Emotional reactions to news coverage of medical errors and the effects of message-induced emotions on risk perceptions, attitudes, and policy preferences

Participants were randomly assigned to one of the four conditions in a 2 (episodic vs. thematic framing) x 2 (anger-inducing vs. fear-inducing) between-participants experiment embedded in an online survey (N = 245). The present study investigated whether emotional appeals, in combination with message frames, would induce emotion-congruent effects with respect to risk perception regarding medical errors, attitudes towards health providers involved in medical errors, and support for punitive and remedial policy measures. In general, this study found expected effects of anger appeal, but not of fear appeal. Exposure to an anger appeal produced greater negative attitudes towards responsible healthcare professionals and enhanced support for punitive policy measures. The fear appeal message did not increase risk perceptions or support for remedial policy measures. Moderated mediation analyses, with experienced anger and fear as mediators and message frame as the moderator, were conducted for attitude and risk perception. Amoderated mediation test with anger as the mediator, message frame as the moderator and attitude as the outcome variable, was close to, but did not reach statistical significance. A serial mediational model was also examined on the relationships between iv anger appeal, experienced anger, negative attitude, and punitive policy support. Specifically, exposure to an anger appeal elicited anger which in turn led to more negative attitudes, and those with more negative attitudes were more likely to support punitive policies. The serial mediation analysis also showed that the simple mediation model via elicited anger explained more variance than the serial mediation model via elicited anger and negative attitude. Counter to expectations, anger appeal, instead of fear appeal, increased risk perceptions through experienced anger, though this effect was significant only for thematic frame. In the serial mediation model, anger manipulation also predicted remedial policy support through elicited anger and then risk perceptions. The serial mediation analysis also indicated a significant simple mediation model: anger appeal predicted remedial policy support via elicited anger. This indirect effect was greater than the indirect effect via elicited anger and risk perceptions. Overall, the findings of the current study highlighted the unique role of anger in influencing human attitude and opinion formation.

EMOTIONAL REACTIONS TO NEWS COVERAGE OF MEDICAL ERRORS AND THE EFFECTS OF MESSAGE-INDUCED EMOTIONS ON RISK PERCEPTIONS, ATTITUDES, AND POLICY PREFERENCES by Tingting Pang A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Communication The University of Utah May 2018 Copyright © Tingting Pang 2018 All Rights Reserved The University of Utah Graduate School STATEMENT OF DISSERTATION APPROVAL The dissertation of Tingting Pang has been approved by the following supervisory committee members: Ye Sun , Chair 12/4/2017 Date Approved Jakob D. Jensen , Member 12/4/2017 Date Approved Kristin Swenson , Member 12/4/2017 Date Approved Kevin DeLuca , Member 12/4/2017 Date Approved Tae Kyoung Lee , Member 12/4/2017 Date Approved and by Danielle Endres the Department/College/School of and by David B. Kieda, Dean of The Graduate School. , Chair/Dean of Communication ABSTRACT Participants were randomly assigned to one of the four conditions in a 2 (episodic vs. thematic framing) x 2 (anger-inducing vs. fear-inducing) between-participants experiment embedded in an online survey (N = 245). The present study investigated whether emotional appeals, in combination with message frames, would induce emotion-congruent effects with respect to risk perception regarding medical errors, attitudes towards health providers involved in medical errors, and support for punitive and remedial policy measures. In general, this study found expected effects of anger appeal, but not of fear appeal. Exposure to an anger appeal produced greater negative attitudes towards responsible healthcare professionals and enhanced support for punitive policy measures. The fear appeal message did not increase risk perceptions or support for remedial policy measures. Moderated mediation analyses, with experienced anger and fear as mediators and message frame as the moderator, were conducted for attitude and risk perception. A moderated mediation test with anger as the mediator, message frame as the moderator and attitude as the outcome variable, was close to, but did not reach statistical significance. A serial mediational model was also examined on the relationships between anger appeal, experienced anger, negative attitude, and punitive policy support. Specifically, exposure to an anger appeal elicited anger which in turn led to more negative attitudes, and those with more negative attitudes were more likely to support punitive policies. The serial mediation analysis also showed that the simple mediation model via elicited anger explained more variance than the serial mediation model via elicited anger and negative attitude. Counter to expectations, anger appeal, instead of fear appeal, increased risk perceptions through experienced anger, though this effect was significant only for thematic frame. In the serial mediation model, anger manipulation also predicted remedial policy support through elicited anger and then risk perceptions. The serial mediation analysis also indicated a significant simple mediation model: anger appeal predicted remedial policy support via elicited anger. This indirect effect was greater than the indirect effect via elicited anger and risk perceptions. Overall, the findings of the current study highlighted the unique role of anger in influencing human attitude and opinion formation. iv TABLE OF CONTENTS ABSTRACT ....................................................................................................................... iii LIST OF TABLES ........................................................................................................... viii LIST OF FIGURES ........................................................................................................... ix ACKNOWLEDGEMENTS ................................................................................................ x CHAPTERS 1: INTRODUCTION ...........................................................................................................1 2: MEDICAL ERRORS ..................................................................................................... 11 The Scope of the Problem ...........................................................................................11 Definitions and Genesis of Medical Errors .................................................................14 Etiology of Medical Errors .........................................................................................17 3: HEALTH NEWS COVERAGE IN THE U.S. MEDIA ................................................28 News Media as a Major Source of Health and Medical Information .........................30 News Coverage of Medical Errors ..............................................................................34 4: FRAMING .....................................................................................................................38 Equivalency Framing ..................................................................................................42 News Media Framing ..................................................................................................46 Mediators and Moderators of News Framing Effects .................................................60 Summary .....................................................................................................................70 5: FRAMING EFFECTS OF EMOTION ..........................................................................73 Emotional Reactions to News Content .......................................................................77 Effects of Emotions on Judgments, Decisions, and Opinions ....................................91 6: METHOD ....................................................................................................................109 Design .......................................................................................................................109 Procedures .................................................................................................................109 Participants ................................................................................................................110 Stimuli .......................................................................................................................111 Measures ................................................................................................................... 114 7: ANALYSES AND RESULTS .................................................................................... 119 Power Analysis .........................................................................................................119 Manipulation Check ..................................................................................................120 Experimental Effects on Anger and Fear Responses ................................................120 Predicting Risk Perception ........................................................................................123 Predicting Attitudes Towards Health Providers Involved in Errors .........................126 Predicting Policy Support .........................................................................................128 8: DISCUSSION ..............................................................................................................146 Experimental Effects of Framing and Emotional Manipulation on Anger and Fear Responses..................................................................................................................149 Effects of Anger Versus Fear Manipulation on Risk Perception, Attitude, and Support for Remedial Versus Punitive Policy Measures ..........................................151 Moderated Mediation and Serial Mediation .............................................................155 Limitations and Future Directions ............................................................................157 Conclusion ................................................................................................................163 Appendices .......................................................................................................................164 A: SWISS CHEESE MODEL OF MEDICAL ERRORS ...............................................164 B: PYRAMID ANALOGY OF MEDICAL ERRORS ....................................................165 C: SUMMARY OF EMOTIONAL MEASURES ...........................................................166 D: EPISODIC ANGER CONDITION ............................................................................170 E: THEMATIC ANGER CONDITION ..........................................................................171 vi F: EPISODIC FEAR CONDITION .................................................................................172 G: THEMATIC FEAR CONDITION .............................................................................173 REFRENCES ...................................................................................................................174 vii LIST OF TABLES 1: Tests of Framing Manipulation, Emotional Manipulation, and Their Interaction on Anger and Fear Responses ..........................................................................................136 2: Conditional Indirect and Direct Effects of Emotional Manipulation on Risk Perception ........................................................................................................................................ 137 3: Conditional Indirect and Direct Effects of Emotional Manipulation on Attitude .......138 4: Tests of Framing Manipulation, Emotional Manipulation, and Their Interaction on Policy support .............................................................................................................139 5: Summary of Emotional Measures................................................................................166 LIST OF FIGURES 1: Interaction of Framing and Emotional Manipulation on Risk Perception .................... 140 2: Moderated Mediation for Risk Perception...................................................................141 3: Moderated Mediation for Attitude ...............................................................................142 4: Indirect Effect of Emotional Appeal on Support for Remedial Measure Through Fear and Risk Perception ..................................................................................143 5: Indirect Effect of Emotional Appeal on Support for Remedial Measure Through Anger and Risk Perception................................................................................144 6: Indirect Effect of Anger Appeal on Support for Punitive Measure Through Anger and Attitude Towards Health Providers Involved in Errors .................................145 ACKNOWLEDGEMENTS This project would not have been possible without the support of many people. First and foremost, I would like to thank the doctoral committee that supported this work from start to finish: my dissertation chair, Ye Sun, and committee members, Jakob D. Jensen, Kristin Swenson, Kevin DeLuca, and Tae Kyoung Lee. I am especially indebted to Ye Sun who has been very supportive of this project and has worked actively to provide me with the protected academic time to overcome all the barriers and problems I had encountered along the way. This work has come out of years of work, support, and perseverance. Without the help of my dissertation chair, I would not have persisted and finally accomplished it. As my teacher and mentor, she has taught me more than I could ever give her credit for here. I owe many thanks to Kristin Swenson who has been very supportive of this work and helped clear up my mind on a number of statistical problems. She has been available to help and reassure me when I need her. She has also been supportive of my career goals and encouraged me to pursue those goals. I am grateful to Jakob Jensen who provided me with directions when I first started on this project and Kevin DeLuca who helped edit my stimuli of news stories. I am also grateful to my friends who have been supportive of this project and encouraged me to complete it: Penchan Pphoborisut, Miao Liu, Wei Wei, and Rui Yan. Last but not least, my families have been so important to me in the pursuit of this dream. I would like to thank my families for their patience, support, and love. xi CHAPTER 1 INTRODUCTION Preventable medical harm has been an ongoing and vexing problem in the U.S. health care system (Andel, Davidow, Hollander, & Moreno, 2012). The magnitude and potential costs of medical errors are tremendous and becoming a fixture in health care (Kalra, Kalra, & Baniak, 2013; Kohn, Corrigan, & Donaldson, 2000). Over the past 2 decades, there has been an increasing concern among the public, health care community and governmental agencies about the safety and quality of modern health care delivery (Robinson et al., 2002). Medical care in the United States is technically complex at the individual provider level, at the system level, and at the national level, creating a highly technical, rapidly changing yet poorly integrated healthcare system that makes patients vulnerable to medical errors (James, 2013). Medical errors are ―diverse, common, and occurring at every level of the system‖ (Kalra et al., 2013, p. 1161). In its seminal paper on medical errors ―To Err Is Human‖ in 1999, the Institute of Medicine (IOM) defined a medical error as ―a failure in the process of delivering care in a complex delivery system‖ (Mclean, 2015, p. 188). The IOM report defined it as ―the failure of the planned action to 2 be completed as intended or the use of a wrong plan to achieve an aim‖ (Tevlin, Doherty, & Traynor, 2013, p. 339). It has become a serious threat to public health and patient safety (Grober & Bohnen, 2005). According to new studies conducted by patient-safety researchers at Johns Hopkins University, medical errors in hospitals and other health-care facilities are common and may now be the third-leading cause of death in the United States (Cha, 2016). The consequences of medical errors are devastating. In addition to loss of human lives, medical errors leave devastating impact on the caretaker, health care system, and society (Varjavand, Nair, & Gracely, 2012). Medical harm imposes a huge financial cost to the nation (Andel et al., 2012). It is estimated to cost between US $17 billion and US $29 billion per year in lost income, lost household production, disability, and additional health care expenses (Grober & Bohnen, 2005, p. 39). The aftermath of a serious adverse event can also have profound consequences on the ―second victims‖ of medical errors: health care providers involved in a medical error incident (Varjavand et al., 2012, p. 1149). This group of victims suffers from emotional distresses such as depression, guilt, self-doubt, and shame as well as anxiety about making errors in the future (Kalra, 2004; Kirby, 2003). Shanafelt et al. (2011) found that 16.2% of 7905 surgeons who reported a recent major error experienced suicidal ideation compared with 5.4% of surgeons not reporting an error. Even the perception of having made a major medical error markedly increased surgeons‘ suicidal ideation by three-fold. Their finding highlights the devastating personal consequences of medical errors on physicians. Moreover, litigation, 3 complaints, and punitive actions can impose additional burden (Kirby, 2003). Kalra et al. (2013) argued that caregivers‘ struggles with emotional turmoils after experiencing a medical error ―deserve equal focus as the other issues of designing safer systems and enhancing the quality of care‖ (p. 1069). As highlighted by Varjavand et al. (2012), errors result in personal distress for doctors which in turn ―contribute to further deficits in patient care‖ and ―increase the risk for damage to patients,‖ creating a ―vicious cycle‖ (p. 1150). It is now widely accepted that reduction of medical errors should be a national priority (Robinson et al., 2002). With increasing media attention, there is growing public awareness of medical errors (Grober & Bohnen, 2005). For health-related issues such as medical errors, research indicates that news media are a major source of information and news for both patients and health care providers alike (Berry, Wharf-Higgins, & Naylor, 2007; Corbett, & Mori, 1999; Grober & Bohnen, 2005; Swift, Koepke, Ferrer, & Miranda, 2001; Viswanath et al., 2008; Wallington, Blake, Taylor-Clark, & Viswanath, 2010; Wang & Gantz, 2007). News coverage of health matters takes on considerable significance, as it can not only influence the perceptions, attitudes, and behaviors of average individuals, but also can shape the decision-making of powerful policy makers (Bryant & Thompson, 2002). Medical errors have received a great amount of attention from the media since the release of the influential Institute of Medicine (IOM) Report, ―To Error is Human‖ in 1999, which called for a national effort to reduce errors rates and improve patient safety (Stebbing, Kaushal, & Bates, 2006; Suresh, 2006). The intensive media coverage of the IOM report 4 ―represented a turning point for the error prevention movement‖ (Millenson, 2002, p. 60) as half of the American public became aware of the report‘s conclusions shortly after it was released, which in turn sparkle a series of legislative and policy responses at both the state and federal levels. To a large extent, the diffusion of innovative efforts in medical error reduction could be traced to ―the public shaming of the profession that has occurred as a result of stories about medical errors in the news media‖ (Millenson, 2002, p. 62). Media effects research has traditionally focused on cognitive effects of media use. Much less attention is paid to the emotional effects of media exposure (Potter & Riddle, 2007). Main theories like agenda setting, priming, and cultivation postulate that media content impacts message recipients primarily via cognitive processes (Bryant & Oliver, 2009; Bryant & Thompson, 2002). In news framing research, for example, cognitive models such as accessibility and applicability effects have been the dominant explanations for underlying psychological mechanisms of news framing effects (e.g., Entman, 1993; Nelson, Clawson, & Oxley, 1997;Scheufele & Tewksbury, 2007; Slothuus, 2008, etc.). Only recently have communication scholars begun to investigate emotional media effects. In particular, the effects of discrete emotions elicited in mediated communication have been studied as mechanisms underlying media effects (Gross, 2008; Konijn, 2013; Lecheler, Kühne & Schemer, 2015; Schuck & De Vreese, 2013). This line of research has generally found that media content not only entails cognitive learning and evaluation processes (Kühne, 2012a), but also could trigger emotional responses (i.e., 5 message-relevant emotions) which may then shape attitudes, judgments, and opinions (e.g., Kim & Cameron, 2011; Nabi, 1999, 2002a, 2002b, 2003). Emotional reactions to media content have been largely explained by cognitive appraisal theories of emotion (Roseman, Spindel, & Jose, 1990; Smith & Ellsworth, 1985). This line of theories posits that if the processing of news content activates a particular interpretation of the issue that matches the key appraisal pattern or a core relational theme (Lazarus, 1991) of a discrete emotion, the corresponding emotion will most likely be elicited. The current study focuses on anger and fear because they seem appropriate given the topic of medical errors and because extant literature tends to indicate that anger and fear could have different effects on judgments and opinions, although they share the same valence. The existing news coverage of medical errors tend to be emotionally provoking. For example, some news stories have depicted how doctors continued to keep practicing medicine and their licenses remained intact after conducting serious medical errors (Eisler & Hansen, 2013). According to appraisal theories of emotion, the perceived injustice will evoke anger and the desire for retribution (Mikula, Scherer, & Athenstaedt, 1998; Nabi, 1999). Other news coverage has highlighted the inherent flaws in the health care system such as communication breakdowns that make patients susceptible to medical errors (Rabin, 2013). According to appraisal theories, the perceived potential threat posed by healthcare (core relational theme of fear) can elicit fear and the corresponding desire to seek protection/prevention (Nabi, 1999, 2003). 6 In addition, a considerable body of social psychological literature indicates that emotions influence attitudes and opinions (Forgas, 2000; Keltner, Ellsworth, & Edwards, 1993; Lerner & Keltner, 2000, 2001, etc.). Early research on influences of affective states on human thoughts, judgments, and decisions has been primarily motivated by a valence-approach, contrasting the effects of positive versus negative affect. The general findings in this domain indicate that global positive moods tend to promote a more optimistic outlook while negative moods tend to produce a more pessimistic outlook (for a review, see Forgas, 2000). A lot of researchers, however, argue that this valence approach lacks power to explain why emotions of the same valence, such as anger and fear, produce opposite effects on judgment and attitude formation. Lerner and Keltner (2000, 2001) proposed the appraisal-tendency framework (ATF) to explain emotion-specific influences on judgment and opinion formation. Their model predicts that ―each emotion activates a cognitive predisposition to appraise future events in line with the central-appraisal dimensions that triggered the emotion- an appraisal tendency‖ (p. 477). For example, anger is associated with high certainty and perceived control while fear is associated with great uncertainty and low perceived control, and according to ATF, anger and fear could produce distinctive effects on risk perception such that angry people tend to make more optimistic judgments and risk-seeking choices while fearful people tend to make more pessimistic judgments and risk-averse choices (Lerner & Keltner, 2000; 2001). Though this body of research has primarily focused on emotional experiences that are ―normatively irrelevant to present judgments and choices‖ (i.e., 7 incidental/unrelated emotions; Han et al., 2007, p. 159), taken together, they are meaningful in that they imply the importance of taking into account emotional responses in examining the news framing effects. This line of research clearly demonstrates that emotions exert influence on subsequent judgments and opinions. Moreover, the general review of extant media coverage of medical errors also revealed that both thematic and episodic frames have been applied to the issue (e.g., Abram, 2014; Daleyfeb, 2015; Gokavi, 2014; Gupta, 2012; McCann, 2014; Ofri, 2013; Southwick, 2013; Weiss, 1999). An episodic frame focuses on personal stories in the coverage of the issue whereas a thematic frame focuses on the underlying political/social factors of the problem. The coverage of medical errors has been characterized by numerous statistical analyses and policy debates (i.e., thematic frames) as well as by lots of highly emotional stories of victims (i.e., episodic frames). The majority of previous studies on episodic and thematic frames focused on their distinctive effects on cognition such as individual versus societal attributions of causal and treatment responsibilities of a social problem as well as policy preferences (e.g., Iyengar, 1990, 1991; Major, 2009). The current study diverges from the previous studies by examining the differential emotional reactions produced by episodic versus thematic news frames and the influence of elicited emotions on ensuing opinions and judgments. Recently, an emerging line of research indicates that episodic and thematic frames not only have distinctive effects on attributions of responsibilities but also could produce differential emotional reactions (Aarøe, 2011; Gross, 2008). This body of research has generally found that episodic 8 frames with a focus on ―human interest details‖ (Baum, 2003, p. 37) are more powerful in eliciting emotions than thematic frames which primarily rely on impassionate statistics and abstract policy analysis (Gross, 2008). Thus, an episodically framed anger or fear-inducing news story about medical errors may be expected to elicit more anger or fear, hence stronger emotion-congruent effects along cognitive tasks (e.g., attitudes, risk perception, and policy preference, etc.) than its thematically framed counterpart. This study manipulates episodic/thematic framing and anger/fear appeal to create four news stories about medical errors as the experimental stimuli. The objectives of the dissertation are twofold: to contribute to existing news framing research by examining whether message-relevant emotions elicited by news frames would influence attitude and opinion formation and whether these effects are moderated by episodic vs. thematic framing. Furthermore, this study examined whether elicited discrete emotions would function as mediators of news framing effects. A great deal of framing research has been focused on the underlying cognitive processes that enable framing effects, failing to systematically address emotions in the framing process. Through an empirical study of the effects of emotional responses to news stories, the dissertation aims to demonstrate that news frames can be designed to elicit specific emotional responses and that frames could rely upon emotional appeals to influence individuals‘ attitude and opinions. This project also extends the existing emotion literature by contrasting the effects of anger and fear as mediators of news framing effects. A number of psychological and political literature clearly suggests that anger and fear are conceptually different and distinct in 9 terms of their effects on opinions and judgments, even if they share similar valence patterns (Nabi, 1999; Lerner & Keltner, 2001). Overall, this study aims to provide further building blocks for the integration of discrete emotions into predominantly cognitive research of framing effects by examining whether and how discrete emotions operate to influence framing effects. The following chapters present a systematic review of the existing literature and outline the methodology for the proposed study. Chapter 2 conducts a substantive review of literature on medical errors in terms of its magnitude and scope, definitions and genesis as well as system and individual causes. It identifies medical errors as a significant issue for public health that needs to be addressed. Chapter 3 presents literature indicating that news media are an important source of medical and health information for the public. It also briefly reviews existing media effects theories (e.g., agenda setting and framing) to argue that the media could exert influence on individual perceptions, attitudes and opinions. Chapter 4 is devoted to the concept of news frames and elaborates on news framing effects. It first provides conceptual clarifications of terms of frame and framing, differentiating frames in the social sciences and news frames in the communication discipline. It further classifies news frames into issue-specific and generic news frames and presents definitions of a topology of generic frames－episodic and thematic frames, which are the focus of this project. Chapter 5 argues that emotions elicited by news frames could function as mediators of news framing effects. While cognitive appraisal theories of emotion can explain why there are emotional reactions to news frames, 10 appraisal tendency framework (ATF) provides the theoretical foundation for the link between elicited emotions and emotion-biased opinion and judgment formation. The chapter primarily relies on prior social psychological research to argue that it is necessary to take into account elicited emotions in framing effects research as this substantial body of research strongly suggests that emotions impacts opinions and judgments. It also identifies a small body of research that directly tested the mediational role of discrete emotions elicited by news frames. Chapter 6 introduces the proposed methodology for this study. It specifies the participant recruitment procedures and provides details of the designing and implementing of the online survey experimental design in order to gather data from the participants. The online survey experiment of a 2 x 2 between-participants design is detailed, and the measures of all the variables in the study are provided. Chapter 7 addressed the proposed research hypotheses by presenting results of statistical analyses and tests conducted in order to test the hypothesized effects of emotional reactions. Last but not least, Chapter 8 is the concluding chapter of the dissertation which provides discussion and conclusions. This chapter summarizes and reviews the major findings of the study. It notes existing limitations of the current project, and explains inconsistent findings with prior studies. Implications of these findings for future framing research are also discussed. Overall, it summarizes the contribution of the dissertation project to the development and refinement of news framing effects theory and to a fuller understanding of the interdependence of emotions and cognitions more broadly. CHAPTER 2 MEDICAL ERRORS The Scope of the Problem In 1999, the Institute of Medicine (IOM) published its landmark report, ―To Err is Human, Building a Safer Health System‖ which brought the issue of medical errors to national prominence (Robinson et al., 2002). The report estimated that between 44,000 and 98,000 patients die every year in U.S. hospitals as the result of medical errors (Hayward & Hofer, 2001). The IOM defines any injury caused ―by medical management rather than by the underlying condition of the patient‖ an adverse event. An adverse event is preventable if it results from errors in execution (failures to complete planned actions as intended), or errors in planning (use of a wrong plan to achieve an aim, Latham, 2001, p. 163). The IOM report focused primarily on medical errors in execution (Latham, 2001). Using the lower estimate, the IOM report asserted that the number of people who die annually from medical errors is greater than the number of people dying from automobile accidents, breast cancer, or AIDS (Harrington, 2005). More importantly, the primary message sent to the public by the IOM report was that the reduction of medical errors 12 would entail systematic changes in the health care system rather than holding individual caregivers accountable (Blendon, et al., 2002; Latham, 2001; Tevlin et al., 2013). IOM crafted a series of legislative and policy recommendations aiming at systematic approaches to resolving medical errors in its 1999 report. The IOM's estimate of error-related deaths has generated controversies and debates since its release (Brennan, 2000; Harrington, 2005). A number of researchers and policymakers questioned the accuracy of the estimate and the reliability of the methodology (e.g., Sox & Woloshin, 2000; Weiner, & Hui, 2000), suggesting that the IOM report may have overestimated the number of deaths due to medical errors (Hayward & Hofer, 2001; McDonald et al., 2000; Robin et al., 2002). On the other hand, a group of researchers proposed that the IOM estimates were actually conservative, and the actual incidence of medical error could be much higher (e.g., Andrew et al., 1997; Leape, 2000; Weingart, Wilson, Gibberd, & Harrison, 2000;, etc.) for two reasons. One was that the study ―included only hospital patients, whereas over half of surgeries were performed in ambulatory surgery centers and millions of patients were in nursing homes‖ (Harrington, 2005, p. 336). The other was that the IOM report may have missed many adverse events and errors that ―are never recorded in the medical record, either because they are concealed or errors not recognized‖ (Leape, 2000, p. 97). Despite such debates, the IOM report has provided impetus for the healthcare system to reform current error reporting systems and make health care safer. 13 The shocking numbers of deaths proposed by the report successfully galvanized public and political attention on the problem of medical error. It was widely discussed by hospital administrators, doctors and health policymakers and was quoted extensively in the media, which deeply shaped public and policymakers‘ perceptions of the issue (Allen, 2013). Since its release in 1999, there has been a flurry of policy and legislative activities focusing primarily on the IOM‘s conclusions of the number of error-caused deaths and its recommendations for the establishment of a comprehensive error-reporting system (Harrington, 2005). Almost two decades have passed since the release of the IOM report in 1999. Despite a national effort to decrease medical errors and improve patient safety, recent research consistently indicates that the improvement in patient safety has remained painstakingly slow (Kalra, 2004; Landrigan et al., 2010; McLean, 2015). Medical errors continue to injure and kill patients at a relentless pace. A 2013 study from the prestigious Journal of Patient Safety estimated that four times as many patients die from preventable medical errors in the United States as the number estimated by the early IOM report, which amounts to 440,000 a year (―Hospital Errors,‖ 2013; Binder, 2013). In 2016, Johns Hopkins patient safety experts announced that approximately 250,000 deaths per year are caused by medical errors in the United States (Rice, 2016). The result was primarily based on an analysis of medical death rate data over an 8-year period. This new figure surpasses the U.S. Centers for Disease Control and Prevention's third leading cause of death—respiratory disease, which kills close to 150,000 people per year (―Medical 14 errors,‖ 2016; Rice, 2016; ―Study suggests,‖ 2016;). These recent studies, along with the early IOM report, converge to show that medical errors have been an increasingly urgent problem for healthcare. Definitions and Genesis of Medical Errors Extant literature indicates that there has been little consensus as to what constitutes a medical error, nor has there been an agreement regarding how medical errors can be classified. The heterogeneity of definitions fundamentally hinders efforts to collect reliable information on medical error and undermines commensurability across results from different studies (Brennan, 2000). The use of multiple and overlapping definitions of medical error is due to ―differing contexts and purposes, such as research, quality control, ethics, insurance, legislation, legal action and statutory regulation‖ (Grober & Bohnen, 2005, p. 40). This section provides a brief review of definitions and classifications of medical errors in the literature. In the IOM report ―To Err Is Human: Building a Safer Health System,‖ medical error was defined as ―the failure of a planned action to be completed as intended or the use of a wrong plan to achieve an aim‖ (Oyebode, 2013, p. 323; Tevlin et al., 2013). Few studies have measured medical errors directly, if this term is ever used at all. Instead, an overwhelming amount of literature adopted surrogate measures of medical error such as noxious episodes, iatrogenic illness, critical incidents, potentially compensatable events, negligence, preventable adverse event, slips, mistakes, and violations (for a review, see 15 Grober & Bohnen, 2005). The five most commonly reported medical errors reported by family physicians in the United States are: (1) errors in prescribing medications (2) errors in getting the right laboratory test done for the right patient at the right time (3) filing system errors (4) errors in dispensing medications and (5) errors in responding to abnormal laboratory test results (Dovey, Phillips, Green, & Fryer, 2003, p. 697). In terms of classification, Kalra et al. (2013) pointed out that medical errors can be classified as skill-based, rule-based, or knowledge-based. Skill-based tasks include previously stored thoughts and actions which are performed unconsciously, and skill-based medical errors are slips such as laboratory errors (also see Ferner & Aronson, 2000). Rule-based tasks refer to situations where the use or disregard of rule(s) results in an undesired outcome. Knowledge-based tasks are those that require the conscious analytic processing of stored knowledge or expertise. Medical errors can also be categorized into system/latent errors or individual/active errors (Kalra et al., 2013; Murphy & Dunn, 2010; Wu, Cavanaugh, Mcphee, Lo, & Micco, 1997). Flaws inherent in the system of medical practice are primarily responsible for the occurrence of individual errors. The system ―sets up‖ individuals to make mistakes, for example, through ―the unavailability of medical records, by confusing labelling of medications, and the like‖ (Wu et al., 1997, p. 770). System errors also include excessive workload, insufficient training, and inadequate maintenance of equipment (Kalra et al., 2013, p. 1164). In system errors, individual caregivers are regarded only partially responsible. By contrast, 16 individual errors are usually caused by deficiencies in the physician‘s health knowledge, professional skill, or attentiveness. Along these lines, medical errors can be further categorized into errors of proficiency, communication, execution, and judgments (Murphy & Dunn, 2010). Errors of proficiency arise when a physician is not equipped with proficient knowledge or skill to perform a specific procedure or examination as required by the professional standard (e.g., a physician made errors when performing a bronchoscopy while he/she lacks training and has not done the procedure independently). Communication errors arise when ―crucial patient information is wrong, missing, misinterpreted, or not appreciated‖ (Murphy & Dunn, 2010, p. 1292). As an example, a pulmonary angiogram is performed in a patient with an elevated creatinine level, but the radiologist is unware that the patient has renal failure (Murphy & Dunn, 2010, p. 1292). Poor communication and disjointed teamwork are now regarded by many researchers as the number one cause of patient harm. As for the solution, it is suggested that physicians learn from other high-risk professions such as aviation and nuclear power to promote their communication skills (Murphy & Dunn, 2010; Reader, Flin, & Cuthbertson, 2007). Execution errors occur when a physician makes a technical error while following the correct protocols despite possessing proficient knowledge and skills. This type of error like a slip or lapse, increases when people are tired and stressed. They are defects in human unconsciousness and hard to target via training or even punishment (Ferner & Aronson, 2000). Finally, judgment errors occur when ―a physician unnecessarily increases patient risk or willfully violates 17 standards of care without a compelling reason‖ (Murphy & Dunn, 2010, p. 1292 & 1293). That a nurse fails to challenge a resident‘s decision not to call for needed help constitutes a judgment error. Etiology of Medical Errors The literature is extensive on etiology factors of medical errors which can be generally summarized into two categories of causes, individual versus system factors underlying medical errors (e.g., Oyebode, 2013). I will provide a discussion of these two causes below. First, the longstanding person approach focuses on blaming individual care providers for errors. It views the unsafe acts of those individuals—errors and procedural violations—as arising primarily from ―aberrant mental processes such as forgetfulness, inattention, poor motivation, carelessness, negligence, and recklessness‖ (Reason, 2000, p. 708). By contrast, a system approach ―identifies the source of the error with the aim of both understanding the origin of errors and building defenses to avert them or mitigate their effects‖ (Oyebode, 2012, p. 325). In the aftermath of a medical error, countermeasures adopting a system approach would be focusing on how and why the defenses and safeguards of the system failed, instead of asking the question of who blundered. The central philosophy of this approach is that medical errors arise within a system, and individual acts are constantly embedded in the system which directs, guides, and constrains the individual‘s behavior (Karlene & Robert, 2001). This concept helps 18 shift the previous exclusive focus on individual error to defective systems underlying errors (Leape, 2000). Correspondingly, a disproportionate number of recommended solutions as well as policy initiatives have been characterized by a systems approach aiming at reforming the existing health care system, building stronger defenses to avert errors, or mitigating their effects (Reason, 2000). Individual Factors Underlying Medical Errors On one hand, the person approach was the dominant tradition in medicine for decades. The extant literature has identified a wide range of individual factors contributing to errors such as slip in attention, the absence of self-awareness of error, personal neglect, illegible handwriting as well as nonadherence of procedures (Karavasiliadou & Athanasakis, 2014; Oyebode, 2013). In particular, cognitive causes of errors have been recognized as important (Stripe, Best, Cole-Harding, Fifield, & Talebdoost, 2006). For example, Stripe et al. (2006) pointed out that medical errors and aviation accidents are similar when it comes to their underlying cognitive causes. They argued that a majority of aviation accidents were due to cognitive errors made by pilots such as poor decision making and risk management. Similarly, physicians become more error-prone when they are affected by a number of factors including fatigue, emotional distress, external pressure, medications, and negative attitudes towards error disclosure, etc. Therefore, they argued that the model of cognitive causes of errors adopted by the aviation community could be applied to that of medicine. The initial aviation model of 19 cognitive errors involved several categories of evaluation of the pilot. Pilots‘ performances were evaluated along several dimensions consisting an acronym called IMSAFE: I, illness in the pilot that could affect the pilot‘s risk management and judgment; M, the effects of medications on reflexes and judgment; S, stress in pilot‘s personal life, family, and so forth, that could influence judgment; A, represents alcohol or other type of substance that could potentially affect judgment; F, physiological factors such as fatigue, hunger, and deprivation of sleep that could influence judgment; E, stands for emotional factors, such as anger, sadness, or depression, which are well known to affect judgment in the emotion literature. Stripe et al. (2006) adapted these criteria of aviation risk factors into a model of Medical Decision Making Risk Management (MDMRM) which was applied to the examination of the causes underlying 126 medical malpractice cases. The results indicated that cognitive factors involving IMSAFE (illness, medications, stress, alcohol, physiology, and emotion) were reliably identified in these malpractice cases. Their study implies that addressing cognitive cause factors or errors on the part of caregivers holds great potential to the reduction of medical errors. Additional support for a cognitive basis of medical error comes from a study conducted by Landrigan et al. (2004), which showed that one of the IMSAFE risk factor, fatigue or lack of sleep, was a major cause in errors made by medical interns in ICU. Similar findings have been reported in nurses and surgery teams. For example, Rogers, Hwang, Scott, Aiken, and Dinges (2004) investigated logbooks completed by 393 hospital staff nurses and they found that extended work shifts and overtime significantly 20 increased both errors and near errors (i.e., an event or situation that did not produce patient injury, but only because of chance, Baldwin & Rodriguez, 2016). Bognár et al. (2008) examined the perceptions and attitudes of surgical teams with regards to committing error and found that team members perceived a high level of workload. In addition, a majority of them felt that their performance was impaired by fatigue caused by excessive workload. It is worth noting, however, that the cognitive factor of fatigue, or lack of sleep, or other seemingly individual causes were attributable to system factors such as hectic work schedule and high workload (Bognár et al., 2008; Gander, Purnell, Garden, & Woodward, 2007; Landrigan et al., 2004; Rogers et al., 2004). The temptation to examine individual error and blame the culprit is hardly resisted. It is, in a way, ―somewhat analogous to the drunk who looks for his lost keys under the streetlight because that is where the light is‖ (Karlene & Robert, 2001, p. 179). There is a growing body of scholars advocating that there should be a move from ―perceiving errors as deriving from individual failures to systems failure and from undue reliance on independent, individual performance excellence to a culture of interdependent, collaborative, interprofessional teamwork‖ (e.g., Leape, 2009; Leape et al., 1995; Murphy & Dunn, 2010; Oyebode, 2012, p. 326; etc.). The section below presents system factors that have been discussed in previous literature. 21 System Factors Underlying Medical Errors Health care is a complex and complicated system which includes multifaceted disease processes, medical staff, equipment, infrastructure, organizational policies, and procedures (Kalra et al., 2013). The multiple components in the system interact with one another and these interactions are ―time dependent‖ (Mclean, 2015, p. 188). With this system getting increasingly complex, ―most errors are due to system failures, not just those of individuals‖ (Kalra et al., 2013, p. 1164). Extant literature has identified key aspects of system failures underlying errors such as high patient-to-nurse ratio, excessive patient workload, problems in supervision and staff composition, communication issues, and the prevailing of a blame culture that thwarts learning from errors, etc. Researchers have long found the nurse-to-patient ratio to be positively related to error rates. For example, Weissman et al. (2007) examined the relationship between peak hospital workload and rates of adverse events (AEs) by reviewing 6841 cases of patients discharged from the medical/surgical services at four U.S. hospitals. The results indicated that for one urban teaching hospital that had enhanced occupancy rates more than 100% for much of the year, a 0.1% increase in the patient-to-nurse ratio led to a 28% increase in the AE rate. Tibby, Correa-West, Durward, Ferguson and Murdoch (2004) conducted a prospective observational study to identify suboptimal working conditions contributing to errors such as workload, supervision issues, training, communication, and equipment. They found that patient workload, staff supervision, and staff composition contributed to errors. For example, during periods of excessive patient workload, there was an increased 22 incidence of patient AE. Permanent staff members acted as a defense mechanism that could hinder the progression of a near miss to an actual event. In addition, the higher percentage of permanent stuff on duty, the lower risk of actual AE. For nursing supervision, the results found a positive association between unsupervised yet non-new residents with an increased risk of AE. In other words, the lack of a senior medical or nurse staff in charge presented a great risk for AE. Overall, their findings showed that multiple organizational factors contributed to AE in various ways. Similarly, Morrison, Beckmann, Durie, Carless, and Gillies (2001) examined the effects of nursing staff inexperience (NSI) on the occurrence of adverse patient events in intensive care units. They found that nursing inexperience and the lack of supervision of experienced staff had played a significant role in causing adverse events. Communication pitfalls and poor teamwork have played a major role in medical errors (e.g., Manning, 2006; Moss, 2005; Murphy & Dunn, 2010). Effective communication and teamwork is essential for quality care (Leonard & Frankel, 2011). The Joint Commission on Accreditation of Healthcare Organizations reported that communication failures accounted for over 60% of root causes of reported sentinel events since 1995 (Manning, 2006; Murphy & Dunn, 2010). Hannaford et al. (2013) examined the type and nature of incidents occurring within medical imaging settings in Australia. They applied 71 search terms related to clinical handover and communication to 3976 incidents in the Radiology Events Registry database, which is open to voluntary reporting of adverse events and near misses involving medical imaging in Australia and New 23 Zealand. Their findings are consistent with previous studies, indicating that incidents involving handover and communication errors were prevalent in medical imaging. Lingard et al. (2004) investigated the pattern and effects of communication failures in the operating room (OR) by recording procedurally relevant communication events using filed notes. They found that out of 421 communication events identified from filed notes, 129 were categorized as communication failures, which accounted for approximately 30% of the total team exchanges. Along these lines, Arora, Johnson, Lovinger, Humphrey, and Meltzer (2005) investigated communication failures that occur during the in-hospital sign-out of patient care between physicians-in-training. The ―sign-out‖ refers to the transfer of care for hospitalized patients between inpatient physicians routinely mediated through written and verbal communication. They interviewed 26 interns caring for 82 patients and found that there were two major types of communication failures: omission (such as medications, active problems, and pending tests), and failure-prone communication processes (such as lack of face-to-face discussion and illegible handwritten notes). All these critical incidents produced uncertainty by the intern during patient care decisions, which may lead to inefficient or suboptimal care such as repeat or unnecessary tests. A growing body of research demonstrates that healthcare providers not only experience excessive communication loads but also are engaged in multiple concurrent tasks and get interrupted/distracted by various sources. The clinical environment in most health service organizations is busy and complex, where multiple sources could produce 24 distractions and interruptions including patients, coworkers, events on the unit, and unexpected tasks, etc. (Karavasiliadou & Athanasakis, 2014). The constant interruptions and distractions in workflow impose tremendous cognitive loads on clinical staff, adversely affecting their working memory, and leading to medical errors (Karavasiliadou & Athanasakis, 2014). Distraction and interruptions are among the top factors contributing to medical errors (Mayo & Duncan, 2004; Mrayyan, Shishani, & Ai-Faouir, 2007; Ozkan, Kocaman, Oztukr, & Seren, 2011; Petrova, 2010; Stratton, Blegen, Pepper, & Vaughn, 2004). Spencer, Coiera, and Logan (2004) conducted an observational study to investigate the differences in role-related communication patterns in the emergency department (ED). Based on a total study observation time of 19 hours and 52 minutes, they identified 831 distinct communication events, an average of 42 events per person per hour. Eighty-nine percentage of clinician spent their time in communication events, among which synchronous communication channels, such as face-to-face communication and the telephone, were used in the majority of the events (84%). It is worth noting that one third of events were classified as interruptions, which produce an average rate of 15 interruptions per person per hour. Also, individuals with greater authority such as senior medical staff and nursing, experienced a higher communication load and higher overall rates of interruptions. Interruptions are common in intensive care units (ICU). For example, Alvarez and Coiera (2005) found that staff in the ICU spent the majority of their time (75%) in communication events during ward rounds and that interruptions were prevalent. Specifically, over a third of communication events were classified as 25 communication-initiating interruptions that disrupt working memory and contribute to error. Moreover, another significant additional source of interruptions through turn-taking, intraconversational disruption was identified, contributing to 5.3% of total communication event time and exerting similar effects on cognitive processes. Combined together, both types of interruptions suggested the overall tremendous burden of interruptions in ICU settings. Furthermore, a rampant ―blame culture‖ and the lack of a ―just culture‖ represents a deeper level of cause underlying medical errors, as learning from errors has been primarily thwarted when a culture of blame is prevalent (Kalra et al., 2013, p. 1164; Tevlin et al., 2013). A blame culture can be defined as ―a set of norms and attitudes within an organization characterized by an unwillingness to take risks or accepting responsibility for mistakes because of fear of criticism or punishment‖ (Kalra et al., 2013, p. 1164). By contrast, a just culture ―recognizes human fallibility but importantly establishes clear expectations of responsibility and does not unfairly or routinely blame those who make mistakes‖ (Waring, 2005, p. 1928). A just culture does not mean abandoning individual accountability for mistakes. Rather, it recognizes that not just those front-line medical staff are accountable, but also people located at multiple levels of the healthcare system (Woodward, Lemer, & Wu. 2009). The chilling effect of blame on reporting of medical errors has been well documented in the literature (e.g., Bognáret al., 2008; Meaney, 2004). Thus, the success of incident reporting is fundamentally premised 26 on the transformation from a blame culture to a just culture that promotes openness, transparency, and learning from errors (Waring, 2015). Summary of Individual and System Causes The above discussion demonstrates that both individual and system factors contribute to medical errors. In the literature, the relationship between latent errors and active errors was illustrated by several visual models. For example, Reason‘s Swiss cheese model has been the dominant paradigm for explaining medical errors (Groper & Bohnen, 2005; Perneger, 2005). Reason proposed the image of ―Swiss cheese‖ to explain the contribution of invisible system failures or latent errors to the observerable, active errors committed by providers (see Appendix A). According to the model, health care organizations have several layers of defense or safeguards that offer protection against the negative consequence of adverse events (signified by several slices of Swiss cheese). However, within each individual safeguard or slice of cheese, there are several flaws or defects (the holes in each individual slice) that allow errors to pass through to reach the patient. According to Reason, these holes are ―continually opening, shutting, and shifting their location‖ (Reason, 2000, p. 769), creating new hazards and risks in the system for both patient and providers. In reality, these defects arise from both active failures such as slips, lapses, fumbles, mistakes, and procedural violations committed by providers, and latent conditions including flaws in regulatory or institutional policies and procedures, problems with leadership, and the working environment (Coolines, Newhouse, Porter, 27 & Talsma, 2014; Reason, 2000). The presence of holes in any one defense layer does not normally lead to an adverse outcome, because if one defense fails, the next down the line should prevent the perpetuation toward an adverse event. However, an adverse event can happen when the holes in all the layers momentarily line up to allow an error to penetrate all the defenses and reach the victim (Grober & Bohnen, 2005). The relationship between system errors and individual errors can also be explained by a pyramid analogy (Kalra et al., 2013) which demonstrates the significant role played by system factors in causing individual errors. The figure clearly demonstrates that the system factors at the bottom, such as organizational culture, managerial policies, hierarchical structure of medicine, and work environment, have a powerful influence on human deficiencies on the top of the pyramid (see Appendix B). Though the immediate error seems to be attributable to the individual provider who committed it (i.e., active error), the latent deficiencies (i.e., latent errors) at the case of the healthcare system play a crucial role in causing mistakes, violations, and incompetence on the part of individual providers. Therefore, efforts directed at the iceberg of the problem will only achieve minimal effects. Remedial actions need be focused on changing the system factors underlying errors such as improving teamwork infrastructure, promoting communication among team members, and transforming from a culture of blame to a just culture. CHAPTER 3 HEALTH NEWS COVERAGE IN THE U.S. MEDIA The news media plays a prominent role as the information provider to a wide and diverse audience, influencing what Americans consider important and how they understand social and political issues (Brodie, Hamel, Altman, & Benson, 2003; Pribble et al., 2006). Mass media communication of medical and health information through various media channels, including television, newspapers, magazines, radio, and the Internet, has proliferated in recent years (De Jesus, 2013). Health communicators, government agencies, and nonprofit organizations have primarily relied on the media to disseminate medical and health-related news to the public (Corbett & Mori, 1999). According to a survey conducted by the Pew Research Center in 2002, 71% of American adults closely follow health news (Mebane, 2003). In addition, the Kaiser Family Foundation (Kaiser)/Harvard School of Public Health (HSPH) Health News Index (HNI), a series of 39 surveys, measured how closely a total number of 42,000 people followed health news stories and health issues between August 1996 and December 2002 (Brodie et al., 2003). The study found that on average, 4 in 10 adults reported following health news closely and that public health, health policy, and diseases related stories were the 29 top three issues that the public reported paying the most attention to. The wide news coverage of health issues makes sense given the ―centrality of this domain to social, economic, and political life‖ (Hallin & Briggs, 2015, p. 86). As an important source of medical and health information, news media exerts influence on news consumers. The media can shape individuals‘ understandings and perceptions of health issues via agenda setting, framing, cultivation, and priming effects. Mass media has the power to raise the salience of a certain health issue and thereby set the agenda for the general public and policymakers, influencing what health issues they think about or regard as important (Kim & Wills, 2007). The media influences how individuals interpret health issues via framing an issue in a specific way (Kim, Scheufele, & Shanahan, 2002). For example, a large body of news framing research has focused on how news framing of a health issue (e.g., obesity) could impact individuals‘ perception of who is responsible for causing and treating the problem (Kim & Wills, 2007). Framing effects in health communication have also focused on the persuasiveness of news frames (Riles, Sangalang, Hurley, & Tewksbury, 2015). For example, there is substantive research on the relative persuasiveness of gain- versus loss-frame in promoting a recommended health behavior. A gain-framed health message emphasizes the benefits of adopting a health behavior whereas a loss-framed health message stresses the negative consequences of not performing the act (Levin, Schneider, & Gaeth, 1998; Meyerowitz & Chaiken, 1987). Communication scholars also generally agree that media coverage can act as a prime for individuals, triggering ―concepts, thoughts, learning, or knowledge 30 acquired in the past and related to the message content‖ (Bryant & Thompson, 2002, p. 96). Along these lines, cultivation theory posits that long-term exposure to health news coverage could shape and cultivate individual‘s health knowledge and beliefs (Wang & Gantz, 2007). News Media as a Major Source of Health and Medical Information Mass media communication of health and medical issues includes print media coverage, television broadcasts, radio, cable, and the Internet (Berry et al., 2007). The media influence personal feelings, attitudes, and behaviors regarding health, illness, and medicine (Clarke & Binns, 2006). The media provide the public with information regarding fundamental health policies related to ―diagnosis, treatment, prevention, health promotion, research directions and supportive services‖ (Clarke & Binns, 2006, p. 39), and help reinforce the belief that the existing medical system is legitimate. Extant research has found that despite the growing ubiquity of Internet access and the proliferation of mobile technologies, traditional news sources such as print newspapers and televisions broadcast still play a dominant role in disseminating important health messages to the public, informing and influencing individual behaviors and policy decisions (Slooten, Friedman, & Tanner, 2013). First of all, the role of mass print media has long been recognized in disseminating health information and research discoveries to the American public, shaping their perceptions and understandings of health issues as well as impacting their behaviors 31 (Moyer, Creener, Beauvais, & Salovey, 1995). Mass print media include newspapers, magazines, newsweeklies, pamphlets, books, and science periodicals (Brown, Zavestoski, McCormic, Mandelbaum, & Luebke, 2001; Clarke & Everest, 2006; Moyer et al., 1995). Print media were identified with a frequency second only to physicians as an important source of illness prevention and health promotion for the American public (Meissner, Potosky, & Convissor, 1992). The print media coverage of health news has many advantages over other media channels. For example, they are ―more permanent than television or radio reports, are inexpensive, can be left in the room, read and reread, and they are in widespread circulation‖ (Clarke & Binns, 2006. P. 40). In rural and isolated areas, print media coverage is a critical media component of health intervention projects. In these areas, local newspapers are the most important source of health information due to ―lower overall availability of health care providers and fewer television and radio stations‖ (Brownson et al., 1996, p. 483). For example, Brownson et al. investigated coverage of cardiovascular diseases (CVD) and CVD-related issues by 23 newspapers in six rural counties in southeastern Missouri for the period October 1988 through August 1993, as part of their evaluation of a community-based health promotion project. Respondents reported hearing of heart health coalitions primarily from local newspapers, a decided advantage of which is its low cost in dissemination information. Their media content analysis documented increasing print media coverage of CVD-related health issues and shows the important of print media in proving CDVD health knowledge and awareness among people in high-risk, rural areas. 32 Although print media coverage is an important source of health information, extant research tends to indicate that television news, in particular, local television news, is the public‘s most preferred source of health news (Tanner, Friedman, & Zheng, 2015). Television has become American‘s single most important source of general news and television reporting of health news has increased significantly during the past decades (Pollard, 2003; Pribble et al., p. 170). According to the Pew Research Center, 57% of Americans watch local television news on a regular basis (Pribble et al., 2006). Americans rated local television news as their primary source of health information and 76% reported acting on health information provided in the television (Pribble, et al., 2006). For example, information and news regarding public health emergencies are disseminated primarily via television news, which provides the public of critical health information that help them navigate through emergencies. Pollard‘s (2003) study found that local television news was the top media source for information regarding bioterrorism, based on an analysis of data from six national surveys before and after the bioterrorist anthrax attacks in the fall of 2001. For other health care emergencies such as Hurricane Katrina, H1N1 influenza, SARS, and September 11, 2011, the public reported relying on mass media, particularly television news, for updated information (Berry, et al., 2007; Hallin & Briggs, 2015; Tanner, Friedman, Barr, & Koskan, 2008). Moreover, with the proliferation of the Internet, the public today ―have more health information at their fingertips than ever before‖ (Slooten et al., 2013, p. 35-36). Individuals are increasingly using the Internet as a source of medical and health 33 information (Diaz et al., 2002; Jensen, King, Davis, & Guntzviller, 2010). Local television news, in addition to its primary function of health news reporting, has also been found to motivate viewers to seek additional health information online (Friedman & Tanner, 2007; Kerschbaumer, 2000; Tanner et al., 2008). Viewers exposed to health information on local TV news are also prompted to turn to the Internet to seek more information regarding a health issue. The internet is perceived by viewers to be advantageous over traditional news sources in many aspects such as ―the potential for interactive learning, regular availability of information, timely updates, ability to tailor information for a diverse group of users and the opportunity to search sensitive health topics in anonymous ways‖ (Slooten et al., 2013, p. 36). Though research is mixed in terms of the accuracy and reliability of online health information (Friedman & Hoffman-Goetz 2007; Friedman & Kao, 2008; Yeaton, Smith, & Rogers, 1990), an increasing number of people now rely on the Internet to aid them in health and medical decision making. In addition to searching health information online, motivated viewers can directly access their local media websites for more detailed health information. Studies found that nearly all local television stations have a website (Papper, 2009, as cited in Slooten et al., 2013), and it is increasingly their top agenda to move their news content online and promote their digital content as an indispensable part of health information accessible to the public (Smith, Tanner, & Duhe, 2007; Tanner and Friedman, 2011; Tanner et al., 2008). 34 To conclude, medicine and health are major topics in the news agenda and are among the most prominent elements of contemporary news coverage (Hallin & Briggs, 2015). Though the Internet and mobile technologies have gained increasing popularity among users as an important source for health news, traditional media such as newspapers and television broadcasts are still playing a dominant role in disseminating health news and information. Americans rely on a blend of both traditional and new sources to get health information (Slooten et al., 2013). News Coverage of Medical Errors For medical errors, extant research demonstrated that the mass media played a pivotal role in raising public awareness of the issue. Medical errors have had a high profile in the media since the release of the influential Institute of Medicine (IOM) Report, ―To Error is Human‖ in 1999 that called for a national effort to make health care safer (Stebbing et al., 2006; Suresh, 2006). The report estimated that, each year, between 44,000 and 98,000 people die in the United States because of medical errors (Macready, 2000). The New York Times compared the IOM‘s estimated number of deaths ―to having three jumbo jets filled with patients crash every two days.‖ (Harrington, 2005). The Kaiser Family Foundation/Harvard School of Public Health Survey indicated that half of the American public closely followed the media coverage of the IOM‘s report about the high number of medical errors (Blendon et al., 2002). 35 The intensive media coverage of the IOM report aroused wide public concern of patient safety and unleashed a torrent of criticism against its estimates of error-related deaths (Hayward & Hofer, 2001; Leape, 2000; McDonald et al., 2000; Sox & Woloshin, 2000; Weingart et al., 2000). The debates around the issue sparked a series of legislative and policy responses. There has been a flurry of activity in congress and in state legislatures focusing on the IOM‘s shocking numbers of patient deaths and its recommended four-part strategy, the most important part of which is the establishment of mandatory and voluntary error-reporting systems (Harrington, 2005). The patient safety movement has also burgeoned since the release of the report. Despite debates around the accuracy of its estimated number of deaths, the media coverage of IOM report ―represented a turning point for the error prevention movement‖ (Millenson, 2002, p. 60), as it successfully galvanized public and political attention on medical error (Harrington, 2005). Medical errors are still top news today, and recent studies indicate that medical errors are now the third leading cause of death in the United States, just behind heart disease and cancer (Marcus, 2016). Scientists from Johns Hopkins published their findings in The BMJ on May 3 of 2016, which estimated that more than 250,000 Americans die due to medical errors every year, including wrong diagnoses, botched surgeries, and medication mistakes (Rice, 2016). The majority of the most influential broadcast and cable networks, including CNN, ABC, NBC, CBS, and Fox News, have devoted coverage to this new estimate of error-related deaths. Mainstream newspapers in the United States have also 36 reported this research finding, including The New York Times (Bakalar, 2016), The Washington Post (Cha, 2016), Los Angeles Times (Netburn, 2016), USA Today (O‘Donnell, 2016), and The Wall Street Journal (Lieber, 2016), just to name a few. The online versions of these coverages can be readily accessed by searching key words of medical errors via Google and other search engines or by directly visiting online news websites. A general search of the newspaper coverage of the topic on LexisNexis using the key term medical errors identified a total of 530 newspaper articles published in the United States. These articles can be further categorized based on their specific newspaper sources. Tribune Content Agency-Business News (96 entries), FierceHealthcare (58 entries), The New York Times (55 entries), The Washington Post (44 entries), and St. Louis Post-Dispatch (31 entries) are the top five sources that produced the most newspaper coverage on the issue. To summarize, for medical errors, news media has played a prominent role in informing the public about the magnitude of the problem and engaging both the public and policymakers in discussion of solutions. The abundance of news reporting on medical errors has the potential to sway the public‘s perceptions and opinions regarding the issue, as indicated by media effects research. This dissertation attempts to investigate the mechanism underlying this influence process and contribute to the existing body of research on media effects via focusing on the mediational role of elicited discrete emotions. 37 Given that the news media serve an integral function in communicating health-related news and disseminating health messages to the general public and policymakers alike, it is necessary to understand what effects health news exert on individual news consumers. Well-established communication theories of media effects such as agenda setting, framing, cultivation, and priming posit that the news media could influence individual perceptions, attitudes, and behaviors regarding a given issue in various ways. In the next chapter, I will elaborate on news framing, a concept related to the agenda-setting tradition but expanding the research by focusing on the promotion of a particular interpretation rather than on the promotion of the salience of the issue. While agenda setting posits that news coverage tells people what health issues to think about, news-framing theory and research indicate that news frames incline individuals to think about a health issues in a particular way and form a particular perception. CHAPTER 4 FRAMING The past decades have witnessed the proliferation of framing research in the area of human judgment and decision making, expanding to ―include domains as diverse as cognition, psycholinguistics, perception, social psychology, health psychology, clinical psychology, education psychology, and business‖ (Levin et al., 1998, p. 150). The terms frame and framing have been used in a relatively broad sense, so much so that they now possesses multiple meanings (Kuhberger, 1998; Shen & Dillard, 2007). There is ―substantial conceptual confusion‖ (Chong & Druckman, 2007, p. 114) about types of framing effects, because scholars have used the terms frame ―in different and conflicting ways‖ (Druckman, 2001, p. 226). Druckman (2001, 2004) distinguishes between emphasis or issue framing effects and equivalency or valence framing effects, and work on framing effects has evolved into these two distinct literatures. Equivalency framing effects occur when different descriptions of the same decision situation cause individuals to produce different preferences, ―despite the fact that the ‗acts, outcomes, and contingencies‘ associated with 39 the decision remain invariant across the descriptions‖ (LeBoeuf & Shafir, 2003, p. 78). It involves casting logically equivalent information in either a positive or negative light to alter individual preferences (Chong & Druckman, 2007). Tversky and Kahneman‘s (1984) classical study of the Asian disease problem provided a prototypical example of equivalency framing effects. They found that participants demonstrated a preference for risk aversion when presented with a gain frame while they made risk-seeking decisions when presented with a loss frame. This type of framing, introduced by Tversky and Kahneman, is risky choice framing, and is ―the form most closely associated with the term framing‖ (Levin et al., 1998, p.150). In addition to risk preferences, extant literature on equivalency framing effects examines specific evaluations or behaviors (Druckman, 2001). Levin et al. (1998) proposed a typology to distinguish between three different types of valence framing effects, all of which involve casting the same critical information in either a positive or a negative way to alter individuals‘ evaluation and perception of the issue. These three framing effects, ―are often treated as a relatively homogeneous set of phenomena explained by a single theory, namely prospect theory‖ (Levin et al., 1998, p. 150). The first type of these framing manipulations is risky choice framing introduced by Tversky and Kahneman (1984) which focuses on how valence of frames affects risk perception. Another form of valence framing is attribute framing, wherein the attributes or characteristics of an object/event is framed to affect individuals‘ evaluation of the item. A third category of these framing manipulations is goal framing, in which the issue may be 40 framed to highlight its potential to provide a benefit (positive frame) or on its potential to prevent a loss (negative frame). Research on goal framing tends to focus on the relative persuasiveness of the positive versus the negative frame on promoting the recommended behavior. The other type of framing effect is referred to as emphasis or issue framing effects (Druckman, 2001). Emphasis or issue framing is what most researchers refer to as news framing. It involves the presentation of news coverage in various ways that aims at promoting a particular interpretation or judgment of the issue. While equivalency frames employ objectively equivalent descriptions, emphasis frames focus on ―qualitatively different yet potentially relevant considerations‖ (e.g., free speech or public safety; Chong & Druckman, 2007, p. 114). News framing is not concerned with making logically equivalent or materially identical statements of the same critical information (Chong & Druckman, 2007). Rather, journalists focus on a certain angle or perspective in composing the news story such that the audience could follow this perspective in understanding the issue. News frames in the domain of health and political communication can influence perceptions and attitudes. For example, if journalists cover obesity in terms of personal laziness (i.e., individual frame), the public will tend to hold individuals as primarily responsible for causing and treating the problem. By contrast, if journalists employ a societal frame which highlights the role of the food industry in causing obesity, the public will be aware of the social factors and hold the society as a whole as responsible for fixing the problem (Kim & Willis, 2007; Lawrence, 2004). In 41 covering electoral politics, if journalists use a conflict or game frame (Price, Tewksbury, & Powers, 1997) to depict the election campaign and present politics as a horse race, a military battle or sport, the voters will tend to focus on the tactics and maneuvers of the political candidates and produce negative emotions such as anger and disgust at the politicians‘ self-interested motivations (Gross & Brewer, 2007). However, if a substance frame is employed, the voters will tend to be directed to focus on the pros and cons of policy proposals and their implications. The objective of the chapter is to examine the different ways that the extant literature has employed the concepts of framing and framing effects. The first part of the chapter provides a brief review of the existing research on three types of equivalency/valence framing effects. The second part of the chapter elaborates on emphasis or news framing, which is the focus of the dissertation study. A review of definitions of news frames and framing in the literature are provided, and a typology of news frames (e.g., generic versus issue-specific frames) is presented. In particular, a typology of generic frames—thematic and episodic news frames—are introduced, as the current study aims to investigate the relative effectiveness of these two types of frames in eliciting individual emotional reactions about medical error. The chapter then concludes with a brief review of traditional models of framing effects in the cognitive domain, indicating a gap in the knowledge base of framing effects. The next chapter then focuses on emotional framing and organizes the literature surrounding the question of whether and how emotions contribute to news framing effects. 42 Equivalency Framing Risky Choice Framing Tversky and Kahneman (1984) described the outbreak of an ―Asian disease problem‖ to demonstrate risky choice framing effects. Specifically, participants read a scenario that the United States was preparing for the outbreak of an Asian disease, which was expected to kill a total of 600 people, and asked to choose from two sets of logically equivalent yet different options. In the positively framed version, which focused on number of lives saved, participants were asked to choose between Program A, the adoption of which could yield a sure saving of 200 people, and Program B, if adopted, which could produce a one-third chance of saving all 600 lives and a two-thirds chance of saving no lives. In the negatively framed version that highlighted the consequence of the option in terms of death, participants were also asked to choose from two programs. If Program C was adopted, 400 people would die, which was a sure loss. And if Program D was adopted, there was a one-third probability of losing no lives and a two-thirds probability of losing all the 600 lives. Their findings demonstrated a ―choice reversal,‖ where the majority of participants (72%) in the positively framed condition chose Program A (a riskless behavior that would yield a certain outcome), the majority of participants (78%) in the negatively framed condition preferred Program D (risk-seeking behavior). Individuals‘ risk preference seemed to depend upon the valence of the framing: They tended to be risk-averse when gains were highlighted whereas they tended to be risk-seeking when losses were salient. 43 Kahneman and Tversky (1979, 1984) used prospect theory to explain this phenomenon. The major premise of the theory is that how the option is framed will yield ―different subjective values for corresponding gains and losses‖ (Kuhberger, 1995, p. 231). Simply speaking, prospect theory posits that people value losses and gains differently and tend to make decisions based on perceived sure gains instead of perceived potential losses. As an example, if an individual is presented with two equal choices, with one framed in terms of sure gains but the other in terms of potential losses, the theory posits that people will mostly prefer the former one. Because people tend to have a strong preference for certainty and an inclination to give losses more weight than gains (i.e., loss aversion). As pointed out by Levin et al. (1998), most people ―have an S-shaped subjective value function that is concave in the domain of gains and convex in the domain of losses,‖ thus, the framing manipulation determines whether individuals evaluate the outcomes in terms of gains or losses (p. 152). Attribute Framing and Goal Framing Attribute framing represents the simplest form of valence framing effect, as it demonstrates how positive or negative framing of the attribute or characteristics of an object will affect information processing and evaluation. For example, one study of attribute framing by Levin and Gaeth (1988) showed that when the beef was labeled as 75% lean (a positive frame), customers‘ rating of the beef quality was more favorable than when it was labeled as 25% fat (a negative frame). Similar examples of attribute 44 framing include describing surgery or other medical treatments in terms of survival rates versus mortality rates (McNeil, Pauker, Sox, & Tversky, 1982), describing the performance of basketball players in terms of the shots missed or made (1987), and describing the result of exams in terms of percentage correct or percentage incorrect (Levin et al., 1998). Goal framing, another type of valence framing, has been widely applied in persuasive communication to promote the adoption of a certain health behavior. Goal framing emphasizes either the positive consequences of performing an act (or gain) or the negative consequences of not performing the act (or loss). An example of a gain-framed message is ―If you follow the Surgeon General‘s recommendations, you will increase your chances of living a long, healthy life,‖ contrasted by a loss-framed message that might state ―If you do not follow the Surgeon General‘s recommendations, you will increase your chances of dying early‖ (Detweiler, Bedell, Salovey, Pronin, & Rothman, 1999, p. 190). Gain- and loss-framed appeals can each take two forms. A gain-framed appeal might be stated as ―If you perform the act, you will obtain the desirable consequence‖ or ―If you perform the act, you will avoid the undesirable consequence.‖ Similarly, a loss-framed appeal may also be expressed by either focusing on the loss of the positive outcome or the obtaining of negative consequence (O‘ Keefe & Jensen, 2007). In goal-framing manipulation, both framing conditions focus on the promotion of the same act (Levin et al., 1998). Research in goal framing is primarily concerned with which 45 frame, positive or negative, will enjoy a persuasive advantage in promoting the same end result. There is a substantive body of empirical studies investigating the relative persuasiveness of gain- and loss-framed appeals on health-relevant decisions and behaviors. Results are mixed as to which type of framing is more persuasive. On one hand, some studies have generally found that gain-framed appeals are more persuasive when the target health behavior is a relatively low-risk action such as preventive health behavior while loss-framed appeals enjoy a persuasive advantage when riskier actions are contemplated such as health detection behaviors (Detweiler et al., 1999; Meyerowitz & Chaiken, 1987; Rothman, Bartels, Walschin, & Salovey, 2006). A well-known example of a goal framing effect is provided by the study of Meyerowitz and Chaiken (1987). They showed that college-aged female participants demonstrated more positive breast self-examination attitudes, intentions, and behaviors when they read a pamphlet stressing the negative consequences of not performing breast self-examination than did those in the gain-framed condition and the control group. Bigman, Cappella, and Hornik (2010) found that respondents exposed to positive framing perceived greater effectiveness of the human papillomavirus (HPV) vaccine and indicated more support of a vaccine mandate policy than those exposed to the negative frame. On the other hand, however, a group of researchers contested that there were no significant differences in persuasiveness between gain- and loss-framed health messages in promoting preventive health behavior such as safer-sex behaviors, skin cancer prevention behaviors or diet and nutrition behaviors (e.g., 46 O‘Keefe & Jensen, 2007, 2009). The mixed results might be due to moderators. For example, Riet, Ruiter, Werrij, and De Vries (2010) found that self-efficacy to perform skin self-examination was a significant moderator of the framing effects. Loss-framed skin-detection messages were more effective than messages stressing gains in promoting active detection of skin-cancer symptoms, but only for those with high self-efficacy. Gerend and Maner (2011) investigated whether anger or fear influenced health decision-making. Participants were subjected to a fear or anger induction task and then they read a gain- or loss-framed pamphlet promoting fruit and vegetable consumption. The fruit and vegetable intake of participants were assessed over the following two weeks. They found that the effectiveness of the framed message was moderated by the emotional state of the recipient such that participants in a fearful emotional state were more responsive to a loss-framed message than to a gain-framed message, whereas participants in an angry emotional state were more responsive to a gain-framed message than to a loss-framed message. News Media Framing In emphasis or issue framing, journalists emphasize a subset of potentially relevant considerations or concepts to lead individuals to focus on these considerations when constructing their perception and opinion (Druckman, 2004). Within the last decade, the concept of emphasis framing has become increasingly attractive in media research, as it has been particularly useful in understanding the media‘s role in political and social life 47 (Reese, 2001). Although equivalency or valence framing effects and prospect theory are valuable in the study of individual-level decision making, emphasis framing plays a more significant role in policy discourse and political communication (Slothuus, 2008). Emphasis framing is often referred to as news media framing, as they are ―closer to ‗real‘ journalistic news coverage and present qualitatively different yet potentially relevant considerations‖ (Lecheler et al., 2013, p. 190). The two terms have been used interchangeably in the literature. The current study uses the term news framing instead of emphasis framing in the following chapters. In contrast to equivalency framing, news framing is not concerned with presenting exactly equivalent information but focuses on a particular aspect of an issue (Slothuus, 2008). News framing is ―the typical manner in which journalists shape news content within a familiar frame of reference and according to some latent structure of meaning‖ (Van Gorp, 2007, p. 61). In the coverage of a news event, the intended focus of the journalist is to make certain interpretations more relevant and applicable in the subsequent opinion formation of the audience (Sun, Krakow, John, Liu, & Weaver, 2016). On the other hand, individuals respond to the news story and interpret a news event via their own frames, or schemata, which might be different from the news frame adopted by the journalist. Extant literature differentiates between news frames and individual frames. The concept of frame ―maintains a useful tension or balance between structure and agency‖ (Gamson, Croteau, Hoynes, & Sasson, 1992, p. 384). On one hand, the news media rely on news frames to organize media content and 48 give specific meanings to a news story; on the other hand, the audience interprets the message via their individual frames (Van Gorp, 2007). News Frame A news frame is a ―storyline or unfolding narrative about an issue‖ (Gamson et al., 1992, p. 385). It is ―a central organizing idea for making sense of relevant events and suggesting what is at issue‖ (Gamson & Modigliani, 1989, p. 57). News frames are also alternative definitions, constructions, or depictions of a policy problem (Nelson & Oxley, 1999, p.1041). Similarly, Gitlin (1980) conceptualized news frames as ―persistent patterns of cognition, interpretation, and presentation, of selection, emphasis, and exclusion, by which symbol-handlers routinely organize discourse‖ (p. 7). Entman (1993) further specified that frames have four functions, namely, frames can define problems, diagnose the underlying causes, make moral judgments about which causal agents should be responsible for the problem, and suggest remedies and evaluate their possible effects. Extant research show that various factors influence the construction of news frames, a process called framing building (Scheufele, 1999). Gurevitch and Levy (1985, as cited in Gamson et al., 1992) stated that ―the media becomes a site on which various social groups, institutions, and ideologies struggle over definition and construction of social reality‖ (p. 385). In general, the extant literature has identified five levels of influences which include ideological or political orientations of journalists, journalistic routines, organizational pressures and constraints, pressures of interest groups, and social norms 49 and values (Scheufele, 1999; Van Gorp, 2007). First of all, journalists approach their work with their own ideologies, schemata and professional norms. In addition, media routines, political orientation, and other organizational-level influences of a media organization may influence journalistic choices of news frames. Moreover, external, social influences of various interest groups, lobbyists, powerful communication agents, and social elites may exert pressure on media organizations and individual journalists in order to promote their preferred frames in mass-media outlets (Nelson & Oxley, 1999). Finally, the dominant ideologies of a culture exert powerful influence on the type of news frames adopted. For an example, Akhavan-Majid and Ramaprasad (1998) conducted a comparative analysis of U.S. and Chinese media framing of the fourth United Nations Conference on Women and the Non-governmental Organizations Forum in Beijing forum. The results indicated that the U.S. coverage was characterized by an anticommunist frame that focused on criticism of China, reflecting the strong influence of anticommunist ideology. The majority of the U.S. news stories depicted China as an oppressive communist nation replete with human rights problems, as a nation with repressiveness, ineptitude and clumsiness, and as a nation that harbors ―dirty secrets‖ and is unwilling to submit to a ―rule of law‖ (p. 145). On the contrary, the Chinese coverage used a proequality frame and was characterized by a clear focus on the central issues of concerns to the conference itself. The Chinese media also downplayed the theme of conflict by presenting ―a strong focus on the theme of cooperation to reach solutions‖ (p. 147). The 50 overall Chinese coverage reflected China‘s clear motivation as host of the international event to cast itself in a positive light. Individual Frame Individual frames have been defined in various ways in the literature. For example, Gamson and Modigliani (1989) stated that individual audience members are active information processors, as they process framed media content with ―some anticipatory schema composed of their own life histories, social interactions, and psychological predispositions‖ (p. 5). An individual frame is an ―existing conceptual system that gives order and meaning‖ to the large array of words, images, actions, or text of any kind that an individual deals with in everyday life (Dorfman, Wallack, & Woodruff, 2005, p. 324). Individual frames are cognitive mental framework or schemata through which individuals come to understand the world (Fiske & Taylor, 1991). A schemata can be defined as ―collections of organized knowledge, develop gradually, become more complex, and related to personal experiences and associated feelings‖ (Van Gorp, 2007, p. 63). Individual frames are also named as ―frame in thought‖ (e.g., Chong & Druckman, 2007), ―audience frames‖ (e.g., Scheufele, 1999), as well as ―conceptual frames‖ (e.g., Dorfman et al., 2005). A frame in thought eventually determines what considerations about the event are the most important in evaluating it, unlike news frames which reflect the journalist‘s choice of organizing schemas of the story. 51 Issue-specific and Generic News Frames News frames could be further distinguished between issue-specific and generic news frames (de Vreese, Peter, & Semetko, 2001). Issue-specific frames are constructed to describe a particular issue or event and are highly contextually bound. Issue-specific frames are context-bound (Matthes, 2009). de Vreese et al. (2001) stated that issue-specific frames provide great specificity and detail in their coverage of a certain issue, and ―capture specific aspects of selection, organization, and elaboration that pertain to a well-defined issue‖ (p. 108). For example, Wu (2006) analyzed and compared the news coverage of HIV/AIDS in China by the Xinhua News Agency of China and the Associated Press of the United States in 2004. Wu found that the two leading news organizations used divergent frames to cover this issue, and these frames are specific to the issue of HIV/AIDS. While an overarching antigovernment frame was dominant in the Associated Press‘s report, Xinhua News Agency adopted a progovernment frame. Three subframes, that is, the dishonesty/oppression frame, the human rights abuser frame, and the incompetence frame, have combined to support the overarching antigovernment frame. Using these subframes, the AP discourse portrayed the Chinese government and its officials as dishonest and inefficient in addressing China‘s HIV problem. Xinhua news agency also applied three distinctive subframes to support its progovernment frame, that is, the defense frame, the progress frame, and the ambivalence/ambiguity frame. The defense frame highlighted the Chinese government‘s open attitude, concrete action, and repeated commitment in addressing the HIV/AIDS problem in China; the progress frame 52 ―cast an optimistic picture of China‘s fight against the deadly epidemic‖ and focused on the change that the Chinese society had experienced in fighting the epidemic (p. 265); and the ambivalence frame demonstrated the Chinese government‘s cautiousness, ambivalence, and ambiguity in revealing statistics and information about the extent of this epidemic. Along these lines, Hong‘s (2013) content analysis revealed that the media coverage of recalled Chinese products in 2007 differed significantly between two Chinese media and two U.S. media. They adopted different issue-specific frames. While the Chinese media primarily relied on ―Not to Fear‖ frames to provide reassurance of safety and demonstrate the Chinese government‘s efforts to increase products‘ quality and product controls, the American media used ―Broken System‖ and ―Worries about Recall‖ to blame China and emphasize the threat of recalled Chinese products. In addition, Menashe and Siegel (1998) conducted a systematic analysis of the predominant framing used by the tobacco industry in arguing tobacco-related public policy issues over the past decade. They found that the tobacco industry used consistent dominant frames that appeal to deeply ingrained American values and principles such as freedom, autonomy, individual rights, economic opportunity or livelihood, capitalism, and so forth. For example, content analysis revealed the prevalence of a ―positive economic force‖ frame that highlights the tobacco industry had boosted the economy and provides thousands of jobs to the public. A ―just doing business‖ frame was identified in 18 articles which emphasized that antismoking advocates are zealots whose real motive is to prohibit tobacco entirely. An 53 accommodation frame identified in 17 articles argued that the public must accommodate all persons, smokers and nonsmokers. By contrast, tobacco control advocates frequently used dominant frames to appeal to the core value of health that included deceit/manipulation, nonsmokers‘ right, kids, and killers. For example, a deceit frame highlights the fact that people are deceived by the tobacco industry in its advertising to believe that smoking is not harmful and are manipulated to smoke. On the contrary, generic frames are ―broadly applicable to a range of different news topics, some even over time, and potentially, in different cultural contexts‖ (de Vreese et al., 2001, p. 108). Generic frames allows the possibility of ―comparisons between frames, topics, and potentially, framing practices in different countries‖ (de Vreese et al., 2001, p.108). Researchers have identified a number of generic news frames in the coverage of political and social issues. For example, Semetko and Valkenburg (2000) identified the prevalence of five generic news frames in previous studies on framing: Conflict, human interest, economic consequences, morality, and attribution of responsibility. Their content analysis was based on more than 4,000 newspaper and television news stories from Dutch national television news in the period surrounding the Amsterdam meetings of European heads of state in 1997. The results indicated that the attribution of responsibility frame was the most prevalent frame used in covering political stories, followed by the conflict, economic consequences, human interest, and morality frames, respectively. Specifically, an attribution of responsibility frame presents an issue by ―attributing responsibility for its cause or solution to either the government or to an individual or group‖ (p. 96). A 54 conflict frame describes ―electoral politics as a game, favoring events that clearly pit candidates against each other, emphasizing conflicts regardless of whether events themselves clearly suggest it‖ (Price et al., 1997, p. 484). Third, the economic consequences frame reflects a ―preoccupation with the ‗bottom line,‘ profit and loss‖ (Neuman et al., 1992, p. 63). This type of frame focuses on the economic impact or other likely effects on stakeholders involved. Fourth, a human interest frame approaches the news story via a personal angle and focuses on ―candidates and other political figures as personalities‖ such that issues or processes might not be ―abstract‖ (Price et al., 1997, p. 484). And finally, a morality frame ―puts the event, problem, or issue in the context of religious tenets or moral prescriptions‖ (p. 96). Morality framing of stories contain moral messages, make reference to morality, God, and other religious tenets, and offer specific social prescriptions about how to behave. A number of other studies have also found that news about politics and the economy are often covered with these generic frames (de Vreese et al., 2001; Han, 2007; Yun, Nah, & McLeod, 2008). These studies demonstrate that generic frames are examples of ―a more generic conceptualization of a kind of news frame that has the capacity to transcend issue, time and space limits‖ (de Vreese, et al., 2001, p. 109). Another typology of generic frames, mainly episodic and thematic frames, are also common news frames in the television coverage of political and social issues such as racial inequality, crime, poverty and unemployment (Iyengar, 1990). Many researchers pointed out that the dominance of episodic frames over thematic frames in television 55 coverage has impoverished political discourse in the United States. Because episodic frames focus on the internal personal deficiencies of the victim rather than analyzing the social factors underlying a problem, leading the public to regard individuals as primarily responsible for what happened to them. The sections below elaborate on the definitions of episodic and thematic frames as well as the effects of these frames on attribution of responsibility. Episodic and Thematic Frames Iyengar‘s episodic and thematic frames are prime examples of generic news frames (de Vreese, 2005; de Vreese, et al., 2001; Matthes, 2009). They are two fundamental types of frames that appear across issues, time, and space in political news communication (Aarøe, 2011). While episodic frames focus on personal stories or case studies, thematic frames present an issue in a broader context and provide abstract and collective evidence (Iyengar, 1990; 1991). Journalists have employed these two generic frames in the coverage of a wide range of social and health issues, such as unemployment, poverty, cancer, obesity, and medical errors (Kim & Willis, 2007; Lawrence, 2004; Iyengar, 1991; Major, 2009, etc.). Episodic frames present an issue to the audience by ―offering a specific example, case study, or event oriented report‖ (Gross, 2008, p. 171). Thematic frames, on the contrary, place issues into a broader context and provide a landscape view of the issue by analyzing underlying social and political factors. Episodic coverage provides little insight into the social structural factors underlying a problem, 56 leading the public to perceive individuals as blameworthy for their own problem, while thematic coverage calls attention to the role of the society and governmental agencies in causing the problem, and helps gulvanizing changes to public policies contributing to the problem. Therefore, for example, episodic framing of unemployment depicted the plight of an unemployed person, focusing on the impacts of unemployment on his or her family and life, whereas thematic framing would provide unemployment rate statistics and commentaries from governmental officers about the causes and consequences of unemployment. Episodic and thematic framing has been found to influence people‘s attribution of responsibility both for causing and treating policy problems, including public health (Jarlenski & Barry, 2013). Iyengar‘s study (1991) found that thematic framing of poverty tended to shift the public‘s assignment of responsibility from the individual victim to the larger social context. When poverty was depicted as a collective outcome, the thematic frame tended to engender a sense of societal responsibility. By contrast, constructing poverty around a specific poor person encouraged blaming that individual for the problem. In the domain of health news framing, a large number of extant research has demonstrated that episodic news framing has been predominant over thematic framing, reducing important health issues to individual-level problems (Kim & Willis, 2007). A person‘s health status is a function of a number of individual and societal factors, and cannot be understood as only being caused by deficiencies and flaws in individual health behaviors or biological makeup. Important social and environmental conditions, such as 57 unequal distribution of economic resources, unsafe environments, or unethical business practices, among other things, contribute significantly to individual health problems (Kim & Willis, 2007). In general, prior studies have found that episodic framing of health issues increases individual causal attribution and leads the public to focus change efforts mostly on modifications of individual behaviors. By contrast, thematic framing promotes societal causal attribution and highlights the importance of collective actions and policy responses in addressing the problem. For example, Major (2009) examined how the intersection of gain/loss frames with episodic/thematic frames influenced respondents‘ attribution of individual and societal responsibility for lung cancer and obesity. Specifically, in the episodic gain lung cancer story, Major depicted ―a man‘s successful battle with lung cancer and his advocacy work for lung cancer prevention‖ (pp. 178-179). By contrast, the episodic loss condition depicted that ―the diagnosis for the man is grim‖ (p. 178-179). For the thematic gain lung cancer story, Major discussed a study which demonstrated that the implementation of the state‘s smoking ban contributed to the improvement of lung functioning. And in the thematic loss lung cancer story, failure to enact smoking bans were associated with the harmful effects of secondhand smoke exposure. Four conditions of obesity were manipulated in a similar way. Their results indicated that a thematic loss frame about lung cancer and obesity significantly increased people‘s attribution of responsibility for the problem to the society at large. Shen, Lee, and Hu (2012) found that participants exposed to episodic framing of obesity, as compared to those who read a 58 thematic frame of the problem, were more likely to assess the message positively and reported more individual-based causal attributions, individual causal beliefs but fewer societal causal beliefs. In a similar vein, Coleman, Thorson, and Wilkins (2011) found that a thematic frame about obesity (compared to an episodic one) led audiences to be more supportive of changes in public policies. A similar concept of individual versus societal framing can also be understood to ―map well onto culpability and controllability‖ in the context of health (Riles et al., 2015, p. 1023). In the context of the U.S. obesity epidemic, a substantive body of research suggests that individual and societal frames have significant influence on the public‘s perception of who should be responsible for causing and treating obesity. Researchers have generally agreed that an individual frame which places the burden of addressing the problem primarily on individuals has taken precedence over a societal frame in national discourse on obesity and a number of other diseases (e.g., Dorfman et al., 2005; Gollust & Lantz, 2009; Hawkins & Linvill, 2010; Yoo & Kim, 2012; Kim & Willis, 2007; Lawrence, 2004). For example, Sun et al. (2016) explored how an individual and a societal frame of obesity affected individuals‘ causal and treatment attributions, which in turn influenced three types of responsibility-taking behaviors at three levels. They defined personal responsibility taking as behaviors performed to maintain individual health and are under the control of the individual agent. Interpersonal responsibility taking refers to behaviors with the intention to help others improve their health such as ―sharing health information with family, friends, or colleagues‖ (p. 142). And social 59 responsibility taking refers to ―civic, participatory actions taken to improve the sociopolitical environment and benefit a larger community or population‖ (p. 142). Their results indicated that the societal frame led the participants to attribute the problem to social conditions, and correspondingly, to hold the government, food industry, and marketing sector responsible for addressing obesity. Respondents‘ belief in the primary responsibility of the social sectors also increased their likelihoods of interpersonal and social responsibility-taking behaviors as well as personal behaviors. Along these lines, Clarke and Everest (2006) identified three frames for the discussion of health and disease: Medical, political/economy and life style, and each of them ―is associated with a different level of disease controllability and all three isolate the root cause of a disease and characterize how people contract it‖ (Riles et al., 2015, p. 1022). A medical frame describes health problems such as cancer as ―biologically based pathologies originating in the malfunctioning of the genes, cells and organs in the individual body‖ (Clarke & Everest, 2006, p. 2592). This perspective treats the body as a machine that when broken needs be fixed or replaced. The lifestyle frame attributes responsibility for contacting the disease to the ―individual choices to engage in unhealthy behaviors such as diet, smoking, alcohol consumption and sexual promiscuity‖ (p. 2592). Both the lifestyle frame and the medical frame tend to blame the individual for outcomes and ―tends to emphasize cure, forecloses on broader understandings of causation, prevention, and possibilities for the promotion of health‖ (p. 2598). By contrast, the political/economy frame is similar to the concept of a societal frame as it focuses on 60 external factors beyond an individual‘s control such as social structural inequalities. The unquestioning preeminent position of a medicine perspective in news framing of cancer and other diseases has been substantiated by other studies. For example, Clarke and Binns (2006) found that the medical frame was most prevalent by lifestyle and social structural frame in the media‘s depictions of heart disease. Pan and Meng (2016) found that the medical/scientific frame was preeminent at the postcrisis stage of the swine flu crisis which highlighted medical treatment and scientific findings in dealing with the epidemic. Mediators and Moderators of News Framing Effects In general, framing researchers have identified three psychological processes to explain news framing effects: accessibility effects, applicability effects/belief importance change and belief content change. Early framing research used the accessibility model as an explanation. Its major premise is that exposure to a frame makes certain considerations or cognitions more accessible, so that they are more likely to be used in subsequent judgment making or opinion formation. Theorists upholding this view argued that individuals possess limited cognitive processing abilities and are more likely to attend to those concepts and ideas that are on top of their mind when making political reasoning (Nelson & Oxley, 1999). Kinder and Sanders (1996) suggested that framing is not different from priming, as both involve the temporary activation of certain considerations and concepts. ―As particular frames rise to prominence, some opinion ingredients are 61 highlighted and made more accessible while others are shunted to the side‖ (p. 174). When forming opinion about a political issue, an individual will not retrieve all the cognitions and ideas stored in long-term memory; rather, only a limited set of considerations that have been primed can rise to the top of the head (Zaller, 1992). Despite that many early framing studies employed the accessibility model as an ―established psychological law and use it analogously to explain framing effects‖ (Pan & Kosicki, 2005, p.181), a majority of researchers stated that news framing involves more than accessibility effects and should be distinguished from other media effects such as agenda setting and priming. Both agenda setting and priming are based on memory-based models of information processing which posit that people form opinions and make judgments about a political issue based on considerations and ideas that are most salient or accessible (Scheufele & Tewksbury, 2007). The news media not only can make certain issues more salient in people‘s mind (agenda setting), but also can alter the standards with which people evaluate political candidates or issues via making certain considerations or aspects of the issue more salient than others (priming). Both agenda setting and priming focus on ―what is reported in the media‖ (Pan & Kosicki, 2005, p. 176) and are essentially accessibility effects. News framing, on the other hand, should be distinguished from these accessibility-based models of media effects (Nelson, Clawson, & Oxley 1997a, 1997b; Pan & Kosicki, 2005; Price et al., 1997; Scheufele & Tewksbury, 2007; Slothuus, 2008). For example, Pan and Kosicki (2005) pointed out the ―accessibility bias‖ wherein more accessible cognitions have a greater likelihood of being activated, should be 62 distinguished from cognitions‘ actual activation and application in social judgments (p. 181). In a similar vein, Price et al. (1997) conceptualized news media framing as an applicability effect which occurs during or immediately following message processing. They differentiated an applicability effect of the news frame from its accessibility effect which occurs at some later point in time. News framing effect occurs when salient features of a news message ―render particular thoughts applicable, resulting in their activation and use in evaluations‖ (p. 486). This effect is different from an accessibility effect, which refers to the residual activation potential of ideas and feelings activated during message processing (i.e., news framing). Specifically, the applicability effect of a news frame is conceptualized as ―a kind of storable relational knowledge defining which beliefs are relevant to one another‖ (Baden & Lecheler, 2012, p. 366). In a similar vein, according to Scheufele and Tewkesbury (2007), a news frame could influence opinion and attitude by suggesting a ―connection between two concepts such that, after exposure to the message, audience accept that they are connected‖ (p. 15). For example, news coverage of the enlargement of the European Union (e.g., de Vreese, 2004), if framed in terms of an economic consequence frame, would suggest a connection between economic implications and considerations and support for the enlargement of the European Union. The economic consequence frame may suggest that the best way to think about whether or not the enlargement of the European Union is desirable is through a consideration of the costs, benefits, and financial ramifications of the issue. Thus, the news frame has said that considerations 63 about economic implications are applicable to questions about policy support. By contrast, a conflict frame may suggest a connection between public and political friction over the issue and support for the EU enlargement. The conflict frame may suggest that controversy and diverging aspects between conflicting parties are important and relevant in thinking about the level of support for the EU enlargement. Thus, the news frame has said that considerations about controversy and disagreement are applicable to questions about policy support. Along these lines, Nelson and colleagues (Nelson, Clawson, & Oxley, 1997a, 1997b; Nelson & Oxley, 1999) conceptualized news framing effects as a result of belief importance change, where news frames affect opinion by ―selectively enhancing the psychological importance, relevance, or weight accorded to specific beliefs with respect to the issue at hand‖ ( Nelson & Oxley, 1999, p. 1043). Studies conducted by Nelson and colleagues supported the psychological model of belief importance change. For example, Nelson, Clawson, and Oxley (1997a) explored how the presentation of the Ku Klux Klan rally using a freedom of speech frame or a public order frame affected respondents‘ tolerance of their activities. Specifically, they assigned half participants to a reaction time task and the other half to an importance- rating task. They used the reaction-time task to test whether or not the cognitive accessibility of the concepts related to free speech and public order would be affected by the two frames. The logic of the reaction-time task is that concepts that are made more accessible by a news frame should be recognized more quickly than those that were less accessible. Results indicated that participants did not 64 differ in their response time. That is, the free speech and public order frames did not increase the speed with which participants recognized words related to these two frames. Thus, the accessibility hypothesis was disconfirmed. Instead, results of the importance rating task showed that participants in the freedom of speech condition rated the free speech value as significantly more important in evaluating their tolerance level of the rally. On the other hand, the ―public order‖ frame might have led participants to think about the likelihood that Ku Klux Klan rallies could descend into violent riots, thus increasing the importance of this consideration in their evaluation of their tolerance level. To summarize, their study showed that the two frames affected respondents‘ tolerance levels ia suggesting to them that the best way to think about Ku Klux Klan rallies was through a consideration of whether one values public order or freedom of speech. Recently, scholars have turned to a model of belief content change as another possible mediator for framing effects (Baden & Lecheler, 2012; Lecheler & de Vreese, 2012; Slothuus, 2008). Chong and Druckman (2007) stated that news frames often ―introduce new considerations about a subject in addition to highlighting existing beliefs‖ (p. 116). Thereby, a news frame not only can change perceived importance of existing beliefs, but also can change the beliefs themselves by offering new considerations to an individual. de Vreese, Boomgaarden, and Semetko (2011) argued that news framing effects are considered indirect via belief importance change and direct via belief content change. 65 The extent to which belief importance change or belief content change takes effect would depend upon moderators. The existing literature has identified a number of individual-level moderator variables that condition news framing effects. For example, Slothuus (2008) proposed a dual-process model of news framing effects that integrates both the how questions concerning mediators of news framing and the who questions concerning the moderators of news framing. The model posits that news frames affect opinion via different psychological routes depending on individual characteristics such as political awareness and political value. In their study, they used a ―job frame‖ and a ―poor frame‖ to describe a welfare policy problem. The ―job frame‖ emphasized that the purpose for the government to cut welfare benefits was to motivate unemployed people on welfare benefit rates to find a job. By contrast, the ―poor frame‖ framed the message by emphasizing that such a decision would not increase employment but would only create more socially marginalized people. Participants were randomly assigned to read one of the three conditions (i.e., two news frames and the control group). Participants were then measured on importance of issue-relevant considerations by rating five different considerations related to the two news frames and/or relevant to the issue (e.g., ―government expenditures on welfare benefits should not be too expensive‖; ―No defrauder should receive welfare benefits‖). The content of such beliefs or considerations with relevance to the issue was also measured and was categorized into internal attribution items (e.g., ―. . . because they lack proper moral standards and ability to pull themselves together‖) and external attribution items t (e.g., ―. . . because today business 66 and industry only employ high-efficiency labor‖). Results indicated that for individuals with a moderate level of political awareness or weak political values, both belief content change and belief importance change mediated the effects of the news frame. On the other hand, for individuals with a high level of political awareness, the news framing effect was primarily mediated via belief importance change. In addition, there was minimal news framing effect for the least politically awareness individuals or people with strong political values. Among the individual-level moderator variables, political knowledge has been identified as a ―dominant moderator of susceptibility to framing effects‖ (Lecheler & de Vreese, 2010, p. 77). Work on political knowledge as a moderator variable has produced conflicting results. One group of framing studies found stronger news framing effects on low knowledge individuals (Hyunseo, Gotlieb, Seungahn & McLeod, 2007; Kinder & Sanders, 1990; Lecheler & de Vreese, 2010; Schuck & de Vreese, 2006); whereas a second group found the opposite (Krosnick & Brannon, 1993; Nelson, Oxley, & Clawson, 1997b). Researchers from the former group argue that low knowledge individuals are unable to counter-argue the framed message due to their lack of issue relevant knowledge and considerations. In addition, they pointed out that low knowledge individuals tend to have weak attitude regarding a given issue, which makes them more susceptible to news framing manipulation. Researchers from the latter group state that politically knowledgeable individuals are more susceptible to news framing effects, because they are more likely to be equipped with a wider variety of relevant considerations and a higher 67 level of comprehension for frame-related concepts than low knowledge people. Since mediation via belief importance requires the availability of issue-relevant considerations and beliefs, high- knowledge people can ―process and integrate framed considerations more quickly and efficiently‖ than low-knowledge people equipped with a smaller set of considerations (Lecheler & de Vreese, 2010, p. 153). Lecheler and de Vreese‘s (2012) empirical study supported the notion that a solid stock of knowledge and considerations on a given issue would facilitate the understanding and processing of a news frame, which would result in larger news framing effects. They conducted an experimental study to examine how news framing of the enlargement of the European Union in light of the accession of Bulgaria and Romania in 2007 would affect individual opinion toward the economic development of the EU‘s two newest members. The issue was framed in terms of an ―opportunity‖ economic consequence frame or a ―risk‖ economic consequence frame. The ―opportunity‖ condition emphasized the opportunities Bulgaria and Romania presented to the EU market, whereas the ―risk‖ condition pointed out the risks the two EU members posed for the EU market. Participants were randomly assigned to one of the two experimental conditions or the control group. After that, participants were measured on belief importance and belief content related to the issue. The analysis of the moderating influence of different levels of political knowledge indicated that participants with higher levels of political knowledge were more susceptible to news framing effects. They were influenced to a greater extent via both belief content and belief importance change than 68 low knowledge people. Specifically, they found belief importance as a significant mediator of news framing effects, supporting the traditional conceptualization of news framing as changing perceived importance of existing considerations within an individual‘s mind. A more interesting finding, as they pointed out, was that belief content emerged as a significant mediator of framing effects. Their mediation analysis indicated that belief content was the more prominent mediator. To explain why belief content generally prevailed over belief importance in mediating the news framing effect, they stated that the issue framed in their study was invisible and remote to most of the public, so even politically knowledgeable individuals did not possess a solid stock of available considerations related to the issue and were thus affected via news beliefs. Along these lines, a number of other studies also found that ―the conditionality of mediated framing effects may vary across issues,‖ depending upon how important or familiar an issue is to an individual (Lecheler & de Vreese, 2012, p. 196). For example, Lecheler, de Vreese and Slothuus (2009) examined the extent to which news framing effects differ in magnitude as well as process, depending on how important an issue is. Importance is defined as ―the concern, caring, and significance an individual attaches to the attitude object, the issue of a news frame‖ (p. 403). Their results indicated that there were no news framing effects for the high-importance welfare issue, whereas there were significant framing effects of the low-importance trade issue. Moreover, in accordance with their expectations, the framing process for effects regarding the low-importance issue was mediated by belief content change. They argued that ―if an issue is unimportant, 69 an individual is less likely to be motivated to acquire attitude-relevant knowledge about this issue‖, thus, frames can affect opinion by adding new considerations and beliefs, ―rather than merely endowing some considerations with greater relevance‖ (p. 416). In addition, issue familiarity has emerged as an important moderator of news framing effects. For example, Han, Chock, and Shoemaker (2009) investigated how a ―game‖ frame and a ―military-consequences‖ frame of Taiwan‘s 2004 presidential election campaign would affect both U.S. and Chinese audiences‘ perception of the event and their attitudes towards Mainland-Taiwan relations. More importantly, they examined the moderating function of issue familiarity on the cognitive and attitudinal effects of the two news frames. In their study, familiarity with the Taiwan election campaign was operationalized by measurement of topical knowledge and personal relevance. Their results indicated that, compared to U.S. participants, Chinese participants were significantly more knowledgeable about Taiwan‘s political issues and also perceived the election as having significantly greater personal relevance. Therefore, it was found that for Chinese participants increasing the intensity of news frame had less impact on their perception and attitude. By contrast, increasing the intensity of news frame had a significant impact on U.S. participants such that even a limited amount of news framing of the issue produced significant attitudinal changes. To summarize, extant literature shows that news frames work via various psychological processes, depending on a number of individual-level moderator variables such as political knowledge, political value, and political awareness as well as message 70 characteristic variables such as issue importance and issue familiarity. Though these processes designate distinctive psychological processes of news framing effects, researchers argued that they might operate simultaneously or be complementary to each other. For example, Scheufele and Tewksbury (2007) stated that ―accessibility and applicability cannot be completely isolated from one another‖ (p. 16). Similarly, Baden and Lecheler (2012) pointed out that the availability and accessibility of certain knowledge and considerations are the preconditions for a news frame to manipulate the applicability of those cognitions. That is, an applicability effect would depend on ―the presence of those frames in audience members‘ existing knowledge or the content of the message‖ (Scheufele & Tewksbury, 2007, p. 16). Chong and Druckman (2007) summarized that ―framing can work on all three levels, by making new beliefs available about an issue, making certain available beliefs accessible, or making beliefs applicable or strong in people‘s evaluations (p.111). Therefore, the three mediating processes involved in news framing processing ―operate complementarily, each contributing to the total framing effect‖ (Baden & Lecheler, 2012, p. 362). Summary Health problems have been extensively covered in the media and a disproportionate amount of research has been focused on the effects of equivalence framing (i.e., gain- vs. loss-framed health messages). These studies focused on the relative persuasiveness of gain and loss frames in terms of compliance with recommended health behaviors, as 71 previously discussed. While this line of framing research provides valuable insights, the effects of frames other than equivalency/valence frames on health decision-making, opinions and behaviors need be examined. Research on news framing of health issues would shed light on how different sets of considerations highlighted in the health message could bring forth different perspectives in construing a given health issue and subsequent attitudinal and behavioral changes (Sun et al., 2016). A growing number of studies show that the news media serve an integral function in influencing how the public thinks about a health issue. News framing in the domain of health and medicine has been operationalized in various ways in the literature, and there are a variety of news frames in the discourse of health communication, such as episodic versus thematic frames, individual vs. societal frame, a typology of medical/scientific, political/economy, and lifestyle frames, and so forth. Research on health news framing have examined how news frames of health issues impact individuals‘ perception and opinions regarding a given health issue. This body of research, in general, has found that an episodic or individual, frame of health issues increases attributions of responsibilities, both for causing and treating the problem, to individuals, whereas a thematic or a societal, frame stresses the society‘s responsibilities in addressing the problem. There is a growing body of research indicating that episodic and thematic news frames not only have divergent effects on cognitions such as attributions of causal and treatment responsibilities for an issue, but also would introduce different emotional reactions (Gross, 2008; Major, 2009). This project adds to this line of research by 72 focusing on the differential emotional reactions these two generic frames would induce and the influence of elicited discrete emotions on subsequent outcomes. News framing effects research has traditionally focused on psychological models of framing processes. Thus, an exploration of the effects of discrete emotional reactions in news framing research would contribute to a better understanding of how news frames influence attitudes and opinions. Therefore, the next chapter provides a discussion of whether and how emotional reactions contribute to news framing effects. CHAPTER 5 FRAMING EFFECTS ON EMOTIONS The previous chapter revealed that prior studies of news framing effects have been predominantly confined to cognitive processes, failing to systematically examine the effects of emotional responses to a news frame. This is not surprising, given that affective influences on political opinion formation have traditionally been disregarded in key theories regarding the effects of mass media (Kühne, 2012a). Studies addressing the effects of media-induced emotions on opinions are still scarce in communication science and political science fields (Kühne, 2012a). Recently, a significant body of research has emerged to examine the persuasive impacts of message-induced or issue-relevant emotions on subsequent opinions and judgments. This body of work provides important insights into the role of discrete emotions in the news framing process. For example, Nabi‘s (1999, 2002a, 2002b, 2003, 2007) research is path-breaking in this area. She proposed the cognitive functional model (CFM) that illustrated how ―discrete, message-induced negative emotions may direct information processing and subsequent attitude change and information recall, particularly when the emotion aroused 74 is substantively linked to a message‘s focal topic‖ (1999, p. 293). Based on appraisal theories of emotion, she argued that persuasive messages could produce emotional reactions in recipients, which in turn could influence the persuasive outcomes. Specifically, centering around the notions of motivated attention and motivated processing, she elaborated on how the emotion type, expectation of the message containing reassuring information or not, presence of peripheral cues, argument strength, and cognitive ability, would mediate both the mode of processing (central vs. peripheral) and the outcomes of persuasion (i.e., message acceptance or rejection). Nabi also formulated a three-step theoretical model to explain the emotional processes that underlie news framing effects on judgments and opinion formation (2003, 2007). First, when the processing of a news frame produces a specific interpretation or meaning of an issue, which in turn activates a specific appraisal pattern, the corresponding discrete emotion is elicited. Second, topic-specific emotions should lead to selective attention, processing, and accessibility of information. In other words, information relevant to the alleviation of the negative emotion will be more readily retained in memories and thus become more accessible for subsequent judgments. In this vein, message-induced emotional reactions could be regarded as similar to an individual‘s cognitive frame that influence information processing and attitude formation. Finally, the model postulates that emotion-biased information processing and accessibility lead to emotion-congruent judgments and attitudes. 75 Corresponding theoretical propositions have been brought forward by other scholars. For example, Kühne (2012b, 2014) outlined a three-step emotional framing model that integrated cognitive framing effects models (accessibility and/or applicability effects) and appraisal theories of emotion. First, news frames make certain interpretation or cognitive appraisals more accessible or applicable. Second, the specific appraisal triggers a corresponding emotion. Third, the induced emotion promotes emotion-congruent judgment and opinion formation, as numerous psychological studies have found that emotions influenced judgments (e.g., Blanchette & Richards, 2010; DeSteno, Petty, Rucker, Wegener, & Braverman, 2004; Lerner & Keltner, 2000, 2001). Along these lines, Nerb and Spada (2001) pointed out that news reports about environmental damages can produce negative emotions which in turn would lead to emotion-specific action tendencies. Accordingly, depictions of accidents in terms of degrees of responsibility can influence cognitive appraisals of human agency and controllability which in turn would produce emotional arousals such as anger and sadness. Finally, the emotional responses can motivate individuals to take a certain course of action. For instance, anger was found to be a potent predictor for action tendencies to boycott the responsible agent. The above review of extant theoretical models revealed two major points in understanding the effects of message-induced emotions. First, appraisal theories of emotion (e.g., Winterhoff-Spurk, 1998; Ellsworth & Scherer, 2003) have been prevalent in explaining emotional reactions to news framing. The current project also builds on this theoretical framework to argue that individuals will experience emotional responses to 76 anger- or fear-inducing news stories of medical errors. Second, these scholars have primarily drawn on prior social psychological research on incidental emotions—emotions unrelated to a judgment or decision (e.g., Han et al., 2007; Lerner & Keltner, 2000, 2001; Lerner & Tiedens, 2006)—to argue that emotions should be taken into account in media effects research. Because affective influences have traditionally been disregarded in key media effects theories, empirical studies investigating the influence of message-induced emotions on opinion are still scarce in communication science and political science fields (Kühne, 2012a). Nonetheless, a significant body of social psychological studies has focused on incidental emotions, which clearly demonstrated that incidental emotions could impact opinion and judgment formation (Lerner & Keltner, 2000, 2001). That is, emotion-specific effects on opinion have predominantly been investigated with regards to incidental emotions. Along these lines, the current project also draws on extant literature on incidental emotions to explain the effects of message-induced emotions on attitude and opinion. More specifically, Lerner and Keltner‘s (2000, 2001) appraisal tendency framework (ATF), is adopted to explain emotion-specific effects on judgment and opinion formation (i.e., appraisal tendencies). ATF explains how emotions affect judgments in a way that directly corresponds to the appraisal dimensions at hand (Han et al., 2007). 77 Emotional Reactions to News Content This section first defines and differentiates affect, mood, and emotions. It then briefly reviews the key premises of cognitive appraisal theories of emotion. Relevant empirical findings on emotional reactions to news coverage are incorporated. Key appraisal dimensions associated with anger and fear are discussed. At the end of this section, the emotion-inducing power of episodic versus thematic frames is compared, followed by the first two hypotheses of the study. Affect, Mood, and Emotions The terms affect, mood, and emotion are sometimes used interchangeably in the literature. Yet researchers have worked on their conceptual clarity and differentiation (e.g., Batson, Shaw, & Oleson, 1992; Dunn & Schweitzer, 2005). Affect is a ―general nonspecific term that includes all the foregoing motivational states and processes,‖ the affective domain encompasses the fundamental emotions, patterns of emotions, drives and their interactions‖ (Izard, 1991. p. 55). Mood and emotions are specific types of affective states or affective phenomena (Dunn & Schweitzer, 2005). Mood and emotions can be differentiated along three dimensions: duration, intensity, and specificity (Kühne, 2012a). Whereas moods are ―low in intensity, high in longevity, and do not refer to an object,‖ emotions are ―high in intensity, low in longevity, and refer to an object‖ (Kühne, 2012b, p. 3). Emotions are generally ―short-lived, intense, and directed at some external 78 stimuli‖ (Nabi, 2003, p. 226). Moods are more diffused whereas emotions concern object relations and reflect an individual‘s specific goal. Cognitive Appraisal Theories of Emotion Cognitive appraisal theories of emotion (e.g., Fridja, 1986; Keltner et al., 1993; Lazarus, 1991; Roseman, 1984; Roseman et al., 1990; Smith & Ellsworth, 1985; Tiedens & Linton, 2001) are the dominant paradigm in the emotion literature. Appraisal theories provide the theoretical underpinnings as to why there are emotional reactions to news stories such as car accidents, alcohol-related crimes, and medical errors. The central argument of appraisal theories states that emotions result from individuals‘ subjective evaluation (appraisals or estimates) of an issue along several cognitive dimensions. For example, Smith, Haynes, Lzarus, and Pope (1993) identified six appraisal components in differentiating a discrete emotion: Two primary appraisals of motivational relevance and motivational congruence, which concern whether and how the situation is relevant to personal well-being, and four secondary appraisals of accountability, problem-focused coping potential, emotion-focused coping potential, and future expectancy. Whereas problem-focused coping potential refers to evaluation of one‘s ability to handle the emotional encounter, emotion-focused appraisal reflects one‘s confidence in psychologically adjusting to the encounter by altering one‘s desires, goals, or beliefs (p. 919). Smith and Ellsworth (1985) identified six orthogonal dimensions to differentiate emotional experience: pleasantness, anticipated effort, certainty, attentional activity, 79 self-other responsibility/control, and situational control. Roseman et al. (1990) differentiated emotions based on the following key components: Appraisals of situational state (motive-inconsistent/motive-consistent), motivational state (punishment/reward), probability (uncertain/certain), power (weak/strong), legitimacy (negative outcome deserved/positive outcome deserved), and agency (circumstances/other person/self). These appraisal theories all suggest that emotions should be distinguished at a more specific level than based merely on their positive or negative valence, and that factors such as human agency, certainty, and controllability are important in differentiating a certain emotion (e.g., Tiedens & Linton, 2001). Appraisals could be understood at two complementary levels of analysis: Appraisal components at a molecular level and a ―core relational theme‖ at a molar level (Smith & Lazarus, 1990, 1993). Each emotion is associated with a distinct ―core relational theme,‖ which is the central or core meaning associated with a specific emotion. First proposed by Lazarus (1991), a core relational theme is defined as a summary of the relational harm or benefit underlying a specific emotion. For example, the core relational theme for anger is other blame, and for sadness the core relational theme is helplessness/loss or others‘ suffering (Smith & Lazarus, 1993). A news story could be developed to induce a specific emotion by emphasizing information related to the emotion‘s core relational theme. For example, to induce anger about an organizational incident, Kim and Cameron (2011) emphasized the company‘s intentional wrongdoing in the anger-inducing news story. To 80 induce sadness, they focused on crisis victims‘ personal lives and suffering in the sadness-inducing story. Moreover, appraisal theories posit that when a certain emotion is produced, its associated key appraisal dimensions will be activated and thus become ―especially available for the interpretation of subsequent events‖ (Keltner et al., 1993, p. 740). Along these lines, Lerner and Ketlner (2000, 2001) proposed the appraisal tendency framework (ATF) that provides the theoretical basis for distinguishing the effects of discrete emotions on judgment and decision making. The ATF posits that emotions carry over from past situations to color future judgments and attitudes. The appraisal patterns associated with each emotion would trigger distinct ―appraisal tendencies,‖ which account for the effects of emotions on judgment and decision making in subsequent situations. For example, anger co-occurs with appraisals of individual control and tends to trigger continuing perceptions of such control in subsequent situations. A number of studies have found that incidental anger triggered attributions of individual blame and control (e.g., Goldberg, Lerner, & Tetlock, 1999; Keltner et al., 1993). Fear arises from and evokes appraisals of situational control and great uncertainty, which impact subsequent judgment and opinion in a corresponding way. Studies have found that fear tended to trigger pessimistic risk perceptions and increase risk-aversive choices (Lerner, Gonzalez, Small, & Fischhoff, 2003; Lerner & Keltner, 2001). Empirical findings on emotional reactions to news content. When applied to mediated communication, appraisal theories of emotion imply that news recipients would 81 experience message-induced, topic-relevant emotions after exposure to media content, as news contents and news story features can function as ―appraisal-pattern signifiers‖ (Nabi, 2003, p. 242). When a news story suggests a particular interpretation or meaning of an issue, a specific cognitive appraisal pattern will be activated, which in turn will elicit the corresponding emotion. This effect has been corroborated in some initial empirical studies. For example, Myers, Nisbet, Maibach, and Leiserowitz (2012) found that a public health frame aroused hopeful emotions about climate whereas a national security frame elicited anger, especially among an audience already doubtful or dismissive of the issue. Aarøe (2011) designed two pro and two con news frames (each with one episodic and one thematic frame) to cover the topic of ―24-year rule‖—a contested Danish law about immigrants‘ marriage patterns. Both the episodic and the thematic pro frames stressed that the law could damage immigrants‘ outdated, involuntary marriage patterns from a specific and a general perspective, respectively. Both the episodic and the thematic con frames stressed that the law represented an injustice against young people who already choose their spouse themselves, from a specific and a general perspective, respectively. The findings substantiated their emotional arousal hypothesis such that the episodic frames triggered significantly stronger pity, compassion, anger, and disgust than the thematic frames. Harth, Leach, and Kessler (2013) found that when fake newspaper articles emphasized the in-group responsibility for environmental damage, individuals produced guilt and anger; whereas articles that emphasized in-group responsibility for environmental protection induced pride. Similarly, Igartua, Moral-Toranzo, and Fernández (2011) found 82 that the delinquency frame focusing on the negative consequences of immigration produced more negative emotions of disgust, contempt, anger, shame, and fear than the positively framed economic contribution frame, which in contrast induced more positive emotions of interest, surprise, and happiness. Further studies demonstrate that multiple factors could moderate emotional reactions to news content. For example, Gross and D‘Ambrosio (2004) found that ideology and racial predispositions influenced the effects of two attributional frames about the 1992 Los Angeles riots on emotional reactions. In their study, the situational frame focused on unfavorable social conditions in southcentral Los Angeles, such as poverty and unemployment, which contributed to the riots. The dispositional frame attributed the cause of the riots to individual responsibility and criminality. They found that conservatives were more likely than liberals to experience anger after being exposed to the dispositional frame. Racially resentful participants were less likely to have disgust than their unresentful counterparts after reading the situational frame. Gross and Brewer (2007) found that the extent to which conflict framing of policy debates provoked anger and disgust was contingent upon participants‘ prior view about campaign finance reform. The more prior support there was for the losing side of the debate, the more anger and disgust they experienced when exposed to the conflict coverage. In addition, other factors such as need for affect—―the propensity to feel and experience emotions‖ (Lecheler et al., 2013, p. 203)—could moderate emotional reactions to media content. Along these lines, Maio and Esses (2001) defined need for affect as ―the general motivation of people to 83 approach or avoid situations and activities that are emotion inducing for themselves and others‖ (p. 585). They stated that emotions have more influence on perceptions, attitudes, and behaviors in people who have high need for affect than those that have low need for affect, as people with high need for affect may be more receptive to influences of emotions. Maio and Esse regarded need for affect as ―an important construct for understanding emotion-related processes‖ (p. 584) and relevant for any emotion-related research. According to appraisal theories, the type of emotions elicited by a news story depends on the news content. Given the context of this study and previous research on anger and fear, this study focuses on these two negative emotions. Both can be experienced as intense and negative, and can be clearly differentiated with respect to their emotivational goals and action tendencies. Fear induces the desire for protection and activates the action tendency to escape from the threatening agent while anger encourages retribution and mobilizes energy for lashing out (Nabi, 1999, 2003). Although both are negatively valenced, anger and fear have distinctive effects on perceptions and judgments, opinions and behaviors (Lerner & Keltner, 2000, 2001). Anger. A substantial amount of theoretical and empirical work has focused on defining anger and fear as two discrete emotions (Averill, 1982; Manstead & Tetlock, 1989; Nabi, 1999; Roseman et al., 1990; Scherer, 1984). The core relational theme of anger has been defined by Lazarus (1991) as ―demeaning offense against me and mine‖ (p. 222). Lerner and Tiedens (2006) pointed out that there has emerged ―a remarkably 84 consistent picture of anger‖ from the extant emotion literature (p. 117). In general, prior research has found that anger is an approach-related, high-action-potential emotion that mobilizes individuals against the source of that anger (Carver & Harmon-Jones, 2009). Researchers tend to agree that anger is the most potent organizer and energizer for behavior (Nabi, 1999), whereas most other negative emotions, such as fear, are characterized by an avoidance tendency (Watson, Wiese, Vaidya, & Tellegen, 1999). Anger is characterized by high other-person control/responsibility and is associated with ―strong attributions of human agency‖ (Smith & Ellsworth, 1985, p. 833). Anger is usually induced as a result of perceiving a negative outcome as being intentional (Smith et al., 1993; Smith & Ellsworth, 1985). Therefore, the blameworthiness of the causal agent is a key appraisal in eliciting anger (Lazarus, 1991; Thomas, McGarty, & Mavor, 2009). In addition, a number of other situations can cause anger. ―Interruptions of joy or interest, perceived external obstructions to the attainment of personally significant goals, being forced to do something against one‘s wishes or being taken advantage of ‖(Nabi, 1999, p. 297), just to name a few ―decisively aversive situations‖ (Berkowitz & Harmon-Jones, 2004, p. 107 ) that can lead to anger induction. Along these lines, anger has been found to be closely related to perceptions of illegitimacy, immorality, unfairness, or injustice (Berkowitz & Harmon-Jones, 2004; Mikula et al., 1998; Thomas et al., 2009). For example, Mikula et al. (1998) found that injustice or unfairness played a fundamental role in the elicitation of anger, and anger-provoking events were most frequently perceived as unfair. Those that are angered 85 make the judgment that the provoking situation is ―contrary to what ought to be‖ (Shaver, Schwartz, Kirson, & O'Connor, 1987, p. 1077). In Frijda‘s (1986) terms, ―An angering event is one in which someone or something challenges what ‗ought‘ to happen‖ (p. 199). Along these lines, Nabi (1999) pointed out that if a public service announcement emphasizes how convicted drunk drivers continued to carry a valid driver‘s license, viewers will perceive injustice, which in turn will provoke anger. Similarly, when a news story about medical errors emphasizes how convicted doctors continue to carry clean licenses, it is expected that participants will perceive injustice or immorality, which in turn will evoke anger. Both an episodic and a thematic news story can elicit anger if they contain information that reflects the theme of injustice. While an episodic frame portrays a patient‘s outrageous personal story to elicit anger, a thematic frame presents disturbing statistics and discussing the incompetent supervisory system to evoke anger. Fear. For fear, the core relational theme is ―the concrete and sudden danger of imminent physical harm‖ (Lazarus, 1991, p. 235). Fear has been found to be associated with high levels of uncertainty, low coping potential, and appraisals of other-responsibility/control (Roseman et al., 1990). Generally, it is aroused ―when a situation is perceived to be threatening to one‘s physical or psychological self and out of one‘s control‖ (Nabi, 1999, p. 297). For Nabi (2003), the core relational theme of fear can be as simply as ―potential threat‖ (p. 227). For example, a fear message about crime that focuses on potential threat can be expected to elicit fear, and a public service announcement about the dangers posed by drunk drivers can be expected to induce fear 86 (Nabi, 1999, 2002a, 2002b, & 2003). Moreover, there are many theoretical models in the persuasion literature that provide conceptualizations of the processes and mechanisms through which fear appeals influence the perceptions and decisions of recipients. These models provide theoretical underpinnings as to the decision-making process involved when message recipients decide whether or not to follow a particular recommended health behavior (Cismaru, Lavack, Hadjistavropoulos, & Dorsch, 2008). These models include Leventhal‘s (1970) parallel processing model (PPM), Roger‘s (1975) protection motivation theory (PMT) and Witte‘s (1992) extended parallel process model (EPPM). Although these theoretical models vary in the components of the threat and coping appraisal, they tend to agree that a recipient‘s perceived vulnerability to a certain threat and its perceived severity are the key components for eliciting fear (Nabi, 1999). Therefore, individuals will experience fear when they are exposed to a news story about the potential risks involved in seeking medical care (i.e., perceived vulnerability to errors). Both an episodic frame that focuses on horrible individual experiences and a thematic frame that provides in-depth analysis of the flawed health-care system are expected to elicit fear. Operationalization of anger and fear. A review of the extant literature revealed that researchers have gauged anger and fear using different measures (e.g., Dillard & Peck, 2000, 2001; Dillard, Plotnick, Godbold, Freimuth, & Edgar, 1996; Gordijn, Yzerbyt, Wigboldus, & Dumont, 2006; Iyer, Schmader, & Lickel, 2007; Kim & Cameron, 2011; Kühne & Schemer, 2015; Mackie, Devos, & Smith, 2000, etc.). A large number of studies 87 measured anger and fear using single-item measures by asking participants to indicate on a likert-type how much anger or fear they felt when reading the story or thinking about the issue (e.g., Fischhoff, Gonzalez, Lerner, & Small, 2005; Goodall, Slater, & Myers, 2013; Gross & Brewer, 2007; Iyer et al. 2007; Lecheler et al., 2013; Malhotra & Kuo, 2009; Nabi, 2002; Nerb & Spada, 2001, etc.). Another group of researchers have utilized factor analytic techniques to establish construct validity. Their attempts have resulted in the development of scales, each of which includes a slightly different set of items to measure anger and fear. For example, Gordijin et al. (2006) developed four items (i.e., angered, outraged, annoyed, and irritated) to measure to what extent participants felt angry about a proposal on a 9-point scale ranging from 1 (absolutely not) to 9 (absolutely). Their principal component analysis with varimax rotation showed that the anger scale included three items: angered, outraged, and annoyed (eigenvalue = 2.69; explained variance = 13.45%; Cronbach‘s α = 0.80). Mackie et al. (2000) asked participants to indicate to what extent the out-group made them feel angry, displeased, irritated, or furious (four items measuring anger), and worried, anxious, afraid, or fearful (four items measuring fear) on 7-point scales ranging from 1 (not at all) to 7 (extremely). A principal components analysis with varimax rotation showed that all fear items loaded strongly on fear (>.76) while all anger items loaded strongly on anger (>.62). Kühne and Schemer (2015) measured anger with three items (―anger,‖ ―furious,‖ ―annoyed‖) and fear with three items (―fear,‖ ―anxiety,‖ ―faint-hearted‖). Their exploratory factor analysis of the items revealed that the three 88 anger items and the three fear items formed two distinct constructs. The α reliability for the anger index was .94, and for the fear index was .86. Along these lines, Dillard and his colleagues (Dillard et al., 1996; Dillard & Anderson, 2004; Dillard, Kinney, & Cruz, 1996; Dillard & Peck, 2000, 2001; Dillard & Shen, 2006; Shen & Dillard, 2007; Yan, Dillard, & Shen, 2012, etc.) developed a set of close-ended scales to measure discrete emotional reactions on a 5-point response scale ranging from 0 (none of this emotion) to 4 (a great deal of this emotion). The anger and fear scales and their corresponding items were: anger (irritated, angry, annoyed, aggravated) and fear (fearful, afraid, scared). Confirmatory factor analyses were conducted for each of the studies in which the scales developed by Dillard et al. were used (Dillard & Kinney, 1994; Dillard et al., 1996; Dillard & Peck, 2000, 2001; Segrin & Dillard, 1991; Shen, 2005; Shen & Dillard, 2007; Yan & Dillard, 2010; Yan et al., 2012). All of these studies indicated that their emotion scales were unidimensional. As stated by Dillard and Shen (2006), the validity of these emotion scales is quite strong, as ―all of these same studies show unique and discriminable relationships for each of the scales with some antecedents such as cognitive appraisals or some outcome variable such as perceived message effectiveness‖ (p. 332). In terms of internal consistency reliability, typically, Cronbach's α for both their anger and fear scales were in the .80 to .90 range. For example, in the Dillard and Peck (2001) data, the reliabilities were anger (α .88) and fear (α .94). In the Shen and Dillard (2007) data, the reliabilities were anger (α .87) and fear (α .92). In the Yan et al. (2012) data, reliabilities for both anger and fear scales were 89 at .95 level. Test-retest reliabilities has not been evaluated for these scales, because they were designed to measure transient states (Dillard & Shen, 2006). Based on these scales developed and verified by Dillard and colleagues, Kim and Cameron (2011) constructed a 17-item measure to assess participant‘s emotional reactions on a 7-point scale ranging from not at all to very much: afraid, scared, fearful, angry, irritated, annoyed, aggravated, sad, dreary, dismal, surprised, startled, disgusted, sickened, revolted, agitated, and empathy. Four factors emerged from their factor analysis (anger, fear, sadness, and surprise). For anger and fear scales, their corresponding items were the following, anger (anger, irritated, aggravated, α .93) and fear (scared, fearful, afraid, α .97). In Appendix C, a summary of anger and fear scales is provided. The emotion-inducing power of episodic versus thematic frames. While news stories can elicit emotional reactions in general, existing literature indicates that some news stories elicit stronger emotional reactions than others when a certain type of news frame is applied, such as an episodic, a conflict, or a human-interest frame (Lecheler et al., 2013; Otieno, Spada, & Renkl, 2013). This study focuses on the relative emotion-engaging power of episodic versus thematic framing regarding anger and fear. Episodic and thematic frames are commonly classified as generic news frames that have been applied to report a wide range of issues (de Vreese, 2005; de Vreese et al., 2001; Matthes, 2009). While episodic frames focus on personal stories or case studies, thematic frames present an issue in a broader context and provide abstract and collective evidence (Iyengar, 1990, 1991). Extant literature has generally indicated that episodic 90 news stories elicited stronger emotional reactions than thematically framed ones. For example, Gross (2008) found that episodic storytelling about mandatory minimum sentencing increased participants‘ expressions of compassion and pity than thematic storytelling. Similarly, Aaroe‘s (2011) study about the Danish immigration law (―24-year-rule‖) found that episodic frames (either pro or con) generated stronger compassion, pity, anger and disgust than the thematic frames and the baseline condition. Concerning the reasons why episodic frames are generally more emotion-inducing, existing research indicates that it lies in their distinctive features. Episodic frames are characterized by salient ―human interest‖ details that provide the receivers with a real and specific ―face‖ on the presentation of an issue (Aarøe, 2011, p. 210; Gross, 2008, 172). On the contrary, thematic frames place issues and events in some general context and present impersonal figures and policy discussions. When exposed to thematic framing, the focal point of emotional reactions is more diffuse, due to lack of presence of a specific character. In addition, episodic news stories share important commonalities with sensational news stories, as argued by Cosand (2014), ―sensational slants are the basis of episodic frames‖ (p. 20). Sensational news stories tend to be oriented towards personalization and dramatization of events or issues, which have great potential to provoke sensory and arouse public emotions (Wang, 2012). To summarize, as constituting two ―intellectually distinct appraisal contexts‖ (Aarøe, 2011, p. 210), episodic and thematic frames could induce varying emotional reactions. Therefore, it postulates that an 91 episodic anger-inducing and an episodic fear-inducing news story about medical errors will elicit more anger and fear than their thematic counterparts. Hypothesis 1a: An episodic frame will elicit stronger emotional responses than a thematic frame. Hypothesis 1b: The difference between episodic and thematic frame on emotional responses will be moderated by emotional manipulation. Effects of Emotions on Judgments, Decisions, and Opinions Emotion-specific effects on attitude and opinion formation have been corroborated for both related and unrelated emotions. An emotion is considered related or integral when it is caused by the object being evaluated, whereas an emotion caused by a different object irrelevant to present judgments and choices is considered as unrelated or incidental (i.e., incidental, Kühne, 2012a, p. 2012; Lerner & Keltner, 2000, p. 474). In general, incidental emotions could be induced by instructing individuals to recall emotional life events or presenting them with emotional stories/vignettes. In social psychological research, researchers have predominantly focused on the puzzling carryover effects of incidental affect on judgments and attitudes, as this type of influence ―shows the fallibility of humans and tells an exciting, counterintuitive story‖ (Zeelenberg et al., 2008, p. 22). Though this body of research primarily tells the story of incidental affects, it has yielded ―substantial insight into the effects of affective states on attitudes‖ (Kühne, 2012a, p. 2012). 92 The past decades have witnessed a burgeoning body of research on the effects of integral emotions. Integral emotions have been induced by instructing individuals to think about certain emotional aspects of an issue (Lerner et al., 2003; Nabi, 2003; Small, Lerner, & Baruch, 2006) or by presenting emotional articles or news reports (Iyer et al., 2007; Kim & Cameron, 2011; Nerb & Spada, 2001). Message-induced emotions are integral emotions. Exposure to news stories of medical errors are expected to induce message-relevant anger or fear, and these two emotions are integral emotions as participants‘ opinions were assessed regarding the issue of medical errors. A number of prior studies have found that integral emotions could differentially influence perceptions and attitudes in an emotion-congruent way (e.g., Kim & Cameron, 2011; Kühne & Schemer, 2015; Nabi, 2003; Nerb & Spada, 2011). The question of whether incidental and integral emotions have the same persuasive impacts has remained largely underexplored. Kühne (2012a) provided some initial insights into this question. He stated that incidental and integral emotions should have similar impacts on opinion and judgment formation because the two kinds of emotions do not differ with regards to their experiential qualities such as thoughts, feelings, action tendencies, and emotivational goals (Roseman et al., 1994). Theoretically, incidental and integral anger, once elicited, both could give rise to appraisals of human agency and certainty. In other words, incidentally angry people are not different from integrally angry people with respect to attribution of causality, evaluations, and blame. Empirical findings did support this notion. For example, Keltner et al. (1993) found that compared to those 93 who were induced to feel sad, incidentally angry individuals regarded dispositional attributions as more likely and dispositional forces more responsible for an ambiguous social event. Similarly, Kim and Cameron (2011) used fictional news stories to elicit anger and sadness about an organizational crisis. They found that relative to sad individuals, angry individuals reported significantly more negative attitudes towards the company involved in the crisis. Therefore, the large volume of psychological research on the effects of incidental emotions has significant implications for understanding the effects of integral emotions. Yet there is a difference between the two, as Kühne (2012a) further pointed out that incidental emotions can be more easily discounted than integral ones when the emotion-induced goal is obtained (the goal attainment hypothesis, e.g., Lerner & Keltner, 2000) or the irrelevance of the emotional state has been recognized by the individual (the cognitive awareness hypothesis, e.g., Desteno, Petty, Wegener, Rucker, & Kruglanski, 2000; Lerner, Goldberg, & Tetlock, 1998). Emotion-induced goals associated with integral emotions are more likely to be still active during judgment formation than goals produced by incidental emotions. For example, Dunn and Schweitzer (2005) found that although trusters‘ incidental emotions (i.e., anger, gratitude, sadness, happiness, guilt, and pride) influenced perceptions of trustworthiness, there were no carryover effects of these incidental emotions on trust judgments when participants were made aware of the source of their incidental emotions or when they were very familiar with the trustee. Similarly, Small and Lerner (2008) found that when participants‘ cognitive resources were limited, 94 there were no significant differences between incidentally angry and sad participants in their depth-of-processing, which in turn eliminated the differential carryover effects of anger versus sadness on welfare policy preferences. As summarized by Kühne (2012a), when individuals have limited cognitive resources (i.e., low motivation and/or ability), emotions could have an impact ―regardless of its relevance,‖ whereas ―integral emotions should be more important with increases in processing resources‖ (p. 14). To summarize, research on integral and incidental affect converges to show that emotions, both related and unrelated, have significant impacts on cognition and thus need be taken into account in media effects research. The present study focuses on two integral emotions—anger and fear induced by exposing participants to emotional stimuli about medical errors. Since related and unrelated affect do not differ with respect to their experiential qualities, this study also draws primarily on theory and research on incidental affect (i.e. appraisal tendency framework; Lerner & Keltner, 2000, 2001) to explain why there are emotion-specific effects of message-induced emotions on opinion formation. Whenever possible, studies on integral emotions from communication science and political science are included. The Appraisal-Tendency Framework Lerner and colleagues (Han et al., 2007; Lerner & Keltner, 2000, 2001; Lerner & Tiedens, 2006) proposed the appraisal-tendency-framework (ATF) to explain influences of incidental emotions on attitude and judgment formation. The central premise of ATF 95 predicts that each emotion activates a cognitive predisposition to appraise future events in line with the central appraisal dimension that triggered the emotion. This appraisal-tendency influences an individual‘s ―interpretation of a situation and subsequent evaluations/judgments about the situation‖ (Kim & Cameron, 2011, p. 833). Many prior theories of affective influences on judgment have taken a valenced-based approach that contrasted only the effects of global positive versus negative affect. ATF moves beyond that by linking ―emotion-specific appraisal processes to a broad array of judgment and choice outcomes‖ (Lerner & Keltner, 2001, p. 146). While the majority of appraisal theories of emotion has been concentrated primarily on the link between appraisals and the elicitation of emotions (cognition-to-emotion), ATF is the first theoretical framework that links elicited emotions and subsequent appraisals (Han et al., 2007; Keltner et al., 1993). According to ATF, emotions not only can arise from but also give rise to core appraisal themes. That is, emotions co-occur with certain cognitive appraisals and would trigger continuing perceptions of such appraisals even in novel situation (Lerner & Tiedens, 2006). ATF is based on cognitive-appraisal theories and functional theories of emotion. On one hand, from appraisal theories of emotion, Lerner and colleagues borrowed the idea that a diverse array of cognitive dimensions (i.e., certainty, pleasantness, attentional activity, control, anticipated effort, and responsibility) could effectively differentiate discrete emotions (Lerner & Kelnter, 2000, p. 476). These core appraisals determine the influence of specific emotions on social judgments—a process 96 called appraisal tendency. On the other hand, from functional theories of emotion, they borrowed the idea that emotions serve adaptive functions by triggering a set of responses to enable the individual to deal quickly with opportunities or crises in the environment. They emphasized that emotions, ―although tailored to help the individual respond to the event that evoked the emotion, often persist beyond the elicitation situation‖ and would carry over to influence one‘s responses to objects or events that are irrelevant to the initial cause of the emotion (Lerner & Keltner, 2001, p. 146). In general, the ATF concentrates on the effects of incidental emotions and has been applied widely in studying consumer decisions and behaviors (Han et al., 2007). The extant literature showed that anger would reduce risk perception while fear would increase it (e.g., Habib, Cassotti, Moutier, Houdé, & Borst, 2015; Lerner et al., 2003; Lu, Xie, & Zhang, 2013). Anger and fear influence risk perception in opposite directions due to their differences in two dimensions of cognitive appraisals. Specifically, anger has been found to be associated with high certainty and control while fear has been found to trigger low certainty and control (Lerner & Keltner, 2000, 2001). Certainty and control are two important factors that impact judgment and decision making involving risk (Lerner & Tiedens, 2006). Fear and anger should therefore induce differential risk perception and preference. Along these lines, anger would be expected to reduce risk perception regarding medical errors while fear would be expected to increase such perception. Furthermore, the extant literature showed that anger tended to trigger more negative evaluation and attitude towards human agents than other negative emotions such 97 as fear or sadness (Goldberg et al., 1999; Keltner et al., 1993; Kim & Cameron, 2011; Mackie et al., 2000; Quigley & Tedeschi, 1996). Anger increases individuals‘ tendency to blame and punish causal agents for harmful behaviors (Lerner et al., 1998; Quigley & Tedeschi, 1996; Wagner, 2014; Weber, 2004). Punitive tendency and attribution of individual blame would then lead to more negative attitudes towards causal agents (Choi & Lin, 2009; Kim & Cameron, 2011). This study therefore expects that relative to fear, anger would induce more negative attitudes towards health providers responsible for medical errors. In addition, this study expects that anger would induce more support for punitive policy initiatives that punish individual health providers, whereas fear would be expected to induce more preference for remedial policy initiatives that deal with the existing healthcare system. While anger tends to trigger appraisals of individual control and blame, fear is associated with appraisals of situational control and perceptions of diffuse causal agents. Effects of anger versus fear on risk perception. In their initial empirical tests of ATF, Lerner and Keltner (2000, 2001) compared the effects of anger and fear on risk perception. According to ATF, a clear empirical strategy researchers should apply when investigating the distinctive effects of emotions is that they should ―compare emotions that are highly differentiated in their appraisal themes on judgments/choices that relate to that appraisal theme‖ (Lerner & Tiedens, 2006, p. 119). Anger and fear, as outlined in the literature, differ markedly in appraisals of certainty and control: fear both arises from and gives birth to appraisals of profound uncertainty and situational control; whereas anger 98 co-occurs with appraisals of high certainty and individual control. Certainty and control, in turn, ―map directly onto the two cognitive metafactors in the risk literature that reliably determine risk assessment‖ (Lerner & Keltner, 2000, p. 480). Therefore, Lerner and Keltner predicted that anger and fear should produce opposite effects on risk perception such that anger would decrease risk perception while fear could increase it. In their studies, they asked participants to estimate the number of annual fatalities due to 12 events that lead to a certain number of deaths each year in the United States (e.g., brain cancer, strokes, floods; Lerner & Keltner, 2000) or to estimate the likelihood of the occurrence of specific positive and negative events in their own life as compared to the lives of relevant peers (Han et al., 2007; Lerner & Keltner, 2001). Consistent with their expectations, the empirical findings showed that angry people made more optimistic judgments and risk-seeking choices, because anger generated confidence, energy, and perceptions of high individual control. By comparison, fearful people made more pessimistic judgments and risk-averse choices, because fear triggered great uncertainties and perceptions of low individual control. Moreover, in their national field experiment of 973 Americans, Lerner et al. (2003) found that experimentally primed anger and fear, created by asking participants to write and think about emotional aspects of the September 11 terrorist attack, produced opposing effects on risk perceptions and policy preferences. They found that primed anger led to more optimistic beliefs and activated more preference for punitive public policies, whereas primed fear triggered more pessimistic beliefs across a range of risk 99 perceptions, both terror and nonterror related (i.e., ―Another major terrorist attack will occur within the next 12 months, (―Being hurt,‖ and ―Getting the flu‖) and increased preferences for conciliatory polices. The carryover effects of appraisal tendencies not only shaped perceptions of an abstract future but also biased perceptions of past experiences. Fischhoff et al. (2005) found that when participants were induced to experience fear about the event (one year after the terrorist attack), they recollected having experienced high levels of risk during that time. By contrast, when induced to feel anger, participants tended to recollect low levels of risk. Similarly, Hemenover and Zhang (2004) showed that incidental anger activated a form of defensive optimism through which they deemphasized the importance and potential impact of two stressful life events. Along these lines, Druckman and McDermott (2008) found that anger and enthusiasm were positively related to risk-seeking behavior whereas distress was negatively correlated with risk-seeking behavior. To summarize, fear and anger produce opposite effects on judgments and decision-making involving risk estimates. The ―the proximal mechanism‖ driving these carryover effects revolved around cognitive appraisals of certainty and control (Lerner & Tiedens, 2006, p. 124). This line of research has primarily focused on the carryover effects of incidental anger and fear. Yet this work has important implications for examining the effects of message-induced anger or fear on risk perception. A few researchers went a step further by examining whether discrete emotional reactions mediated the news framing effects on opinion and attitude. For example, Kühne and 100 Schemer (2015) found that anger mediated the effects of the anger frame on both the preference for punitive policies and the behavioral intention to punish those who caused accidents. Along these lines, Goodall et al. (2013) found that there was an indirect path from exposure to news coverage of alcohol-related incidents (i.e., anger-inducing) to individual-level causal attributions and individually oriented alcohol-control policy support via the elicited anger. They also found an indirect path from exposure to the nonalcohol version (i.e., fear-inducing) to societal-level causal attributions and societally oriented alcohol-control policy support via the elicited fear. Gross (2008) exposed participants to read either a thematically framed or one of two episodically framed column arguing against mandatory minimum sentencing and found that sympathy and pity, elicited by the episodic frame, mediated the relationship of exposure to news frames and opposition to mandatory minimum sentencing. Not all elicited discrete emotions mediated the effects of exposure to news frame. For example, Lecheler et al. (2013) found that while exposure to the positive valence version of the economic consequences frame induced both enthusiasm and contentment, only enthusiasm mediated the relationship between exposure to the news frame and policy opinion. Similarly, exposure to the negative valence version of the same generic frame induced both anger and fear, while anger was a significant mediator, fear was not. Similarly, Major (2011) examined whether emotional reactions to episodically/thematically framed loss/gain stories of obesity and lung cancer mediated the effect of frames on attributions of responsibility. It was found that guilt was a significant mediator of the relationship between the gain 101 frames and societal responsibility attribution. As argued by Lecheler et al. (2013), ―a next step in integrating the role of emotions into framing effects theory is to examine whether emotional responses also function as mediators when it comes to the effects of news frames on attitudes and opinions‖ (p. 190). Regarding the issue of medical error, it predicts the following: Hypothesis 2a: Participants exposed to a fear-inducing news story will have higher levels of risk perception about medical errors than those exposed to an anger-inducing news story. Hypothesis 2b: The effect hypothesized in H2a will be mediated via elicited anger and fear. Hypothesis 2c: The effect hypothesized in H2b will be moderated by framing manipulation. The effects of anger versus fear on evaluation, blame, and attitude. Fear is associated with strong attributions of situational control; it arises from attributions to external factors (Lerner & Keltner, 2000; Small & Lerner, 2008; Smith & Ellsworth, 1985). Those that experience fear perceive negative events as unpredictable and feel great uncertainty (Nabi, 2003). Fear triggers ―an implicit tendency to focus on external/uncontrollable causes‖ for negative events (Small & Lerner, 2008, p. 153) and thus should induce less negative evaluations towards causal agents than anger. By contrast, anger is characterized with perceived high certainty of what has happened, a confidence of individual control over the situation and a strong individual causal 102 attribution (e.g., Keltner et al., 1993; Nabi, 1999, 2003). Thus, anger would trigger more negative evaluations of human agents than fear. Lerner and Tiedens (2006) pointed out that ―anger merits special attention‖ in the study of judgment and decision-making because its effects distinctively diverge from the effects of other negative emotions (e.g., sadness and fear) on a number of outcomes. Indeed, recent research has shown pervasive carryover effects of anger not only on choices and decision-making involving risk estimation as outlined above, but also on attributions of causality, evaluation, and attitudes. In general, this group of studies found that incidental anger could increase attributions of responsibility to human agents and induce more negative attitudes and evaluations than fear or sadness. For example, Keltner et al. (1993) found that participants induced to feel sad perceived situationally caused events as more likely and situational forces more responsible for an ambiguous event. On the contrary, angry participants perceived events caused by dispositional forces as more likely and other people more responsible for an ambiguous social event. The results were consistent with the central idea of the ATF that sadness triggered appraisals of situational control even in novel situations whereas anger activated appraisals of individual control. Studies by Lerner and colleagues further identified the carryover tendencies of incidental anger to trigger individual causal attributions. They found that experimentally primed anger in a normatively unrelated situation would lead participants to make more punitive attributions to defendants and prescribed more punishment in ensuing fictional tort cases, 103 even though the defendants were not responsible for the original anger-provoking situation (Goldberg et al., 1999; Lerner et al., 1998). Moreover, feelings of anger and appraisal of blame reinforce each other in a recursive loop such that ―the more anger, the more blame placed on others and vice versa‖ (Lerner & Tiedens, 2006, p. 123). Anger results from and gives rise to attribution of blame (Meimer & Robinson, 2004; Goldberg et al., 1999). Furthermore, anger triggers more negative attitudes and evaluations. For example, DeSteno, Dasgupta, Bartlett, and Cajdric (2004) found that incidental feelings of anger created automatic prejudice against outgroups, as anger has a basic association with intergroup competition and conflict which in turn evoked a psychological readiness to evaluate outgroups negatively. Mackie et al. (2000) found that the more group members perceived the in-group as being in a strong position relative to the out-group, the more they expressed anger toward the out-group and aggressive behavioral tendencies to oppose or confront the out-group. Furthermore, anger has been found to induce more negative attitudes towards the poor, because angry people are more likely to blame poor individuals for their poverty (Cozzarelli, Tagler, & Wilkinson, 2001). For example, Small and Lerner‘s (2008) study revealed that incidental anger decreased participants‘ willingness to provide welfare assistance while incidental sadness led participants to recommend higher levels of welfare assistance, because sadness is associated with appraisals of situational control. This appraisal tendency led sad people to perceive that external circumstances and situational forces caused a person‘s misfortune. On the contrary, anger triggered 104 appraisals of individual control and human agency, which in turn led angry people to perceive that an individual caused his or her misfortune and was primarily responsible for his or her plight. Accordingly, their study showed that angry people chose lower levels of welfare assistance than did sad people. Along these lines, anger has been found to undermine trust. Dunn and Schweitzer (2005) found that incidental happiness and gratitude increased trust while incidental anger decreased trust. Specifically, individuals who were induced to experience happiness and gratitude in unrelated settings tended to misattribute the positive valence of these two emotions to their judgments of other people. As a result, these individuals expressed more trust towards other people than those that were induced to feel angry. By contrast, individuals who were induced to feel anger in unrelated situations tended to misattribute the negative valence of anger to their judgments of other people who were not involved in evoking their anger, thus expressing a lower level of trust towards other people. Together, this set of literature (for a review, see Lerner & Tiedens, 2006) demonstrates that incidental anger gives rise to a stronger blaming tendency and more negative evaluations and attitudes than other negative emotions such as fear or sadness. This line of research has important implications for the study of integral emotions. It can be posited that message-induced anger or fear, once activated, will produce the same patterns of appraisals as incidental anger or fear and hence similar effects with regards to blaming tendency, attitude, and evaluation. Some initial studies have shed light on the effects of discrete negative emotions on attitudes and evaluations. For example, Nabi 105 (2003) investigated whether experimentally primed anger and fear would have differential impacts on information accessibility, information seeking, and policy preferences when participants evaluate a social issue. She proposed the emotion-as-frame hypothesis which predicted that emotional experiences would guide individuals‘ perceptions and processing of media messages. The results indicated that for the more familiar issue of drunk driving, anger-primed participants were more likely to have an individual agent, or blame-related cause and a punitive solution on the top of their mind compared to the fear-primed group, which had more societal-level causes and protection-related solutions accessible. Anger was also found to increase participants‘ desire for retribution-related information and support for retributive policies; by contrast, fear increased the desire for protection-related information and participants‘ preference for protective solutions. These findings supported her notion of emotion-as frame. For the less familiar topic of gun violence, no significant differences between anger and fear were found, because individuals tended to have a less-developed schema with this issue. Nabi explained that if emotions serve a framing function, then ―the greater the schema development, the stronger the framing effects‖ (p. 240). Therefore, emotions should have more pronounced impacts on processing, information seeking, and policy preferences when individuals are exposed to highly familiar issues such as drunk driving, as evidenced in her study. Similarly, Kühne and Schemer (2015) found that anger induced by having participants read a news story about a traffic accident not only increased the accessibility of information about punishment, but also promoted the individual‘s 106 preference for punitive policy initiatives and behavioral intention to punish the causal agents. By contrast, message-induced sadness increased accessibility of information regarding helping the victim. Sad participants were more likely to support remedial policy initiatives and had a stronger behavioral intention to help the victims. Along these lines, Kim and Cameron (2011) investigated the public‘s emotional responses to news framing about an organizational crisis and the implications of their emotions for subsequent attitudes and perceptions regarding the organization. By focusing on the intentional wrongdoings of the company, they elicited anger in participants (i.e., integral anger); whereas an emphasis on human‘s sufferings made participants to feel sad (i.e., integral sadness). The results indicated that angry participants evaluated the related company significantly more negatively than those that felt sad. Therefore, they pointed out that ―in a negative event, anger-induced people perceive the related party more negatively, than do sadness-induced people‖ (p. 834). Their findings were in line with Choi and Lin‘s (2009) findings, which indicated that consumers‘ anger as a result of exposure to the Mattel product recalls posted on online bulletins (i.e., integral anger) negatively predicted the perceived reputation of Mattel and positively predicted behavioral intentions to boycott Mattel‘s products. Moreover, in the context of crisis communication, a number of other researchers have consistently found that anger directed towards the related parties was a robust predictor of stakeholders‘ negative word-of-mouth intentions (i.e., talking negatively about the organization), more negative attitudes towards the responsible organization and a stronger behavioral 107 intention to boycott the organization or sever interactions with it (Coombs, 2007; Coombs & Holladay, 2009; McDonald, Sparks, & Glendon, 2010; Utz, Schultz, & Glocka, 2013; Yang, Kang, & Johnson, 2010). As indicated above, message-induced anger or fear could produce divergent evaluations of human agents and differential policy preferences. Concerning the issue of medical errors, it is therefore expected that angry individuals would be more likely to attribute medical errors to internal/dispositional causes of individual providers than those exposed to fear-inducing news, as anger activates strong blame tendencies and perceptions of human causality. In contrast, participants exposed to fear-inducing news would be more likely to see the causes of medical errors as stemming from forces over which individuals have little control, like system failures, as fear co-occurs with appraisals of situational control and perceptions of diffuse causal agents. As a result, angry individuals would form more negative attitudes towards health providers involved in errors and prefer punitive measures that focus on punishing individual providers. By contrast, fear would ameliorate respondents‘ attribution of blame to individual providers and thus alleviate their negative attitudes towards those involved in errors. The great uncertainty associated with fear would also lead individuals to consider various contributing factors underlying medical errors and thus would increase preference for remedial measures that deal with the existing healthcare system where errors are made and perpetuated. Moreover, since episodic news stories are expected to elicit more anger or fear than their thematic counterparts, it is also posited that episodic news stories would 108 produce stronger effects than their thematic counterparts concerning attitudes and policy preferences. The following hypotheses are therefore constructed: Hypothesis 3a: Participants exposed to an anger-inducing news story will have more negative attitudes towards health providers involved in medical errors than those exposed to a fear-inducing news story. Hypothesis 3b: The effect hypothesized in H3a will be mediated via elicited anger and fear. Hypothesis 3c: The effect hypothesized in H3b will be moderated by framing manipulation. Hypothesis 4a: A fear-inducing news story, as compared with an anger-inducing news story, increases support for remedial policy measures. Hypothesis 4b: The effect of a fear-inducing news story on support for remedial policy measures will be moderated by framing manipulation. Hypothesis 4c: The effect hypothesized in H4a will be mediated by elicited fear and risk perception regarding medical errors. Hypothesis 5a: An anger-inducing news story, as compared with a fear-inducing news story, increases support for punitive policy measures. Hypothesis 5b: The effect of an anger-inducing news story on support for punitive policy measures will be moderated by framing manipulation. Hypothesis 5c: The effect hypothesized in H5a will be mediated by elicited anger and attitude towards health providers involved in errors. CHAPTER 6 METHOD Design This study used an experimental 2x2 between-participant design. The study was approved by the University of Utah‘s Institutional Review Board. One experimental factor was the type of emotion-inducing condition (anger-inducing vs. fear-inducing), and the other was framing condition (episodic vs. thematic). Dependent variables included participants‘ emotional response, risk perceptions regarding medical errors, attitudes towards health providers responsible for errors, preference towards punitive and remedial measures and behavioral intentions to support a punitive or remedial patient advocacy group. Procedures Participants were recruited from the department of Communication at the University of Utah. Students participated in this experimental study for extra course credit. They completed an online survey in which they were randomly assigned to reading one of four 110 emotion-inducing news stories about medical errors. Then they completed a series of questions assessing their emotional reactions to the news story, opinions and behavioral intentions regarding medical errors, and demographic questions. The survey was created using Qualtrics, an online survey software that enables users to create and distribute the survey in various ways (Qualtrics, Provo, UT). Students can click on a Qualtrics-generated link to take the survey anywhere with Internet connection. Participants A total of 245 participants completed the survey. The participant pool consisted of 151 female (61.6%) and 94 male students (38.4%), with an average age of 24.3 (SD = 4.9 years). The majority of the participants were White (74.7%), 8.6% were Asian, 6.9% were Hispanic/Latino, 4.5% were Black/African American, .8% were Pacific Islanders, 0.4% were American Indian/Native Americans, and 0.4% reported ―other.‖ Regarding the year of university, 42.9% were seniors, 38.8% were juniors, 10.2% were sophomores, 4.9% were in 5th year or higher, 2.4% were freshman, and 0.8% reported as graduate students. Of all the respondents, 151 (61.6%) reported that they have had a personal experience with medical errors, which either happened to themselves, or to a family member or friend. Of this group of people, 23 (24.5%) reported that the medical error they or their friend/family had experienced was ―extremely serious,‖ 27 (28.7%) reported that it was ―very serious,‖ 31 (33.3%) reported that it was ―moderately serious,‖ two 111 (12.8%) reported that it was a little bit serious, and only one participant (1.1%) reported that it was ―not at all serious.‖ With respect to perceptions of the seriousness of medical errors, 37.1% regarded it as ―extremely serious,‖ 28.2% regarded it as ―very serious,‖ 23.7% regarded it as ―moderately serious,‖ 10.2% regarded as ―slightly serious,‖ and only two participants (0.8%) regarded as ―not serious at all.‖ Stimuli Participants were randomly assigned to one of the four constructed emotion-inducing news stories about medical errors: episodic anger-inducing condition, thematic anger-inducing condition, episodic fear-inducing condition, and thematic fear-inducing condition. The stimulus materials are included in the appendix D, E, F, and G. As introduced in the previous sections, medical errors have been selected as the issue context for the empirical investigation of the hypotheses about the relative capacity of episodic and thematic emotion-inducing conditions to influence individuals‘ opinions. Both thematic and episodic frames were salient because medical errors have been in the news and the public debate has been characterized by statistical studies and investigations as well as by a number of highly emotional stories of individual victims‘ sufferings (Aarøe, 2011). In other words, medical errors is a ―balanced case that makes it possible to construct the experimental stimuli for both the thematic and the episodic frame in a realistic way, enhancing external validity‖ (Aarøe, 2011, p. 212). Manipulation of news 112 framing (episodic vs. thematic) and emotion-inducing condition (anger- vs. fear-inducing) involved using information, words, and depictions that appeared in real news stories. The basic information about medical errors within each news article was kept identical. That is, each article included an identical core section of two paragraphs describing the factual information about medical errors. For episodic framing manipulation, the two episodic frames featured one main character, Carol Smith, as an individual victim of medical errors. The two episodic stories included exactly the same depictions of her personal experiences with medical errors. For thematic framing manipulation, the two thematic frames featured statistics and quotes from authorities. The two thematic stories highlighted statistics from the same source (i.e., USA TODAY investigation), presented survey results from the same survey (i.e., Pew Research Center) and provided quotes from the same authority (i.e., head of the Citizen Advocacy Center). For emotion manipulation, anger-inducing stories emphasized information related to anger‘s core relational theme (i.e., injustice), and fear-inducing stories emphasized information related to fear‘s core relational theme (i.e., threat). The two thematic conditions emphasized the same core relational themes as their episodic counterparts. The difference is that the two episodic conditions focused on a specific story and the main character‘s personal comments regarding her experiences to elicit emotions while the two thematic conditions used statistics and authorities‘ quotes to induce emotions. Specifically, in the episodic anger condition, the main character expressed her anger over the fact that responsible health providers went unpunished in her case, despite her pain and suffering 113 (i.e., perceived injustice for anger induction). Key words such as ―outrage and infuriated‖ were incorporated in her account of what happened to her. On the other hand, the thematic anger condition presented statistics at the nationwide level about how many health providers went unpunished and how many doctors‘ licenses remained intact after they committed medical errors from 2001 to 2014. The statistics were based on the result of a USA TODAY investigation. It also included a survey result from the Pew Research Center indicating that 85% if the respondents were outraged by the prevalence of medical errors. In addition, the last paragraph of the thematic anger story presented a quote from an authority—the head of the Citizen Advocacy Center—who criticized the deficient oversight systems which allowed doctors to continue practicing after committing medical errors. In the episodic fear condition, the main character expressed her fear of going to hospitals in the future as well as her concern that more people will be victimized by medical errors if the system is not improved to reduce the risk. Key words such as ―frightened, terrifying and afraid‖ were incorporated in her personal account of what happened to her. On the other hand, the thematic fear condition provided statistics of the prevalence of medical errors and emphasized how breakdowns in communication has made healthcare conducive to medical errors. It also presented a survey result from the Pew Research Center which showed that 85% of the respondents were worried about the prevalence of medical errors and were afraid that they would be the victims. In addition, 114 the quote from the same authority pointed out how vulnerable healthcare is to communication pitfalls which are a major source of medical errors. The online website of USA TODAY (USATODAY.com) was selected to be the news source for the constructed news stories. The website provides the public with up-to-date online coverage of medical and health news, including research advances in medical errors and news stories of individual victims. Moreover, research indicates that news websites have been increasingly popular as a major source of health and medical information for individuals. Care was taken to adjust the four constructed articles to the common layout and editorial style of online health news published and publically available on USATODAY.com. Measures Demographic Variables Participants‘ age, gender, race, and educational level (i.e., year in university) were demographic variables included in this study. Furthermore, they were also asked about their perceived importance of religion (i.e., the amount of guidance provided by religion in one‘s life) and ideology. Perceived importance of religion was measured on a 7-point scale ranging from 0 (no guidance at all) to 6 (a great deal of guidance). Zero was recoded into 1 and 6 recoded into 7. For ideology, participants were asked to indicate how conservative they think they are in terms of economic and social issues on a 7-point 115 semantic differential scale ranging from 1 (very liberal) to 7 (very Conservative). A mean index for ideology was computed (M = 3.43, SD = 1.51). Dependent Variables Risk perceptions regarding medical errors. Participants were measured on risk perceptions regarding medical errors. They were asked the following question: ―To what extent do you agree with each of the following statements regarding your own risks for experiencing a medical error?‖ Scale items were adapted from the perceived risk index of Major (2011). Specifically, the scale included four items: (a) ―I am likely to be at risk for medical errors‖; (b) ―It is possible I will experience medical errors‖; (c) ―I am able to protect myself from medical errors‖; and (d) ―Preventing medical errors is easy for me.‖ Participants were asked to rate each item using a 5-point scale that ranged from 1(strongly disagree) to 5(strongly agree). The negatively keyed items (i.e., items c and d) were reverse coded and the total of four measures were submitted to an exploratory factor analysis. Analyses indicated a two-factor solution with the two reversely coded items loaded on another factor. The scale consisting of all four measures had an α coefficient of .64. After deleting the two reverse-coded items, the scale had an α coefficient of .74. So the scale included two items: (a) ―I am likely to be at risk for medical errors‖; (b) ―It is possible I will experience medical errors.‖ Accordingly, the mean index for risk estimate was computed (α reliability = .74, M = 3.39, SD = .79). 116 Attitudes towards health providers responsible for errors. Participants‘ attitudes towards health providers responsible for causing medical errors in general were measured by five items on a 5-point scale that ranged from 1(strongly disagree) to 5(strongly agree): dependent, honest, reliable, trustworthy, and experienced (Ohanian, 1990; Newell & Goldsmith, 2001). Factor analysis of these five items revealed a single factor structure, explaining 59.36% of the variance. So the index of attitude was computed (M = 2.81, SD = .76). It was quite reliable; Cronbach‘s α = .86. The lower the score, the more negative attitude towards health providers responsible for errors. Policy preferences. The preference for punitive and the preference for remedial measures were assessed with five items each which were specifically designed for the present study (e.g., ―suspending the licenses of health professionals who make medical errors,‖ ―Having a government agency fine health professionals who make medical errors, ―requiring hospitals to develop systems for preventing medical errors,‖ and ―increasing the number of nurses in hospitals‖). Participants were asked to state how much they agreed or disagreed with each of the 10 measures using a 5-point scale that ranged from 1(strongly agree) to 5(strongly disagree). These 10 measures were submitted to an exploratory factor analysis which indicated a two-factor solution. Three items related to punishment loaded on one factor: (a) ―Increasing malpractice lawsuits against professionals who make medical errors,‖ (b) ―Suspending the licenses of health professionals who make medical errors,‖ and (c) ―Having a government agency fine health professionals who make medical errors.‖ A mean index for the preference for 117 punitive measures was computed (M = 3.63, SD = .87, Cronbach‘s α = .77). Four items related to precautions/protections loaded on a second factor: (a) ―Improving the training of health professionals,‖ (b) ―Requiring hospitals to develop systems for preventing medical errors,‖ (c) ―Responsible health professionals should be required to undergo mandatory training and reeducation,‖ and (d) ―Requiring a showing of professional competence before reinstating a license.‖ Also, a mean index for the preference for remedial measures was computed (M = 4.27, SD = .64, Cronbach‘s α = .81). Emotional Responses as Mediators Mediators and dependent variables were measured after participants were exposed to news stories. Participants‘ emotional responses to news stories were measured to assess whether the target emotions (anger vs. fear) have been successfully induced. Based on prior work conducted by Dillard and colleagues (Dillard et al., 1996; Dillard, Kinney, & Cruz, 1996; Dillard & Peck, 2000, 2001; Dillard & Shen, 2006), this study developed a 10-item measure to gauge participants‘ emotional responses. Specifically, immediately after exposure to the news stories, participants were asked ―While reading the new story, to what extent did you experience the following emotions?‖ They indicated the extent to which they experienced the following emotions on a 5-point scale ranging from 1 (not at all) to 5 (a great deal): Irritated, angry, outraged, fearful, scared, terrified, annoyed, worried, surprised, and disgusted. Factor analysis was conducted to reduce the data and explore the possible underlying factor structure of the 118 observed variables (Burton & Mazerolle, 2011). As a result, two factors emerged. The unidimensional emotion scales and their corresponding items were the following: Anger (irritated, angry, outraged, α .883), and fear (fearful, scared, terrified, α .90). Accordingly, the anger items were summarized to form a measure of anger (M = 3.07, SD = 1.08), and the fear items to form a measure of fear (M = 2.91, SD = 1.06). Control Variables Participants‘ gender, ideology, and race (recoded as binary, White vs. others) were included as possible control variables. A zero-order correlation was conducted with anger, fear, age, ideology, gender, race, religion, and attention to news stories of medical errors entered as the variables. The correlation matrix showed that ideology was negatively related to anger (r = -.21, p <. 01) and fear (r = -.20, p < .01) such that the more conservative participants were with respect to economic and social issues, the less likely that individuals would experience emotional reactions to news coverage of medical errors. Gender was negatively related to anger (r = -.21, p < .01) and fear (r = -.30, p < .01) such that female participants were more likely than male participants to experience emotional reactions to news stories of medical errors. Ethnicity was negatively related to fear (r = -.18, p < .01) such that relative to others, white individuals were less likely to experience emotions after they read news stories of medical errors. Therefore, gender, race, and ideology were control variables included in the analyses. CHAPTER 7 ANALYSES AND RESULTS Power Analysis Power analysis refers to the ability of the experimental design to detect a difference when the difference in reality exists and reject the null hypotheses. The GPower was utilized to calculate the power of the current study design to detect significant effects in analyses of variance (α = .05, 4 groups, N = 245). Three power analyses were conducted for three standard effect sizes, small (f = .10), medium (f = .25), and large (f = .40, Cohen, 1988). The experimental design had excellent power to detect a large (.99) or medium effect (.97), but very little power to detect a small effect (.34). Since the sample of the current study was not large enough to have adequate power to detect small effects, the following statistical analyses and hypothesis testing will make note of results that approached but failed to meet the standard statistical significance of .05. Thus, results of p values less than .10 are also reported and noted both in text and in tables and figures. 120 Manipulation Check To check whether there was successful anger induction, a one-way ANOVA was conducted with emotion condition (anger-inducing vs. fear-inducing) as the independent variable and anger as the dependent variable. Results indicated that there was a significant main effect for emotion on elicited anger, F (1, 243) = 4.03, p < .05. Those that read anger-inducing news stories experienced significantly more anger (M = 3.21, SD = 1.08) than those that read fear-inducing news stories (M = 2.93, SD = 1.06). To check whether there was a successful fear induction, a second one-way ANOVA was conducted with emotion condition as the independent variable and fear as the dependent variable. Results showed that there was no significant main effect for emotion on elicited fear, F (1, 243) = 1.53, p = .22. The anger-inducing news group (M = 2.83, SE = 1.11) and the fear-inducing news group (M = 3.00, SE = 1.01) did not differ significantly with respect to fear reactions. Thus, there was successful anger induction but no successful fear induction. Experimental Effects on Anger and Fear Responses Hypothesis 1 predicted that an episodic frame would elicit stronger emotional responses than a thematic frame and this effect would be moderated by emotional manipulation. To test these hypotheses, a 2 (episodic/thematic framing) x 2 (anger-inducing/fear-inducing) between-participants multivariate analysis of covariance was performed on anger and fear, controlling for the three covariates. Using Wilks’ 121 Lambda, the analysis showed a significant main effect for emotional manipulation, Wilks F (2, 237) = 8.182, p <.001,  p2 = .065, observed power = .958, but not for framing manipulation, Wilks F (2, 237) = 1.987, p = .139. This analysis revealed a marginally significant interaction between framing and emotional manipulation, Wilks F (2, 237) = 2.556, p = .080,  p2 = .021, observed power = .508. Univariate tests indicated that emotional manipulation was significantly related to anger response, F (1, 238) = 9.014, p = .003, but not significantly related to fear response, F (1, 238) = .066, p = .798. Framing manipulation was marginally significantly related to anger response, F (1, 238) = 3.301, p = .071, and marginally significantly related to fear response, F (1, 238) = 3.176, p = .076. There was significant interaction between framing and emotional manipulation on anger response, F (1, 238) = 4.307, p = .039 as well as on fear response, F (1, 238) = 4.021, p = .046. Effects of emotional manipulation and framing after adjustment for covariates were further investigated through Roy-Bargman Stepdown analyses. Stepdown analyses follow a univariate analysis of the significance of the most optimal combination of the dependent variables (DVs) by providing estimates of the significance of the first DV and then the significance of the second DV with the first controlled for. This analysis allows a ―statistically pure look at the significance of DVs, in context, with Type I error rate controlled‖ (Tabachnick & Fidell, 2007, p. 286). Specifically, two step-down analyses were conducted. The first analysis gave priority to fear response to estimate anger response after fear response was controlled and the 122 second gave priority to anger response to estimate fear response after anger response was controlled. Stepdown analysis for emotional manipulation on anger response showed that after adjusting for difference on fear response, there was significant mean difference between anger-inducing and fear-inducing conditions on anger response, F (1, 237) = 16.293, p < .001. Anger-inducing condition elicited significantly more anger response than fear-inducing condition after controlling for fear and the other three covariates (M = 3.276 vs. M = 2.858, Mdiff = .418, SE = .104). Stepdown analysis for framing manipulation on anger response showed that after adjusting for difference on fear response, there was no significant mean difference between episodic and thematic framing conditions on anger response, F (1, 237) = .800, p =.372. Furthermore, stepdown analysis for the interaction between framing and emotional manipulation on fear response showed no significant interaction effect on anger response after fear response was controlled, F (1, 237) = 1.090, p = .298. Stepdown analysis for emotional manipulation on fear response showed that after adjusting for difference on anger response, there was significant mean difference between anger-inducing and fear-inducing conditions on fear response, F (1, 237) = 7.118, p = .008. Fear-inducing condition elicited significantly more fear responses than anger-inducing condition after controlling for anger and the other three covariates (M = 3.052 vs. M = 2.778, Mdiff = .274, SE = .103). Stepdown analysis for framing manipulation on fear response showed that after adjusting for difference on anger response, there was no significant mean difference between episodic and thematic 123 framing conditions on fear response, F (1, 237) = .677, p =.412. Furthermore, stepdown analysis for the interaction between framing and emotional manipulation on anger response showed no significant interaction effect on fear response after anger response was controlled, F (1, 237) = .809, p = .369. To summarize, the above stepdown F tests showed that there was a significant main effect of emotional manipulation on anger response after adjustment was made for fear response. There was a significant main effect of emotional manipulation on fear response after adjustment was made for anger response. There was no significant main effect of framing manipulation on anger response after fear was controlled. There was also no significant main effect of framing manipulation on fear response after anger was controlled. Thus, Hypothesis 1a was not supported. Moreover, there was no significant interaction between framing and emotional manipulation on anger response after fear was controlled. There was also no significant interaction between framing and manipulation on fear response after anger was controlled. Thus, Hypothesis 1b was not supported. Stepdown analysis for the interaction and two main effects is summarized in Table 1. Predicting Risk Perception H2a asked whether fear-inducing stories will elicit more risk perceptions about medical errors than anger-inducing stories. To test H2a, a two-way ANCOVA was conducted with framing and emotional manipulation as the two independent variables, risk perception as the dependent variable, and gender, ideology, and race as covariates. 124 There was no significant main effect of emotional manipulation on risk perception such that anger-inducing and fear-inducing conditions elicited almost comparable level of risk perceptions regarding medical errors, F (1, 236) = 1.187, p = .277. Thus, H2a was not supported. There was no significant main effect of framing, F (1, 236) = .009, p = .926. There was a significant interaction of emotion and framing manipulation on risk perceptions, F (1, 236) = 5.904, p = .016 (see Figure 1). In other words, the effect of emotional manipulation on risk perceptions was different across episodic and thematic conditions. Analyses of simple effects indicated that for thematic condition, the difference between fear-inducing stories and anger-inducing stories on risk perceptions was in the right direction but not statistically significant (M = 3.464 vs. M = 3.331, Mdiff = .133, SE = .143. p = .351). The pattern was reversed for episodic conditions such that fear-inducing stories elicited significantly less risk perceptions than anger-inducing stories (M = 3.210 vs. M = 3.566, Mdiff = -.357, SE = .145, p = .015). H2b and H2c asked whether there was moderated mediation. To test these hypotheses, PROCESS Model 8 was run with risk perception as the outcome variable, the emotional manipulation as the predictor, framing as the moderator, and elicited anger and fear as two mediators, controlling for ideology, gender, and race. Emotional manipulation was recoded to be -.5 for the anger-inducing condition and .5 for the fear-inducing condition, and framing manipulation was recoded to be -.5 for episodic framing and .5 for thematic framing. Emotional and framing manipulation in all subsequent mediational analyses were coded in this way. Results indicated that there was a significant interaction 125 effect between framing and emotional manipulation on anger, b = -.538, SE = .262, t = -2.055, p = .041. There was a significant interaction effect between framing and emotional manipulation on fear, b = -.528, SE = .254, t = -2.075, p = .039. Emotional manipulation was negatively related to anger (b = -.397, SE = .133, t = -2.979, p = .003) but not significantly related to fear (b = .045, SE = .129, t = .350, p = .726). There was no significant direct effect of emotional manipulation on risk perception, b = -.036, SE = .103, t = -.345, p = .731. The mediator of elicited anger was positively related to risk perceptions, b = .193, SE = .062, t = 3.095, p = .002. The mediator of elicited fear was not related to risk perceptions, b = .007, SE = .064, t = .109, p = .914. The index of moderated mediation was significant for elicited anger (B = -.104, SE = .061, 95% confidence interval: -.2602, -.0125) but not for elicited fear (B = -.004, SE = .039, 95% confidence interval: -.0925, .0682). Specifically, results of moderated mediation analysis indicated that there was a significant negative relationship between emotional manipulation and risk perception through elicited anger for thematic frames (b = -.128, SE = .054, 95% bootstrap confidence interval: -.2602, -.0439) but not for episodic frames (b = -.025, SE = .039, 95% bootstrap confidence interval: -.1172, .0418). This indicates that the indirect effect of anger manipulation on risk perception which occurred via elicited anger was significantly stronger for the thematic condition than for the episodic condition. That is, for those reading thematic stories, anger manipulation elicited anger responses which in turn increased risk perceptions regarding medical errors. 126 Fear did not play a role in mediating the indirect effect of emotional manipulation on risk perception regarding medical errors. In addition, for the direct effects, there was a significant interaction effect between framing and emotional manipulation on risk perception (b = .597, SE = .198, t = 3.018, p = .003). That is, there was a direct effect of emotional manipulation on risk perception under the episodic frame (b = -.334, SE = .143, 95% bootstrap confidence interval: -.6163, -.0522) but not under the thematic frame (b = .263, SE = .143, 95% bootstrap confidence interval: -.0178, .5441). This indicates that framing manipulation moderates the direct effect of emotional manipulation on risk perception. To summarize, moderated mediation effect was significant such that elicited anger mediated anger manipulation‘s effect on risk perceptions only for thematic frames. Fear response did not play a mediating role. In addition, there was residual direct effect of anger manipulation, especially under the episodic frame, that was not explained by experienced anger (see Figure 2). Thus, H2b and H2c were partially supported. Results are summarized in Table 2. Predicting Attitudes Towards Health Providers Involved in Errors H3a asked whether anger-inducing stories would elicit more negative attitudes towards health providers involved in errors than fear-inducing stories. To test H3a, a two-way ANCOVA was conducted with framing and emotional manipulation as the two independent variables, attitude as the dependent variable, and gender, ideology and race 127 as covariates. Results indicated that there was a significant main effect of emotional manipulation on attitude, F (1, 238) = 7.200, p = .008, such that anger-inducing stories induced significantly more negative attitudes towards those providers involved in errors than fear-inducing stories, regardless of message framing (M = 2.677 vs. M = 2.940, Mdiff = -.263). Thus, H3a was supported. There was no significant main effect of framing, F (1, 238) = .290, p = .590. There was no significant interaction between framing and emotion, F (1, 238) = .097, p = .756. H3b and H3c assessed whether there was moderated mediation. To test these two hypotheses, PROCESS Model 8 was run with attitude as the outcome variable, emotional manipulation as the predictor, framing as the moderator, and elicited anger and fear as two mediators, controlling for ideology, gender and race. Results indicated that there was a significant interaction effect between emotional and framing manipulation on anger, b = -.540, SE = .260, t = -2.075, p = .039 (see Figure 3). There was a significant interaction effect between emotional and framing manipulation on fear, b = -.508, SE = .253, t = -2.005, p = .046. Emotional manipulation was negatively related to anger (b = -.397, SE = .132, t = -3.002, p = .003) but not significantly related to fear (b = .033, SE = .129, t = .256, p = .798). For the direct effects, there was no significant interaction effect between emotional and framing manipulation on attitude, b = -.029, SE = .192, t = -.149, p = .882. There was a significant direct effect of emotional manipulation on attitude, b = .226, SE = .100, t = 2.266, p = .024. The mediator of elicited anger was negatively related to attitudes, 128 approaching significance at the .10 level, b = -.100, SE = .061, t = -1.646, p = .101. The mediator of elicited fear was not related to attitudes, b = -.069, SE = .062, t = -1.104, p = .271. The index of moderated mediation was not significant for elicited anger (B = .054, SE = .049, 95% confidence interval: -.0063, .1936) and not significant for elicited fear (B = .035, SE = .041, 95% confidence interval: -.0200, .1495). There was no significant moderated mediation. Thus, H3b and H3c were not supported. Results are summarized in Table 3. Predicting Policy Support Hypotheses 4 and 5 compared the effects of a fear-inducing condition and an anger-inducing condition on support for two types of policy measures and asked whether these effects would be moderated by episodic/thematic framing. To test these hypotheses, a 2 (episodic/thematic framing) x 2 (anger-inducing/fear-inducing) between-participants multivariate analysis of covariance was performed on support for punitive and remedial policy measures, controlling for the three covariates. Using Wilks’ Lambda, the analysis showed a significant main effect for emotional manipulation (Wilks F (2, 237) = 3.764, p = .025,  p2 = .031, observed power = .683), but not for framing manipulation (Wilks F (2, 237) = .078, p = .925). There was no significant interaction between framing and emotional manipulation (Wilks F (2, 237) = .072, p = .930). Univariate tests indicated that emotional manipulation was significantly related to support for punitive measures (F (1, 238) = 6.764, p = .010), but not significantly related to support for remedial measures(F 129 (1, 238) = .168, p = .682). Framing manipulation was not related to support for punitive measures (F (1, 238) = .085, p = .770) and not related to support for remedial measures (F (1, 238) = .139, p = .710). There was no significant interaction between framing and emotional manipulation on support for punitive measures (F (1, 238) = .046, p = .829), and no significant interaction on support for remedial measures (F (1, 238) = .143, p = .705). Effects of emotional manipulation and framing after adjustment for covariates were further investigated through Roy-Bargman Stepdown analyses. Specifically, two step down analyses were conducted. The first analysis gave priority to support for punitive measures to estimate support for remedial measures after support for punitive measures was controlled and the second gave priority to support for remedial measures to estimate support for punitive measures after support for remedial measures was controlled. Stepdown analysis for emotional manipulation on support for remedial measures showed that after adjusting for difference on support for punitive measures, there was no significant mean difference between anger-inducing and fear-inducing conditions on support for remedial measures, F (1, 237) = .771, p = .381. Thus, H4a was not supported. Stepdown analysis for framing manipulation on support for remedial measures showed that after adjusting for difference on support for punitive measures, there was no significant mean difference between episodic and thematic framing conditions on support for remedial measures, F (1, 237) = .071, p =.790. Furthermore, stepdown analysis for the interaction between framing and emotional manipulation on support for punitive 130 measures showed no significant interaction effect on support for remedial measures after support for punitive measures was controlled, F (1, 237) = .098, p = .754. Thus, H4b was not supported. Stepdown analysis for emotional manipulation on support for punitive measures showed that after adjusting for difference on support for remedial measures, there was significant mean difference between anger-inducing and fear-inducing conditions on support for punitive measures, F (1, 237) = 7.356, p = .007. Anger-inducing condition elicited significantly more support for punitive measures than fear-inducing condition (M = 3.776 vs. M = .3.502, Mdiff = .264, SE = .097). Thus, H5a was supported. Stepdown analysis for framing manipulation on support for punitive measures showed that after adjusting for difference on support for remedial measures, there was no significant mean difference between episodic and thematic framing conditions on support for punitive measures, F (1, 237) = .018, p =.892. Furthermore, stepdown analysis for the interaction between framing and emotional manipulation on support for remedial measures showed no significant interaction effect on support for punitive measures after support for remedial measures was controlled , F (1, 237) = .002, p = .963. Thus, H5b was not supported. Stepdown analysis for the interaction and two main effects is summarized in Table 4. H4c asked whether the effect of fear manipulation on support for remedial policy measures would be mediated by elicited fear and risk perceptions regarding medical errors (see Figure 4). To test this hypothesis, PROCESS Model 6 (serial mediation) was 131 run. Support for remedial measures was entered as the outcome variable, emotional manipulation as the predictor variable. Mediator variables were entered in the following order: fear, risk perception, controlling for ideology, gender, and race. Three models of indirect effects were tested. Specifically, Model 1 tested whether elicited fear mediated the relationship between fear manipulation and support for remedial policy measures, Model 2 tested the following casual chain: fear manipulation  fear  risk perception  support for remedial measures, and Model 3 tested whether risk perception mediated the effect of fear manipulation on support for remedial measures. The total effect model was significant, R2 = .060, F (4, 238) = 3.781, p = .005. Emotional manipulation was not significantly related to support for remedial measures, b = -.030, SE = .082, t = -.363, p = .717. The serial mediation model was significant, R2 = .191, F (6, 236) = 9.290, p = .001. Emotional manipulation was not significantly related to support for remedial measures after mediators were taken into account, b = -.006, SE = .076, t = -.081, p = .936. None of these three models were viable. Indirect effect of Model 1 was not significant, b = .004, SE = .014, 95% bootstrap confidence interval: -.0198, .0376. Indirect effect of Model 2 was not significant, b = .001, SE= .004, 95% bootstrap confidence interval: -.0057, .0112. Indirect effect of Model 3 was also not significant, = -.028, SE = .028, 95% bootstrap confidence interval: -.0946, .0166. In other words, fear or risk perception did not play a role in mediating the effect of fear appeal on support for remedial measures and there was no significant serial mediation. Thus, H4c was not supported. 132 However, previous analyses indicated that anger appeal elicited anger responses, which in turn increased risk perceptions for those in the thematic framing condition. Thus, to explore the indirect effect of anger appeal on support for remedial measures via elicited anger responses and risk perceptions, PROCESS Model 6 was run again to test this serial mediation (see Figure 5). Three models of indirect effects were tested. Specifically, Model 1 tested whether elicited anger mediated the relationship between anger manipulation and support for remedial policy measures, Model 2 tested the following casual chain: anger manipulation  anger  risk perception  support for remedial measures, and Model 3 tested whether risk perception mediated the effect of anger manipulation on support for remedial measures. The total effect model was significant, R2 = .060, F (4, 238) = 3.781, p = .005. Emotional manipulation was not significantly related to support for remedial measures, b = -.030, SE = .082, t = -.363, p = .717. The serial mediation model was significant, R2 = .203, F (6, 236) = 10.024, p = .001. Emotional manipulation was not significantly related to support for remedial measures after mediators were taken into account, b = .044, SE = .077, t = .565, p = .573. Two models were viable. Indirect effect of Model 1 was significant (b = -.048, SE = .022, 95% bootstrap confidence interval: -.1020, -.0144). Anger appeal elicited anger responses which in turn increased support for remedial measures. Indirect effect of Model2 was significant (b = -.017, SE = .009, 95% bootstrap confidence interval: -.0423, 133 -.0053). An examination of the coefficients revealed that emotional manipulation was negatively related to elicited anger (b = -.405, SE = .135, p = .003), elicited anger was positively related to risk perception (b = .175, SE = .049, p = .0004), and risk perception was positively related to support for remedial measures (b = .235, SE = .049, p = .001). In other words, anger appeal elicited anger responses, which in turn increased risk perceptions of medical errors, and those with more risk perceptions were more likely to support remedial measures. Indirect effect of Model 3 was not significant (b = -.009, SE= .024, 95% bootstrap confidence interval: -.0610, .0362). Pairwise comparisons between individual indirect effects can be requested for serial mediation in PROCESS (Preacher & Hayes, 2008). The strength of the three specific indirect effects were compared against each other. Bootstrap confidence intervals are provided for inference for these pairwise comparisons when the contrast option is used in conjunction with bootstrapping. Confidence intervals that do not contain zero represent significant contrasts (Preacher & Hayes, 2008). In this case, there were three pairwise contrasts between the three indirect effects. Results indicated that contrast between Model 1 and Model 2 indicates significant difference in the strength of the two indirect effects, b = -.031, SE = .019, 95% bootstrap confidence interval: -.0807, -.0026. That is, the indirect effect via elicited anger responses was significantly greater than the indirect effect via anger responses and risk perception. H5c asked whether the effect of anger appeal on support for punitive policy measures would be mediated by elicited anger and attitudes towards those involved in errors (see 134 Figure 6). To test this hypothesis, PROCESS Model 6 was run. Support for punitive measures was entered as the outcome variable, emotional manipulation as the predictor variable. Mediator variables were entered in the following order: anger, attitude, controlling for ideology, gender, and race. Three models of indirect effects were tested. Specifically, Model 1 tested whether elicited anger mediated the relationship between anger manipulation and support for punitive policy measures, Model 2 tested the following casual chain: anger manipulation  anger attitude  support for punitive measures, and Model 3 tested whether attitude mediated the effect of anger manipulation on support for punitive measures. The total effect model was significant, R2 = .103, F (4, 240) = 6.860, p = .001. Emotional manipulation was negatively related to support for punitive measures, b = -.284, SE = .109, t = -2.607, p = .010. The serial mediation model was significant, R2 = .232, F (6, 238) = 11.967, p = .001. Emotional manipulation was not significantly related to support for punitive measures after mediators were taken into account, b = -.131, SE = .104, t = -1.260, p = .209. All of the three models were viable. Indirect effect of Model 1 was significant, b = -.098, SE = .038, 95% bootstrap confidence interval: -.1881, -.0342. Anger appeal elicited anger responses which in turn increased support for punitive measures. Indirect effect of Model 2 was significant, b = -.012, SE = .007, 95% bootstrap confidence interval: -.0339, -.0028. An examination of the coefficients revealed that emotional manipulation was negatively related to elicited anger (b = -.405, SE = .134, p = .003), elicited anger was 135 negatively related to attitude (b = -.136, SE = .047, p = .004), and attitude was negatively related to support for punitive measures (b = -.210, SE = .068, p = .002). In other words, anger appeal elicited anger responses which in turn led to more negative attitudes towards health providers involved in errors, and those with more negative attitudes were more likely to support punitive measures. Thus, H5c was supported. In addition, indirect effect of Model 3 was also significant, b = -.044, SE= .028, 95% bootstrap confidence interval: -.1211, -.0051. Anger appeal led to more negative attitudes which in turn increased support for punitive measures. There were three pairwise contrasts between the three indirect effects. Results indicated that contrast between Model 1 and Model 2 indicates significant difference in the strength of the two indirect effects, b = -.086, SE = .036, 95% bootstrap confidence interval: -.1726, -.0280. That is, the indirect effect via elicited anger responses was significantly greater than the indirect effect via anger responses and attitude. 136 Table 1 Tests of Framing Manipulation, Emotional Manipulation, and Their Interaction on Anger and Fear Responses IV Framing Manipulation Emotional Manipulation Framing x Emotional Manipulation Interaction Multivariate F DV Univariate F df Stepdown F df 1.987 Anger Response 3.301† 1/238 0.8 1/237 Fear Response 3.176† 1/238 0.677 1/237 8.182*** Anger Response 9.014** 1/238 16.293*** 1/237 2.556† Fear Response Anger Response 0.066 4.307* 1/238 1/238 7.118** 1.09 1/237 1/237 Fear Response 4.021* 1/238 0.809 1/237 Note: N = 245, †p < .10 *p < .05 **p < .01 ***p< .001 137 Table 2 Conditional Indirect and Direct Effects of Emotional Manipulation on Risk Perception Mediator Anger Fear Moderator Framing Episodic Framing Thematic Framing Framing Episodic Framing Thematic Framing Conditional Direct Effects Moderator Framing Episodic Framing Thematic Framing b (SE) 95% CI -.0246 (.0394) -.1284 (.0543) -.1172, .0418 -.2602, -.0439 .0022 (.0244) -.0015 (.0179) -.0406, .0649 -.0485, .0298 b (SE) 95% CI -.3342 (.1432) .2632 (.1426) -.6163, -.0522 -.0178, .5441 Notes. 1000 bootstrap samples with 95% confidence intervals. Bootstrapping reveals that elicited anger reliably mediates the relationship between the anger manipulation and risk perception for the thematic framing condition as the 95% confidence intervals for the coefficients do not overlap zero. Bootstrapping also reveals that there was residual direct effect of anger manipulation on risk perception for episodic framing condition, as the 95% confidence interval for the coefficient does not overlap zero. 138 Table 3 Conditional Indirect and Direct Effects of Emotional Manipulation on Attitude Mediator Anger Fear Moderator Framing Episodic Framing Thematic Framing Framing Episodic Framing Thematic Framing Conditional Direct Effects Moderator Framing Episodic Framing Thematic Framing b (SE) 95% CI .0126 (.0238) .0664 (.0496) -.0160, .0859 -.0154, .1852 -.0197 (.0252) .0152 (.0218) -.0937, .0110 -.0085, .0906 b (SE) 95% CI .2402 (.1379) .2116 (.1388) -.0316, .5119 -.0618, .4849 Notes: 1000 bootstrap samples with 95% confidence intervals. Bootstrapping reveals that there was no significant moderated mediation as the 95% confidence intervals for the coefficients overlapped zero. 139 Table 4 Tests of Framing Manipulation, Emotional Manipulation, and Their Interaction on Policy Support IV Framing Manipulation Emotional Manipulation Multivariate F .078 3.764* Framing x Emotional .072 Manipulation Interaction Note: N = 245, †p < .10 Univariate DV F df Stepdown F df Punitive Measures .085 1/238 .018 1/237 Remedial Measures .139 1/238 .071 1/237 Punitive Measures 6.764* 1/238 7.356** 1/237 Remedial Measures .168 1/238 .771 1/237 Punitive Measures .046 1/238 .002 1/237 Remedial Measures .143 1/238 .098 1/237 *p < .05 **p < .01 ***p < .001 140 Figure 1. Interaction of Framing and Emotional Manipulation on Risk Perception. 141 Figure 2. Moderated Mediation for Risk Perception. Note: N = 243. †p < .10 *p < .05 **p < .01 ***p < .001 142 Figure 3. Moderated Mediation for Attitude. Note: N = 245. †p < .10 *p < .05 **p < .01 ***p <. 001 143 Figure 4: Indirect Effect of Emotional Appeal on Support for Remedial Measure Through Fear and Risk Perception. Note: N = 243. †p < .10 *p < .05 **p < .01 ***p < .001 144 Figure 5. Indirect Effect of Emotional Appeal on Support for Remedial Measure Through Anger and Risk Perception. Note: N = 243. †p < .10 *p < .05 **p < .01 ***p < .001 145 Figure 6. Indirect Effect of Anger Appeal on Support for Punitive Measure Through Anger and Atititude Towards Health Providers Involved in Errors Note: N = 245. †p < .10 *p < .05 **p < .01 ***p < .001 CHAPTER 8 DISCUSSION News framing research has mainly focused on cognitive processes such as accessibility and applicability effects as the primary mechanisms underlying framing effects. Recently, an increasing number of researchers have started to integrate discrete emotional reactions into the study of news framing effects (e.g., Goodall, et al., 2013; Gross, 2008; Gross & D‘Ambrosio, 2004; Kühne & Scheme, 2015; Lecheler et al., 2013). These studies examined whether emotional responses to news frames would function as mediators when it comes to the effects of news frames on information processing, attitude, and opinion formation. In general, their findings show that message-relevant discrete emotions could have specific consequences for attitude and opinion formation. As such, there is a pressing need to ―supplement the classical framing effects approaches with emotion theories‖ to better understand how news frame affect opinion and attitude via both cognitive and affective approaches (Kühne & Scheme, 2015, p. 401). The current study followed this line of research and examined anger and fear as two discrete emotional reactions to news coverage of medical errors and explicitly tested 147 whether these two emotions functioned as mediators of the effects of news frames. There are two reasons for the emphasis on anger and fear. First, extant literature tends to indicate that anger and fear are active and intense emotions associated with differential appraisal tendencies and goals. For example, anger has been found to be associated with risk-seeking tendencies and decreased risk perceptions while fear has been found to be associated with risk-aversive tendencies and increased risk perceptions. Anger tends to trigger appraisals of individual control and thus more negative attitudes towards causal agents while fear tends to trigger appraisals of situational control and thus would alleviate negative attitudes towards causal agents. In addition, anger has been found to be associated with preference for punitive solutions while fear has been found to be associated with preference for solutions that focus on protection and precaution. Therefore, risk perception regarding medical errors, attitudes towards responsible health providers and support for remedial and punitive policy measures were the outcome variables of interest for this study. Moreover, a group of researchers argued that intersecting news frames will influence framing effects (Major, 2009; Shah, Kwak, Schmierbach, & Zubric, 2004). According to Major (2009), the scarcity of research on the intensifying or diminishing effects of combined frames is a ―serious oversight in framing theory research because single-frame stories rarely, if ever, occur in real news coverage‖ (p. 186). Prior studies of the emotional effects of news frames employed either emotional frames or episodic/thematic frames, exposing participants to only one type of frame. For example, Kühne and 148 Schemer (2015) investigated an ―anger‖ frame and a ―sadness‖ frame about a traffic accident on information processing and policy preference. Kim and Cameron (2011) compared the effects of an ―anger‖ frame and a ―sadness‖ frame about an organizational crisis on subsequent perception and attitude towards the causal agent. Goodall et al. (2012) elicited anger or fear responses to news of crimes/accidents by making or not making reference to alcohol. Both their anger-inducing condition (i.e., alcohol version) and fear-inducing condition (i.e., nonalcohol version) were episodically framed that described the incidents with victims and perpetrators. This group of studies compared the effects of emotional frames on emotional reactions and opinions without manipulating episodic versus thematic frames. Another group of studies compared the emotional effects of episodic and thematic frames without manipulating the type of discrete emotions. For example, Gross (2008) explored the effects of episodic and thematic frames about a woman receiving an unduly harsh sentence on emotional response and policy opinion. Results indicated that multiple emotional reactions were elicited by episodic and thematic framing including pity, anger, disgust, worry, sympathy, and fear. Similarly, Aarøe (2011) compared episodic and thematic framing about a Danish immigration policy on emotional reactions and found that episodic framing elicited more compassion, pity, anger, and disgust than thematic framing. As indicated by these two groups of studies, the lack of attention to the effects of intersecting episodic/thematic framing and emotional manipulation has left a sizable gap 149 in the knowledge base of framing effects on emotional responses. Thus, this study provides some initial insights into this aspect by intersecting episodic/thematic framing with anger versus fear manipulation to investigate whether and how emotional manipulation, in combination with episodic/thematic framing, would affect individuals‘ emotional responses and those outcome variables of interest. Experimental Effects of Framing and Emotional Manipulation on Anger and Fear Responses There was successful anger induction such that people reading anger-inducing news stories expressed significantly more anger responses than those reading fear-inducing stories. Yet there was no successful fear induction such that people reading fear-inducing stories reported experiencing almost the same amount of fear response as those reading anger-inducing stories. This might be due to the fact that medical errors is a topic for which individuals‘ baseline emotion is fear, thus it was relatively hard to manipulate their level of fear with emotional stimuli. In the most general sense, findings from the current study showed that ―exposure to news frames that bear emotional relevance will invoke emotional reactions‖ (Lecheler et al., 2013, p. 202). Previous research indicates that episodic frames tend to be more emotionally provoking and could elicit stronger emotional responses than thematic frames (e.g., Aarøe, 2011; Gross, 2008), as episodic frames present vivid depictions of personal characters and individual experiences. The human interest details presented in the episodic frame 150 put a real face on the presentation of the problem (Aarøe, 2011； Semetko & Valkenburg, 2000) and thus provide a specific personal victim at which individuals could direct their emotional reactions. Inconsistent with previous findings, the current study found that there was no significant mean difference between episodic and thematic framing conditions with respect to anger and fear responses. In other words, the episodic frame elicited comparable level of anger and fear responses as the thematic frame. In addition, there was not a significant interaction between framing and emotional manipulation on anger and fear responses. The episodic fear story that delineated the horrifying experiences of a patient did not elicit more fear responses than the thematic fear story that underscored system failures such as communication breakdowns. The episodic anger story that presented the infuriating experiences of the patient did not elicit more anger responses than the thematic anger story that provides analyses of why responsible doctors went unpunished nationwide. The reason might be that the thematic fear story explained the fundamental risks and broader causes beyond the control of individual health providers, making it as effective, if not more so, than the episodic fear story in eliciting fear. By the same token, the thematic anger story presented a compelling picture of the injustice embedded in the supervisory system, making it as powerful, if not more so, than the episodic anger story in stimulating anger. 151 Effects of Anger Versus Fear Manipulation on Risk Perception, Attitude, and Support for Remedial Versus Punitive Policy Measures Previous researchers found that message-relevant emotions would entail specific consequences for subsequent attitude and opinion formation (e.g., Goodall et al., 2013; Gross, 2008; Kühne & Scheme, 2015; Nabi, 2003, 2007). The results of this study agree with previous research findings that discrete emotions induced by a news frame would promote emotion-congruent attitude and opinion. The current study examined the effects of anger versus fear manipulation on four dependent variables regarding medical errors and whether these effects were moderated by framing condition. In general, this study found significant influence of anger manipulation but not fear manipulation. Message-induced fear did not produce the patterns of result that was hypothesized. Specifically, concerning risk perception of medical errors, it was found that there was no significant main effect of emotional manipulation such that fear-inducing stories failed to elicit higher level of risk perceptions than anger-inducing stories. In other words, fear appeal did not work for risk perception. Concerning policy support, it was found that there was no significant main effect of emotional manipulation on support for remedial measures. That is, fear manipulation did not elicit more support for remedial policy measures that focus on protective/precautionary measures when compared to anger manipulation. To summarize, message-induced fear did not give rise to emotion-congruent risk perceptions and policy preference. 152 The failure to establish a successful manipulation of fear can be one explanation for these findings. The unsuccessful fear induction means that the induction of fear also increased the tendency of the fear-inducing news group to experience anger. That is, those in the fear-inducing condition experienced almost comparable level of anger and fear. Kühne and Scheme (2015) pointed out that mixed emotions elicited in response to media content can ―complicate the identification of specific emotional effects, because particular emotions may have contradictory effects on attitude formation‖ (p. 402). Anger and fear, for instance, should have contradictory effects on appraisals of certainty and individual control and hence on risk perceptions regarding medical errors. While anger reduces perceived risk for experiencing errors, fear should increase such perceptions. It may be posited that a mix of anger and fear reactions actually promoted ambivalent risk perceptions. Anger and fear should also have differential effects on policy preference such that while anger promotes support for punitive policies, fear should increase support for remedial policies. A mix of anger and fear would, therefore, promote ambivalent attitudes towards policy measures. As a result, although it can be posited that fear could have dominated the ―perception, processing, and opinion formation‖ for the fear-inducing group, ―the experience of anger may have interfered with these processes‖ (Kühne & Scheme, 2012a, p. 22). In other words, due to the influence of anger, the impact of message-relevant fear may have been suppressed such that the preference for punitive policy measures activated by anger may have suppressed the tendency of the fear-inducing group to be attentive to reparative or mitigation solutions and that the 153 interference of anger may have also watered down the tendency of the fear-inducing news group to perceive risks regarding medical errors. In this case, anger appeared to be a powerful emotion that could have suppressed the effect of fear. Future studies should further explore the effects of mixed emotions on attitude and opinion formation and whether the effect of anger would still suppress other type of negative emotions. Moreover, it is noticeable that there was a significant interaction of emotional and framing manipulation on risk perception such that the effect of emotional appeal on risk perception was different across episodic and thematic framing conditions. Specifically, for the thematic framing condition, the difference between a fear-inducing and an anger-inducing story on risk perception was in the right direction but not statistically significant, whereas for the episodic framing condition, a fear-inducing story elicited significantly less risk perception than an anger-inducing story. Such a finding is relatively inconsistent with previous studies that have generally found that anger would reduce risk perception and increase risk-seeking tendencies while fear would increase risk perception and decrease risk-seeking tendencies. Theoretically, since anger co-occurs with perceived high certainty and control over what has happened, it should encourage risk-seeking behavior and reduce risk estimation, contrasted with fear which activates perceptions of great uncertainty and low individual control, thus increasing risk perceptions and risk-seeking tendencies. The finding from this study indicated that when episodically framed, an anger appeal could elicit more risk perception than a fear appeal. This finding has practical implications for public health message design in health promotion 154 campaigns which are typically designed to elicit fear to motivate the public to take a particular action. By contrast, the present research provides strong evidence for the impact of message-induced anger on attitudes towards health providers involved in medical errors and preference for punitive measures. Specifically, it was found that anger-inducing stories, when compared with fear-inducing stories, significantly enhanced respondents‘ negative attitudes towards health providers involved in medical errors and increased their support for punitive policy measures that focus on punishing individual health providers. In addition, there was not a significant interaction between framing and emotional manipulation on attitude and support for punitive measures. These findings confirm the results obtained by previous studies (e.g., Goodall, et al., 2013, Kühne & Scheme, 2012a; Kühne & Scheme, 2015; Nabi, 2003) that anger was a strong predictor for preference for retributive measures and more negative attitudes towards causal agents involved in a negative issue. In a more general sense, the findings from the current study support existing research and theory about the discrete emotion of anger. Researchers have generally agreed that anger is an intense emotion that could mobilize a strong blaming tendency and activate appraisals of individual responsibility for a negative event. Consistent with previous research, this study found that those integrally angry participants tended to perceive medical errors as primarily caused by individual human error on the part of health providers such as recklessness or negligence. As a result, they tended to regard individual 155 health providers as highly blameworthy for errors, and hence had more negative attitudes and greater preference for individually oriented punitive policies. Moderated Mediation and Serial Mediation This study also examined whether elicited anger and fear mediated the effects of emotional manipulation on risk perception and attitude and whether these indirect effects would be moderated by framing. Results of moderated mediation showed that anger manipulation increased risk perceptions via elicited anger and that fear manipulation did not play a role with respect to risk perceptions. Specifically, there was a positive relationship between anger manipulation and risk perception through elicited anger for thematic frames but not for episodic frames. That is, exposure to a thematically framed anger-inducing story would elicit anger responses, which in turn increased risk perceptions regarding medical errors. In addition, anger manipulation had a residual direct effect on risk perception under the episodic frame that was not explained by elicited anger. In other words, exposure to an episodically framed anger-inducing story would directly lead to increased risk perceptions. In addition, moderated mediation was not significant with respect to attitude towards health providers involved in errors. To predict the effects of emotional manipulation on two types of policy measures, serial mediation analyses were conducted. Specifically, it was hypothesized that fear appeal would increase support for remedial measures via elicited fear and risk perceptions. This hypothesis was not supported. In addition, results indicated that fear did 156 not mediate the relationship between fear appeal and support for remedial measures and that risk perception did not mediate the relationship between fear appeal and support for remedial measures. Conversely, there was a significant indirect effect of anger appeal on support for remedial measures via elicited anger and risk perceptions. The significant serial mediation indicated that exposure to an anger appeal elicited anger responses which in turn increased risk perceptions of medical errors, and those with more risk perceptions showed more support for remedial policy measures. This serial mediational effect was smaller than the simple mediation effect via elicited anger (i.e., exposure to anger-inducing stories elicited anger which in turn led to more support for remedial measures). In other words, the indirect effect of anger appeal on support for remedial measures via elicited anger was significantly greater than the indirect effect via anger and risk perception. Concerning support for punitive policy measures, there was significant serial mediation such that exposure to an anger appeal would elicit anger responses which in turn enhanced negative attitudes towards those involved in errors, and those with more negative attitudes tended to show more support for punitive policy measures. This serial mediational effect was smaller than the simple mediation effect via elicited anger (i.e., exposure to anger-inducing stories elicited anger which in turn led to more support for punitive measures). In other words, the indirect effect of anger appeal on support for punitive measures via elicited anger was significantly greater than the indirect effect via anger and attitude. In addition, serial mediation analysis revealed another significant 157 simple mediation via attitude. That is, anger appeal increased negative attitude towards responsible health providers, which in turn led to more support for punitive measures. This simple mediation model did not differ significantly from the other two models in terms of explained variance. To summarize, in this study, the role of anger is highlighted as anger appeal was found to be a potent predictor for negative attitude towards causal agents and support for punitive policy measures while fear appeal was not a significant predictor for risk perception and support for remedial measures. Elicited anger was a significant mediator such that it mediated the indirect effect of anger appeal on risk perception for those in the thematic framing condition. It also mediated the indirect effect of anger appeal on support for punitive and remedial policy measures. Future research should further explore the role of anger and its unique effects concerning human judgments and decisions, as this study as well as a number of prior studies consistently found that anger is a robust predictor for emotion-congruent attitude and opinion formation. Anger deserves special attention in the study of emotional effects of news frames. Limitations and Future Directions There are a number of limitations to the study. First and foremost, according to O‘Keefe (2003), the current study employed ―effect-based message variable definitions‖ in which a message variable is conceptualized in terms of its effects on psychological states or emotional arousals (p. 251). The definitions of fear and anger appeal variations 158 in the current study have relied on effect-based definitions of that message variations. That is, fear appeal is defined on the basis of its effect on fear arousal and anger appeal is defined on the basis of differences in aroused anger. According to O‘Keefe, such effect-based message variable definitions ―should be avoided in favor of definitions expressed in terms of intrinsic message features‖ as the former do not offer much illumination on the role of message features in persuasion (p. 251). One would not have learned much about exactly how to design a message to arouse a certain emotional or psychological state by reviewing studies that employed effect-based message definitions. Findings from these studies might be significant in terms of revealing the relationship between psychological states and persuasive outcomes, but they provide little guidance or help to message designers who would want to know exactly what message features or characteristics need be incorporated to induce an emotional state. By contrast, studies defining message variables in terms of intrinsic message features would shed light on the question of specific message traits or properties that produce the observed effects on psychological states. Furthermore, a manipulation check would be unnecessary for studies defining message variables in terms of intrinsic message features. According to O‘Keefe (2003), such assessments should not be treated as a reflection of the adequacy of the manipulation of message variables and are better analyzed as potential assessment of mediating states. For example, no manipulation check is needed for a study that compared the persuasive effects of two messages differing in the length (i.e., number of 159 words) as participants‘ perception of the length of the message did not affect the actual length of the message. In other words, the length of the message, which is an intrinsic message feature, is independent of participants‘ perception and thus the manipulation check would not be an appropriate assessment of whether the message property has been manipulated. Another example would be that an experimental variation of message containing or not containing an introductory metaphor does not necessitate manipulation check (O‘Keefe, 2003). For the current study, episodic and thematic news frames could be understood as definitions based on intrinsic message features as it is clear that an episodic frame involves episodic elements such as personal accounts or individual experiences and a thematic frame includes thematic elements such as statistics, policy discussions, and analysis of the broader trends. According to O‘Keefe‘s conceptualization, no manipulation check is needed as these elements are already incorporated in the stimuli articles, regardless of participants‘ perception. Anger and fear appeal, on the other hand, are effect-based message variable definitions. Therefore, future studies would substitute them for definitions expressed in terms of intrinsic message features, such as an injustice frame for anger appeal and an uncertainty frame for fear appeal. In other words, an injustice frame focuses on injustice-related message features to induce anger and an uncertainty frame focuses on uncertainty-related message features to induce fear. Future studies orientating towards specific message features in eliciting anger or fear would be more insightful when it comes to designing communication messages and theorizing about what message features influence persuasive outcomes. 160 Second, participants were exposed to either anger or fear manipulation in the experiment. However, realistic news coverage of medical errors would apply both anger and fear appeals such that individuals would experience mixed emotions. For example, news coverage of medical errors could depict the victim‘s unsuccessful attempt to sue the involved doctors which could induce anger and could also delineate the horrifying consequences of errors which could cause fear. If the story emphasizes the suffering of the victim, individuals will experience sadness and/or compassion. This implies that realistic news coverage that applies different perspectives or emotional appeals may contribute to ambivalent attitudes and opinions. Therefore, future studies are needed to investigate how mixed emotions induced by multiple emotional appeals or competitive news frames would influence perception of the issue and what their consequences are for the structure of attitude and opinion. Such studies should also ―include additional measures of policy preferences that are not related to punishment or remediation‖ to further explore what preferences are promoted by alternative emotions that may be elicited (Kühne & Schemer, 2015, p. 402). In a similar vein, future studies need to employ measures of additional emotional responses to test whether the news story has elicited unanticipated emotions. The present study used a 10-item measure to gauge respondents‘ emotional reactions to news stories of medical errors. Factor analyses indicated a two-factor solution of anger and fear. However, it is likely that the manipulation also elicited other emotions, such as sadness or compassion. It is possible that participants would also experience sadness and/or 161 compassion to news stories of medical errors. The core relational theme of sadness is irreversible loss (Nabi, 1999) and in mediated communication, news coverage of victims‘ personal experiences can make people feel sad (Kim & Cameron, 2011). Compassion, according to Lazarus (1991), is felt by ―feeling personal distress at the suffering of another and wanting to ameliorate it,‖ and the core relational theme is ―being moved by another‘s suffering and want to help‖ (p. 289). More importantly, similar to fear, previous studies have shown that sadness and compassion could increase the preference for remedial, mitigation measures that provide protection against future occurrence of negative events (Goetz, Keltner, & Simon-Thomas, 2010; Kühne & Scheme, 2015). Future studies could include more emotion items to capture the complexity of emotional responses to experimental stimuli of this nature, as research indicates that pure emotional states are exceedingly rare in normal daily life and, more likely, individuals would experience mixed or blended motions at the same time (e.g., Dillard et al. 1996). Another issue that requires further empirical investigation concerns the process underlying news framing effects via elicited emotions. Based on the theoretical models proposed by a group of researchers (e.g., Kühne & Scheme, 2015; Nabi, 2007; Nerb & Spada, 2001), the present study posited that exposure to news stories activates cognitive appraisals, which in turn gives rises to specific discrete emotions. The induced emotions then influence subsequent attitude and opinion formation in an emotion-congruent way. The results of mediational analyses indicated that message-relevant anger was a significant mediator between exposure to an anger appeal and risk perception, support for 162 remedial and punitive measures, making these process assumptions highly plausible. Nevertheless, additional empirical research is needed to explicitly test the specific steps underlying the emotional framing effects. In particular, future studies need to examine the link between cognitive appraisal dimensions and elicitation of discrete emotions. More understanding about the processes of emotional framing effects could be gained by taking into account potentially moderating variables of this mediated effect. While the present study has analyzed how framing moderated the effects of anger versus fear manipulation on attitude and opinion, future studies need to include variables that influence emotional elicitation. Extant literature shows that there are variables that might moderate individuals‘ emotional reactions to news stories, such as the need for affect which influence individuals‘ propensity to experience emotions. There may be other moderators that could impact this mediated effect such as prior attitudes and beliefs, or political sophistication and knowledge. Future studies would gain deeper insights into the emotional effects of news frame with the integration of specific moderators. Last but not least, a convenience sample of students was employed. Although it could be posited that ―basic psychological processes of emotional response are universal‖ and that the findings from this study could be extrapolated to other populations, it is suggested that future studies should use representative sample (Lecheler et al. 2013, p. 203). 163 Conclusion This study provides building blocks for the integration of discrete emotions into predominantly cognitive framing research. This study clearly demonstrates that cognitive effects models such as accessibility and applicability effects are not the only mechanisms underlying framing effects, as news frames would induce discrete emotions, which in turn entail specific attitudinal consequences. Message-relevant anger induced more negative attitude towards causal agents and promoted support for punitive policy measures. More importantly, this study contributes to the burgeoning body of research testing discrete emotions as mediators of news framing effects. Results indicated that elicited anger mediated the effects of the anger appeal on risk perception and support for punitive and remedial policy measures. These effects imply that anger appeals have great potential to be an effective way of mobilizing people to support a certain policy initiative. Overall, findings from this study indicated the importance of discrete emotions in directing human thoughts, judgments, and opinions in the setting of mediated communication. APPENDIX A SWISS CHEESE MODEL OF MEDICAL ERRORS APPENDIX B PYRAMID ANALOGY OF MEDICAL ERRORS APPENDIX C SUMMARY OF EMOTIONAL MEASURES Table 5 Summary of Emotional Measures Author(s) Number Scales used to measure Reliability Validity Scale Method of of Items anger and fear Checks Checks Type Analysis Likert Factor Scale Analysis Likert Factor Scale Analysis Dillard, Plotnick, Godbold, Freimuth, and Anger 3 (angry, irritated, and annoyed) Yes, α= .84 Yes Yes, α= .94 Yes Yes, α= .94 No Yes, α= .95 No Yes, α= .90 Yes Yes, α= .93 Yes Yes, α= .95 No Yes, α= .95 No Edgar (1996) Fear Yan and Dillard (2010) (2000) Anger Shen (2012) annoyed, and aggravated) 3 (fear, afraid, and scared) 4 (irritated, angry, Anger annoyed, and aggravated) Fear Yan, Dillard, and scared) 4 (irritated, angry, Fear Dillard and Peck 3 (fearful, afraid, and 3 (fearful, afraid, and scared) 4 (irritated, angry, Anger annoyed, and aggravated) Fear 3 (fearful, afraid, and scared) Likert Scale Likert Scale Likert Factor Scale Analysis Likert Factor Scale Analysis Likert Scale Likert Scale 167 Table 5 continued Author(s) Number Scales used to measure Reliability Validity Scale Method of of Items anger and fear Checks Checks Type Analysis Dillard, Kinney, and 4 (angry, annoyed, Anger Cruz (1996) aggravated, and Yes, α= .95 Yes Yes, α= .92 Yes irritated) Fear 3 (fearful, afraid, and scared) Likert Scale Likert Scale Dillard and Anderson Anger (2004) Yes, α= .83; Fear 3 (fearful, afraid, and Yes, scared) α= .94 ;Yes, No Likert Scale α= .91 Shen and Dillard 4 (irritated, angry, Anger (2007) Anger (primed anger) Fear (primed fear) Kim and Cameron Anger (2011) Fear Kühne and Schemer Yes, α= .87 Yes Yes, α= .92 Yes No No No No Yes, α= .93 Yes Yes, α= .97 Yes Yes, α= .94 Yes Yes, α= .86 Yes aggravated) Fear Nabi (2003) annoyed, and Anger (2015) Fear 3 (fearful, afraid, and scared) Likert Factor Scales Analysis Likert Factor Scales Analysis 8 (angry, irritated, tense, annoyed, frustrated, irate, pissed Likert Scale off, and mad) 8 (frightened, anxious, tense, fearful, uneasy, alarmed, nervous, and Likert Scale afraid) 3 (angry, irritated, and aggravated) 3 (scared, fearful, and afraid) 3 (anger, furious, and annoyed) 3 (fear, anxiety, and faint-hearted) Likert Factor Scales Analysis Likert Factor Scales Analysis Likert Factor Scales Analysis Likert Factor Scales Analysis 168 Table 5 continued Author(s) Number Scales used to measure Reliability Validity Scale Method of of Items anger and fear Checks Checks Type Analysis Lecheler, Schuck, and De Vreese Afraid 1 (single-item) No No Angry 1 (single-item) No No Angry 1 (single-item) No No Fear 1 (single-item) No No Anger 1 (single-item) No No Fear 1 (single-item) No No Anger 1 (single-item) No No Fear 1 (single-item) No No Anger 1 (single-item) No No Fear 1 (single-item) No No (2013) Gross and Brewer (2007) Nerb and Spada (2001) Goodall, Slater, and Myers (2013) Nabi (2002) Dunn and Schweitzer (2005) Anger 3 (angry, mad, and irritated) Yes, α= .78; No α= .89 Likert Scales Likert Scales Likert Scales Likert Scales Likert Scales Likert Scales Likert Scales Likert Scales Likert Scales Likert Scales Likert Scales Fear Gordijn, Yzerbyt, Wigboldus, and Anger Dumont (2006) 3 (angered, outraged, Yes, and annoyed) α= .80 Fear Iyer, Schmader, and Lickel (2007) Anger Fear Yes Yes 4 (furious, outraged, Yes, angry, incensed) α= .80 Yes Likert Factor Scales Analysis Likert Factor Scales Analysis Likert Factor Scales Analysis 169 Table 5 (continued) Author(s) Number Scales used to measure Reliability Validity Scale Method of of Items anger and fear Checks Checks Type Analysis Likert Facotr Scales Analysis Likert Facotr Scales Analysis Likert Facotr Scales Analysis Likert Facotr Scales Analysis Mackie, Devos, and Smith Anger (2000) Fear Dillard and Peck (2001) 4 (angry, displeased, Yes, irritated, and furious) α> .62 4 (worried, anxious, Yes, afraid, and fearful) α> .76 4 (irritated, angry, Anger annoyed, and aggravated) Fear Yes, α = .88 3 (fearful, afraid, and Yes, α scared) = .94 Yes Yes Yes Yes APPENDIX D EPISODIC ANGER CONDITION APPENDIX E THEMATIC ANGER CONDITION APPENDIX F EPISODIC FEAR CONDITION APPENDIX G THEMATIC FEAR CONDITION REFERENCES Aarøe, L. (2011). Investigating frame strength: The case of episodic and thematic frames. Political Communication, 28, 207-226. doi:10.1080/10584609.2011.568041 Abram, S. (2014, April 26). 325, 000 deaths in American hospitals due to medical errors. San Gabriel Valley Tribune. Retrieved from LexisNexis Academic Database. Akhavan-Majid, R., & Ramaprasad, J. (1998). Framing and ideology: A comparative analysis of U.S. and Chinese newspaper coverage of the Fourth United Nations Conference on Women and the NGO Forum. Mass Communication & Society, 1, 131-152. Allen, M. (2013, September 19). How many die from medical mistakes in U.S. hospitals? Retrieved from https://www.propublica.org/article/how-many-die-from-medical -mistakes-in-us-hospitals. Alvarez, G., & Coiera, E. (2005). Interruptive communication patterns in the intensive care unit ward round. International Journal of Medical Informatics, 74, 791-796. Andel, C., Davidow, S. L., Hollander, M., & Moreno, D. A. (2012). The economics of health care quality and medical errors. Journal of Health Care Finance, 39, 3950. Andrews, L. B., Stocking, C., Krizek, T., Gottlieb, L., Krizek, C., Vargish, T., & Siegler, M. (199). An alternative strategy for studying adverse events in medical care. LANCET, 349, 309-312. Arora, V., Johnson, J., Lovinger, D., Humphrey, H. J., & Meltzer, D.O. (2005). Communication failures in patient sign-out and suggestions for improvement: A critical incident analysis. Quality and Safety in Health Care, 14, 401-407. Averill, J. R. (1982). Anger and aggression. An essay on emotion. (Springer series in social psychology). R.F. Kidd (Ed.). New York, NY: Springer- Verlag. 175 Baden, C., & Lecheler, S. (2012). Fleeting, fading, or far-reaching? A knowledge-based model of the persistence of framing effects. Communication Theory, 22, 359-382. Bakalar, N. (2016, May 3). Medical errors cause over 250, 000 deaths a year. The New York Times. Retrieved from https://well.blogs.nytimes.com/2016/05/03/ medical-errors-may-cause-over-250000-deaths-a-year/ Baldwin, A., & Rodriguez, E. S. (2016). Improving patient safety with error identification in chemotherapy orders by verification nurses. Clinical Journal of Oncology Nursing, 20, 59–65. Batson, C. D., Shaw, L. L., & Oleson, K. C. (1992). Differentiating affect, mood, and emotion: Toward functionally-based conceptual distinctions. In M. S. Clark (Ed.), Review of personality and social psychology (Vol. 13, pp. 294-326). Newbury Park, CA: Sage. Baum, M. (2003). Soft news goes to war: Public opinion and American foreign policy in the news media age. Princeton, NJ: Princeton University Press. Berkowitz, L., & Harmon-Jones, E. (2004). Toward an understanding of the determinants of anger. Emotion, 4, 107-130. Berry, T. R., Wharf-Higgins, J. & Naylor, P. J. (2007). SARS wars: An examination of the quantity and construction of health information in the news media. Health Communication, 21, 35-44. Bigman, C., Cappella, J. N., & Hornik, R. C. (2010). Effective or ineffective: Attribute framing and the human papillomavirus (HPV) vaccine. Patient Education and Counseling, 81, 70-76 Binder, L. (2013, September 23). Stunning News on Preventable Deaths in Hospitals. Retrieved from http://www.forbes.com/sites/leahbinder/2013/09/23/ stunningnews-on-preventable/deaths-in-hospitals/#114bc54aad08 Blanchette, I., & Richards, A. (2010). The influence of affect on higher level cognition: A review of research on interpretation, judgment, decision making and reasoning. Cognition & Emotion, 24, 561-595. Blendon, R. J., Desroches, C. M., Brodie, M, Benson, J. M., Rosen, A. B., Schneider, E., Altman, D. E., Zapert, K., Herrmann, M. J., & Steffenson, A. E. (2002). 176 Views of practicing physicians and the public on medical errors. The New England Journal of Medicine, 347, 1933-1940. Bognár, A., Barach, P., Johnson, J. K., Duncan, R. C., Birnbach, D., Woods, D. (2008). Errors and the burden of errors: Attitudes, perceptions, and the culture of safety in pediatric cardiac surgical teams. The Annals of Thoracic Surgery, 85, 1375, 1374-1381. Brennan, T. A. (2000). The Institute of Medicine Report on Medical Errors — Could it do harm? The New England Journal of Medicine, 342, 1123-1125. DOI: 10.1056/NEJM200004133421510 Brodie, M., Hamel, E. C., Altman, D. E., Blendon, R. J., & Benson, J. M. (2003). Health news and the American public, 1996-2002. Journal of Health Politics, Policy and Law, 28, 927-950. Brown, P., Zavestoski, S. M., Mccormick, S., Mandelbaum, J., & Luebke, T. (2001). Print media coverage of environmental causation of breast cancer. Sociology of Health & Illness, 23, 747-775. Brownson, R. C., Mack, N. E., Meegama, N. I., Pratt, M., Browson, C. A. (1996). Changes in newspaper coverage of cardiovascular health issues in conjunction with a community-based intervention. Health Education Research, 11, 479-486. Bryant, J., & Oliver, M. B. (Eds.). (2009). Media effects: Advances in theory and research (3rd ed.). New York, NY: Routledge. Bryant, J., & Thompson, S. (2002). Fundamentals of media effects. Boston, Mass: McGraw-Hill Carver, C. S., & Harmon-Jones, E. (2009). Anger is an approach-related affect: Evidence and implications. Psychological Bulletin, 135, 183-204. doi:10.1037/a0013965 Cha, A. E. (2016, May 3). Researchers: Medical errors now third leading cause of death in United States. The Washington Post. Retrieved from https://www.washingtonpost.com/news/to-your-health/wp/2016/05/03/researcher s-medical-errors-now-third-leading-cause-of-death-in-united-states/ 177 Choi, Y., & Lin, Y.H. (2009). Consumer responses to Mattel product recalls posted on online bulletin boards: Exploring two types of emotion. Journal of Public Relations Research, 21, 198-207. Chong, D., & Druckman, J.N. (2007). Framing theory. Annual review of political Science, 10, 103-122. Cismaru, M., Lavack, A. M., Hadjistavropoulos, H., & Dorsch, K. D. (2008). Understanding health behavior: An integrated model for social marketers. Social Marketing Quarterly, 14, 2-32. Clarke, J. N., & Binns, J. (2006). The portrayal of heart disease in mass print magazines, 1991-2001. Health Communication, 19, 39-48. Clarke, J. N., & Everest, M. M. (2006). Cancer in the mass print media: Fear, uncertainty and the medical model. Social Science & Medicine, 62, 2591-2600. Coleman, R., Thorson, E., & Wilkins, L. (2011). Testing the effect of framing and sourcing in health news stories. Journal of Health Communication, 16, 941-954. doi:10.1080/10810730.2011.561918 Collins, S. J., Newhouse, R., Porter, J., & Talsma, A. (2014). Effectiveness of the surgical safety checklist in correcting errors: A literature review applying reason's Swiss cheese model. AORN Journal, 100, 65-79. doi:10.1016/j.aorn.2013.07.024 Corbett, J. B., & Mori, M. (1999). Medicine, media, and celebrities: News coverage of breast cancer, 1960-1995. Journalism & Mass Communication Quarterly, 76, 229-249. Coombs, W. T. (2007). Protecting organization reputations during a crisis: The development and application of situational crisis communication theory. Corporate Reputation Review, 10, 163-176. doi:10.1057/palgrave.crr.1550049 Coombs, W. T., & Holladay, S. J. (2009). Further explorations of post-crisis communication: Effects of media and response strategies on perceptions and intentions. Public Relations Review, 35, 1-6. Cosand, K. Q. (2014). Black and blue and read all over: News framing and the coverage of crime. Portland State University (Master‘s Thesis). Available from ProQuest Database. (UMI No. 1560453). 178 Cozzarelli, C., Tagler, M. J., & Wilkinson, A. V. (2001). Attitudes toward the poor and attributions for poverty. Journal of Social Issues, 57, 207-227. Daleyfeb, J. (2015, February 17). For Colorado mom, story of a daughter‘s hospital death is key to others‘ safety. Colorado Public Radio. Retrieved from http://www.cpr.org/news DeSteno, D., Dasgupta, N., Bartlett, M. Y., & Cajdric, A. (2004). Research article prejudice from thin air the effect of emotion on automatic intergroup attitudes. Psychological Science, 15 (0956-7976), 319-324. Desteno, D., Petty, R. E., Rucker, D. D., Wegener, D. T., & Braverman, J. (2004). Discrete emotions and persuasion: The role of emotion-Induced expectancies. Journal of Personality and Social Psychology, 86, 43-56. Desteno, D., Petty, R. E., Wegener, D. T., & Rucker, D. D. (2000). Beyond Valence in the Perception of Likelihood: The role of emotion specificity. Journal of Personality and Social Psychology, 78, 397-416. de Vreese, C. H. (2004). The effects of frames in political television news on issue interpretation and frame salience. Journalism & Mass Communication Quarterly, 81, 36-52. de Vreese, C. H. (2005). News framing: Theory and typology. Information Design Journal & Document Design, 13, 51-62. de Vreese, C.H., Peter, P., & Semetko, H.A. (2001). Framing politics at the launch of the euro: a cross-national comparative study of frames in the news. Political Communication, 18, 107-122. de Vreese, C. H., Boomgaarden, H. G., & Semetko, H. A. (2011). (In)direct framing effects: The effects of news media framing on public support for Turkish membership in the European Union. Communication Research, 38, 179-205. Dillard, J. P., & Anderson, J. W. (2004). The role of fear in persuasion. Psychology & Marketing, 21, 909-926. doi:10. 1002/mar.20040 Dillard, J. P., & Kinney, T. A. (1994). Experiential and physiological responses to interpersonal self-report measures of discrete emotions influence. Human Communication Research, 20, 502-528. 179 Dillard, J. P., & Kinney, T. A., & Cruz, M.G. (1996). Influence, appraisals, and emotions in close relationships. Communication Monographs, 63, 105-130. Dillard, J. P., Plotnick, C. A., Godbold, L.C., Freimuth, V. S., & Edgar, T. (1996). The multiple affective outcomes of AIDS PSAs: Fear appeals do more than scare people. Communication Research, 23, 44-72. Dillard, J. P., & Peck, E. (2000). Affect and persuasion: Emotional responses to public service announcements. Communication Research, 27, 461-435. Dillard, J. P., & Peck, E. (2001). Persuasion and the structure of affect. Human Communication Research, 27, 38-68. Dillard, J. P., & Shen, L. J. (2006). Self-reports of discrete emotions. In R. A. Reynolds, R. Woods, & J. D. Baker (Eds.), Handbook of research on electronic surveys and measurements. Hershey, PA: Idea Group Reference. doi: 10.4018/978-1-59140-792-8 Dunn, J. R. & Schweitzer, M. E. (2005). Feeling and believing: The influence of emotion on trust. Journal of Personality and Social Psychology, 88, 736–748. Edy, J. A., & Meirick, P. C. (2007). Wanted, dead or alive: Media frames, frame adoption, and support for the war in Afghanistan. Journal of Communication, 57, 119-141. doi:10.1111/j.1460-2466.2006.00332.x De Jesus, M. (2013). The impact of mass media health communication on health decision-making and medical advice-seeking behavior of U.S. Hispanic population. Health Communication, 11, 1-5. Ellsworth, P. C., & Scherer, K. R. (2003). Appraisal processes in emotion. In R. J. Davidson, K. R. Scherer, & H. H. Goldsmith (Eds), Handbook of affective sciences (pp. 572-595). New York, NY: Oxford University Press. Detweiler, J. B., Bedell, B. T., Salovey, P., Pronin, E., & Rothman, A. J. (1999). Message framing and sunscreen use: Gain-framed messages motivate beach-goers. Health Psychology, 18, 189. Diaz, J. A., Griffith, R. A., Ng, J. J., Reinert, S. E., Friedmann, P. D., & Moulton, A. W. (2002). Patients' use of the Internet for medical information. Journal of General Internal Medicine, 17, 180-185. 180 Dorfman, L., Wallack, L., & Woodruff, K. (2005). More than a message: Framing public health advocacy to change corporate practices. Health Education & Behavior, 32, 320-336. Dovey, S. M., Phillips, R. L., Green, L. A., & Fryer, G. E. (2003). Types of medical errors are commonly reported by family physicians. American Family Physician, 67, 697. Druckman, J. N. (2001). The implications of framing effects for citizen competence Political Behavior, 23, 225-256. Druckman, J. N. (2004). Political preference formation: Competition, deliberation, and the (ir)relevance of framing effects. American Political Science Review, 98, 671686. Druckman, J., & McDermott, R. (2008). Emotion and the framing of risk choice. Political Behavior, 30, 297-321. doi:10.1007/s11109-008-9056-y Eisler, P., & Hansen, B. (2013, August 20). Thousands of doctors practicing despite errors, misconduct. USA Today. Retrieved from http://www.usatoday.com/story/news/nation/2013/08/20/doctors-licenses-medica l-boards/2655513 Entman, R. M. (1993). Framing: Toward a clarification of a fractured paradigm. Journal of Communication, 43, 51-58. Ferner, R., & Aronson, J. (2000). Medication errors, worse than a crime. The Lancet, 355, 947-948. Fischhoff, B., Gonzalez, R. M., Lerner, J. S., & Small, D. A. (2005). Evolving judgments of judgments of terror risks: Foresight, hindsight, and emotion. Journal of Experimental Psychology: Applied, 11, 124-139. Fiske, S. T., & Taylor, S. E. (1991). Social cognition (2nd ed.). New York: McGraw-Hill. Forgas, J. P. (2000). Feeling and thinking: The role of affect in social cognition. Cambridge: Cambridge University Press. Friedman, D. B., & Hoffman-Goetz, L. (2007). An exploratory study of older adults‘ comprehension of printed cancer information: is readability a key factor? Journal of Health Communication, 12, 423-437. 181 Friedman D. B. & Kao E. K. (2008). A comprehensive assessment of the difficulty level and cultural sensitivity of online cancer prevention resources for older minority men. Preventing Chronic Diseases, 5, 1-12. Friedman, D. B., & Tanner, A. (2007). Evaluating the reading difficulty level of medical sources on television Web sites: Recommendations for a social practices approach to consumer health literacy. Journal of Consumer Health on the Internet, 11, 43-60. Frijda, N. H. (1986). The emotions. Cambridge, UK: Cambridge University Press. Gamson, W. A., Croteau D., Hoynes, W., & Sasson. T. (1992). Media images and the social construction of reality. Annual Review of Sociology, 18, 373-393. Gamson, W. A., & Modigliani, A. (1989). Media discourse and public opinion on nuclear power: A constructionist approach. American Journal of Sociology, 95, 1-37. Gander, P., Purnell, H., Garden, A., & Woodward, A. (2007). Work patterns and fatiguerelated risk among junior doctors. Occupational and Environmental Medicine, 64, 733- 738. Gerend, M. A., & Maner, J. K. (2011). Fear, anger, fruits, and veggies: Interactive effects of emotion and message framing on health behavior. Health Psychology, 30, 420-423. Gitlin, T. (1980). The whole world is watching. Berkeley: University of California Press. Goodall, C. E., Slater, M. D., & Myers, T. A. (2013). Fear and anger responses to local news coverage of alcohol-related crimes, accidents, and injuries: Explaining news effects on policy support using a representative sample of messages and people. Journal of Communication, 63, 373-392. Gokavi, M. (2014, July 14). Sponge left in woman‘s body leads to medical malpractice suit. Dayton Daily News. Retrieved from http://www/daytondailynews.com/new/news/crime-law/sponge-left-in-womans-b ody-leads-to-medical-malpra/ngfnb/ Goldberg, J. H., Lerner, J. S., & Tetlock, P. E. (1999). Rage and reason: The psychology of the intuitive prosecutor. European Journal of Social Psychology, 72, 1017-1033. 182 Gollust, S. E., & Lantz, P. M. (2009). Communicating population health: Print news media coverage of type 2 diabetes. Social Science & Medicine, 69, 1091-1098. Gordijn, E. H., Yzerbyt, V., Wigboldus, D., & Dumont, M. (2006). Emotional reactions to harmful intergroup behavior. European Journal of Social Psychology, 36, 431-432. doi:10.1002/ejsp.346 Grober, E. D., & Bohnen, J. M. A. (2005). Defining medical error. Canadian Journal of Surgery, 48, 39-44. Gross, K. (2008). Framing persuasive appeals: Episodic and thematic framing, emotional response, and policy opinion. Political Psychology, 29, 169-192. Gross, K., & Brewer, P.R. (2007). Sore losers: News frames, policy debates, and emotions. Press/Politics, 12, 122-133. Gross, K., & D‘Ambrosio, L. (2004). Framing emotional response. Political Psychology, 25, 1-29. Gupta, S. (2012, July 31). More treatment, more mistakes. The New York Times Retrieved from http://www.nytimes.com/2012/08/01/opinion/more-treatment-more-mistakes opinion/more-treatment-more-mistakes.html Habib, M., Cassotti, M., Moutier, S., Houdé, O., & Borst, G. (2015). Fear and anger have opposite effects on risk seeking in the gain frame. Frontiers in Psychology, 6, 1-7. doi: 10.3389/fpsyg.2015.00253 Hallin, D. C., & Briggs, C. L. (2015). Transcending the medical/media opposition in research on news coverage of health and medicine. Media, Culture & Society, 37, 85-100. doi:10.1177/0163443714549090 Han, G. (2007). Mainland China frames Taiwan: How China's news websites covered Taiwan's 2004 presidential election. Asian Journal of Communication, 17, 40-57. doi:10.1080/01292980601114547 Han, G., Chock, T. M., & Shoemaker, P. J. (2009). Issue familiarity and framing effects of online campaign coverage: Event perception, issue attitude, and the 2004 presidential election in Taiwan. Journalism & Mass Communication Quarterly, 86, 739-755. 183 Han, S., Lerner, J. S., & Keltner, D. (2007). Feelings and consumer decision making: The appraisal-tendency framework. Journal of Consumer Psychology, 17, 158-168. Hannaford, N., Mandel, C., Crock, C., Buckley, K., Magrabi, F., Ong, M ... Schultz, T. (2013). Learning from incident reports in the Australian medical imaging setting: Handover and communication errors. The British Journal of Radiology, 86(1022), 20120336. doi:10.1259/bjr.20120336 Harth, N. S., Leach, C. W., & Kessler, T. (2013). Guilt, anger, and pride about in-group environmental behavior: Different emotions predict distinct intentions. Journal of Environmental Psychology, 34, 18-26. Hawkins, K. W., & Linvill, D. L. (2010). Public health framing of news regarding childhood obesity in the United States. Health Communication, 25, 709-717. doi:10.1080/10410236.2010.521913 Hayward, R. A., & Hofer, T. P. (2001). Estimating Hospital Deaths Due to Medical errors: Preventability is in the eye of the reviewer. The Journal of the American Medical Association, 286, 415-419. Harrington, M. M. (2005). Revisiting medical error: Five years after the IOM report, have reporting systems made a measurable difference? Health Matrix: Journal of Law-Medicine, 15, 329. Hemenover, S. H., & Zhang, S. (2004). Anger, personality and optimistic stress appraisals. Cognition and Emotion, 18, 363-382. Hong, S. C. (2013). Scare sells? A framing analysis of news coverage of recalled Chinese products. Asian Journal of Communication, 23, 86-106. Doi:10.1080/01292986.2012.717090 Hospital Errors are the Third Leading Cause of Death in U.S., and New Hospital Safety Scores Show Improvements Are Too Slow. (2013, October 23). Retrieved from http://www.hospitalsafetyscore.org/newsroom/display/ hospitalerrors-thirdleading-causeofdeathinus-improvementstooslow Huang, K. S. (1995). A comparison between media frames and audience frames: The case of the Hill-Thomas controversy (Doctoral dissertation). Retrieved from ProQuest Dissertations & Theses database. (9538560). 184 Hyunseo, H., Gotlieb, M. R., Seungahn, N., & McLeod, D. M. (2007). Applying a cognitive-processing model to presidential debate effects: Postdebate news analysis and primed reflection. Journal of Communication, 57, 40-59. doi:10.1111/j.1460-2466.2006.00328.x Igartua, J., Moral-Toranzo, F., & Fernández, I. (2011). Cognitive, attitudinal, and emotional effects of news frame and group cues, on processing news about immigration. Journal of Media Psychology: Theories, Methods, and Applications, 23, 174-185. Iyengar, S. (1990). Framing responsibility for political issues: The case of poverty. Political Behavior, 12, 19-40. Iyengar, S. (1991). Is anyone responsible? How television frames political issues. Chicago, IL: University of Chicago Press. Iyer, A., Schmader, T., & Lickel, B. (2007). Why individuals protest the perceived transgressions of their country: The role of anger, shame, and guilt. Personality and Social Psychology Bulletin, 33, 572-587. Izard, C.E. (1991). The psychology of emotions. New York, NY: Plenum Press. James, J. T. (2013). A new, evidence-based estimate of patient harms associated with hospital care. Journal of Patient Safety, 9, 122-128. Jensen, J. D., King, A. J., Davis, L. A., & Guntzviller, L. M. (2010). Utilization of internet technology by low-income adults: The role of health literacy, health numeracy, and computer assistance. Journal of Aging and Health, 22, 804-826. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263-291. Kahneman, D., & Tversky, A. (1984). Choices, values, and frames. American Psychologist, 39, 341-350. Kalra, J. (2004). Medical errors: Overcoming the challenges. Clinical Biochemistry, 37, 1063-1071. Kalra, J., Kalra, N., & Baniak, N. (2013). Medical error, disclosure and patient safety: A global view of quality care. Clinical Biochemistry, 46, 1161-1169. 185 Karavasiliadou, S., & Athanasakis, E. (2014). An inside look into the factors contributing to medication errors in the clinical nursing practice. Health Science Journal, 8, 32-44. Karlene, H. R., & Robert, G. B. (2001). When systems fail. Organizational Dynamics, 29, 179-191. Keltner, D., Ellsworth, P. C., & Edwards, K. (1993). Beyond simple pessimism: Effects of sadness and anger on social perception. Journal of Personality and Social Psychology, 64, 740-752. doi:10.1037/0022-3514.64.5.740 Kerschbaumer, K. (2000). Television and the Internet: Old meets news. Broadcasting and Cable, 130, 55-58. Kim, H.J., & Cameron, G.T. (2011). Emotions matter in crisis: The role of anger and sadness in the publics‘ responses to crisis news framing and corporate crisis response. Communication Research, 38, 826-855. Kim, S., & Willis, L. A. (2007). Talking about obesity: News framing of who is responsible for causing and fixing the problem. Journal of Health Communication, 12, 59-376. Kinder, D. R. & Sanders, L. M. (1990). Mimicking political debate with survey questions: The case of white opinion on affirmative action for blacks. Social Cognition, 8, 73-103. Kinder, D. R. & Sanders, L. M. (1996). Divided by color: Racial politics and democratic ideals. Chicago: University of Chicago Press. Kirby, R. S. (2003). Learning the lesson from medical errors. BJU International, 92, 4-5. Kohn, L. T., Corrigan, J. M., & Donaldson, M. S. (Eds.) (2000). Committee on quality of healthcare in America, Institute of Medicine. To err is human: Building a safer health system, National Academy Press, Washington (DC) Konijn, E. A. (2013). The role of emotion in media use and effects [Abstract]. In K.E. Dill (Ed.). The Oxford handbook of media psychology (pp. 186-211). New York, NY: Oxford University Press. Krosnick, J. A., & Brannon, L. (1993). The impact of the Gulf war on the ingredients of presidential evaluations: Multidimensional effects of political involvement. American Political Science Review, 87, 963-975. 186 Kuhberger, A. (1995). The framing of decisions: A new look at old problems. Organizational Behavior and Human Decision Processes, 62, 230-240. Kuhberger, A. (1998). The influence of framing on risky decisions: A meta-analysis. Organizational Behavior and Human Decision Process, 75, 23-55. Kühne, R (2012a): Media-induced affects and opinion formation: How related and unrelated affects influence political opinions; Center for Comparative and International Studies & NCCR. http://dx.doi.org/10.5167/uzh-68721 Kühne, R. (2012b). Political news, emotions, and opinion formation: Toward a model of emotional framing effects. Paper presented at the annual meeting of the International Communication Association, Sheraton Phoenix Downtown, Phoenix, AZ Kühne, R. (2014). Political news, emotions, and opinion formation: Toward a model of emotional framing effects. Working Paper No. 68. National Centre of Competence in Research (NCCR) Challenges to Democracy in the 21st Century. Kühne, R., & Schemer, C. (2015). The emotional effects of news frames on information processing and opinion formation. Communication Research, 42, 387-407. Landrigan, C. P., Parry, G. J., Bones, C. B., Hackbarth, A. D., Goldmann, D. A., & Sharek, P. J. (2010). Temporal trends in the trends of patient harm resulting from medical care. The New England Journal of Medicine, 363, 2124-2134. Landrigan, C. P., Rothschild, J. M., Cronin, J. W., Kaushal, R., Burdick, E., Katz, J.T. (2004). Effect of reducing interns' work hours on serious medical errors in intensive care units. The New England Journal of Medicine, 351, 1838 -1848. Lawrence, R. G. (2004). Framing obesity: The evolution of news discourse on a public health issue. Press/Politics, 9, 56-75. Jarlenski, M., & Barry, C. L. (2013). News media coverage of trans fat: Health risks and policy responses. Health Communication, 28, 209-216. doi: 10.1080/ 10410236.2012.669670 Latham, S. R. (2001). System and sesponsibility: Three readings of the IOM report medical error. American Journal of Law & Medicine, 27, 163- 179. Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford University Press. 187 Leape, L. L. (2000). Institute of medicine medical error figures are not exaggerated. Journal of American Medical Association, 284, 95-97. Leape, L. L. (2009). Errors in medicine. Clinica Chimica Acta, 404, 2-5. Leape, L. L., Bates, D. W., Cullen, D. J., Cooper, J., Demonaco, H. J., Gallivan, T., Hallisey, R., Ives, J., Laird, N., & Laffel, G. (1995). Systems analysis of adverse drug events. Journal of American Medical Association, 274, 35-43. Leboeuf, R. A. & Shafir, E. (2003). Deep thoughts and shallow frames: On the susceptibility to framing effects. Journal of Behavioral Decision Making, 16, 77-92. Lecheler, S., & de Vreese, C.H. (2010). Framing Serbia. The effects of news framing on public support for EU enlargement. European Political Science Review, 2, 73-93. Lecheler, S., & de Vreese, C.H. (2012). News framing and public opinion: A mediation analysis of framing effects on political attitudes. Journalism & Mass Communication, 89, 185-204. Lecheler, S., de Vreese, C. H., & Slothuus, R. (2009). Issue importance as a moderator of framing effects. Communication Research, 36, 400-425. Lecheler, S., Schuck, A. T., & de Vreese, C. H. (2013). Dealing with feelings: Positive and negative discrete emotions as mediators of news framing effects. Communications: The European Journal of Communication Research, 38, 189-209. doi:10.1515/commun-2013-0011 Leonard, M. W., & Frankel, A. S. (2011). Role of effective teamwork and communication in delivering safe, high-quality care. The Mount Sinai Journal of Medicine, 78, 820-826. doi:10.1002/msj.20295 Lerner, J. S., Goldberg, J. H., & Tetlock, P. E. (1998). Sober second thought: The effects of accountability, anger, and authoritarianism on attributions of responsibility. Personality and Social Psychology Bulletin, 24, 563-574. Lerner, J. S., Gonzalez, R. M., Small, D. A., & Fischhoff, B. (2003). Effects of fear and anger on perceived risks of rerrorism: A national field experiment. Psychological Science (Wiley-Blackwell), 14, 144-150. 188 Lerner, J. S., & Keltner, D. (2000). Beyond valence: Toward a model of emotion-specific influences on judgment and choice. Cognition & Emotion, 14, 473-493. Lerner, J. S., & Keltner, D. (2001). Fear, anger, and risk. Journal of Personality and Social Psychology, 81, 146-159. doi:10.1037/0022-3514.81.1.146 Lerner, J. S., & Tiedens, L.Z. (2006). Portrait of the angry decision makers: How appraisal tendencies shape anger‘s influence on cognition. Journal of Behavioral Decision Making, 19, 115-137. Leventhal, H. (1970). Findings and theory in the study of fear communications. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 5, pp. 119-186). New York: Academic Press. Levin, I. P., & Gaeth, G. J. (1988). How consumers are affected by the framing of attribute information before and after consuming the product. Journal of Consumer Research, 15, 374–378. Levin, I. P., Schneider, S. L., & Gaeth, G. J. (1998). All frames are not created equal: A typology and critical analysis of framing effects. Organizational Behavior and Human Decision Processes, 76, 149–188. Lieber, J. B. (2016, May 17). How to make hospitals less deadly. The Wall Street Journal. Retrieved from https://www.wsj.com/articles/how-to-make-hospitals -less-deadly-1463526075 Lingard, L., Espin, S., Whyte, S., Regehr, G., Baker, G. R., Reznick, R. (2004). Communication failures in the operating room: An observational classificaiton of the recurrent types and effects. Quality and Safety in Health Care, 13, 30-334. Lu, J. Y., Xie, X. F., & Zhang, R. (2013). Focusing on appraisals: How and why anger and fear influence driving risk perception. Journal of Safety Research, 45, 65-73. Mackie, D. M., Devos, T., & Smith, E. R. (2000). Intergroup emotions: Explaining offensive action tendencies in an intergroup context. Journal of Personality and Social Psychology, 79, 602-616. doi:10.1037/0022-3514.79.4.602 MacReady, N. (2000). Second stories, sharp ends: Dissecting medical errors. Lancet (London, England), 355(9208), 994. 189 Maio, G. R., & Esses, V. M. (2001). The need for affect: Individual differences in the motivation to approach or avoid emotions. Journal of Personality, 69, 583-615. Major, L. H. (2009). Break it to me harshly: The effects of intersecting news frames in lung cancer and obesity coverage. Journal of Health Communication, 14, 174188. Major, L. H. (2011). The Mediating role of emotions in the relationship between frames and attribution of responsbility for health problems. Journalism & Mass Communication Quarterly, 88, 502-522. Manning, M. L. (2006). Improving clinical communication through structured conversation. Nursing Economics, 24, 268-271. Manstead, A. S. R. & Tetlock, P. E. (1989). Cognitive appraisals and emotional experience: Further evidence. Cognition and Emotion, 3, 225-240. Marcus, M. B. (May 4, 2006). Medical errors now 3rd leading cause of death in U.S., study suggests. Retrieved from https://www.cbsnews.com/news/medical-errors-now-3rd-leading-cause-of-death-in-u-s-study-suggests/ Matthes, J. (2009). What‘s in a frame? A content analysis of media framing studies in the world‘s leading communication journals, 1990-2005. Journalism & Mass Communication Quarterly, 86, 349-367. Malhotra, N., & Kuo, A. G. (2008). Attributing blame: The public‘s response to Hurricane Katrina. Journal of Politics, 70, 120-135. Mayo, A. M., & Duncan, D. (2004). Nurse perceptions of medication errors: what we need to know for patient safety. Journal of Nursing Care Quality, 19, 209-217. McCann, E. (2014, July 18). Deaths by medical mistakes hit records. Healthcare IT News. Retrieved from http://www.healthcareitnews.com/news/deaths-by-medical-mistakes-hitrecords McDonalda, L. M., Sparksb, B., & Glendon, A. I. (2010). Stakeholder reactions to company crisis communication and causes. Public Relations Review, 36, 263-271. McDonald, C. J., Weiner, M., & Hui, S. L. (2000). Deaths due to medical errors are exaggerated in Institute of Medicine report. JAMA, 284, 93-95. 190 Mclean, T. R. (2015). Why the distribution of medical errors matters. The American Journal of Surgery, 210, 188-192. McNeil, B. J., Pauker, S. G., Sox, Jr, H. C., & Tversky, A. (1982). On the elicitation of preferences for alternative therapies. The New England journal of medicine, 306, 1259-1262. Meaney, M. E. (2004). Error reduction, patient safety and institutional ethics committees. Journal of Law, Medicine & Ethics, 32, 358-364. Mebane, F. (2003). Examining the content of health care reporting: Neither the health care system or policies creating it receive coverage they deserve. Nieman Reports, 57(1). Medical errors cause 250,000 U.S. deaths a year. (2016, May 6). Retrieved from http://www.futurity.org/medical-errors-deaths-1157152/ Meier, B. P., & Robinson, M. D. (2004). Does quick to blame mean quick to anger? The role of agreeableness in dissociating blame and anger. Personality and Social Psychology Bulletin, 30, 856-867. Meissner, H. I., Potosky, A. L., & Convissor, R. (1992). How sources of health information relate to knowledge and use of cancer screening exams. Journal of Community Health, 17, 153-165. Meyerowitz, B. E., & Chaiken, S. (1987). The effect of message framing on breast self-examination attitudes, intentions, and behavior. Journal of Personality and Social Psychology, 52, 500–510. Mikula, G., Scherer, K. R., & Athenstaedt, U. (1998). The role of injustice in the elicitation of differential emotional reactions. Personality and Social Psychology Bulletin, 24, 769-783. Millenson, M. L. (2002). Pushing the profession: How the news media turned patient safety into a priority. Quality and Safety in Health Care, 11, 57-63. Morrison, A. L., Beckmann, U., Durie, M., Carless, R., & Gillies, D. M. (2001). The effects of nursing staff inexperience (NSI) on the occurrence of adverse patient experiences in ICUs. Australian Critical Care, 14, 116-121. 191 Moss, J. (2005). Technological system solutions to clinical communication error, The Journal of Nursing Administration, 35, 51-53. Moyer, A., Greener, S., Beauvais, J., & Salovey, P. (1995). Accuracy of health research reported in the popular press: Breast cancer and mammography. Health Communication, 7, 147. Mrayyan, M. T., Shishani, K., & Al-Faouri, I. (2007). Rate, causes and reporting of medication errors in Jordan: nurses' perspectives. Journal of Nursing Management, 15, 659-670. Murphy, J. G., & Dunn, W. F. (2010). Medical errors and poor communication (Editorial). Chest, 138, 1292. Myers, T. A., Nisbet, M. C., Maibach, E. W., & Leiserowitz, A. A. (2012). A public health frame arouses hopeful emotions about climate change. Climate Change, 113, 1105-1112. Nabi, R. L. (1999). A cognitive-functional model for the effects of discrete negative emotions on information processing, attitude change, and recall. Communication Theory, 9, 292-320. Nabi, R. L. (2002a). Anger, fear, uncertainty, and attitudes: A test of the cognitive-functional model. Communication Monographs, 69, 204-216. Nabi, R. L. (2002b). Discrete emotions and persuasion. In J.P. Dillard & M. Pfau (Eds.), The persuasion handbook (pp. 289-308). Thousand Oaks, CA: Sage. Nabi, R. L. (2003). Exploring the framing effects of emotion: Do discrete emotions differentially influence information accessibility, information seeking, and policy preference? Communication Research, 30, 224-247. Nabi, R. L. (2007). Emotion and persuasion: A social-cognitive perspective. In D. R. Roskos-Ewoldsen & J. L. Monahan (Eds.), Communication and social cognition. Theories and methods (pp. 377-398). Mahwah, NJ: Lawrence Erlbaum. Nelson, T. E., Clawson, R. A., & Oxley, Z. M. (1997a). Media framing of a civil liberties conflict and its effect on tolerance. American Political Science Review, 91, 567-583. 192 Nelson, T. E., Oxley, Z. M., & Clawson, R. A. (1997b). Toward a psychology of framing effects. Political Behavior, 19, 221-246. Nelson, T. E., & Oxley, Z. M. (1999). Issue framing effects on belief importance and opinion. The Journal of Politics, 61, 1040-1067. Nerb, J., & Spada, H. (2001). Evaluation of environmental problems: A coherence model of cognition and emotion. Cognition & Emotion, 15, 521-551. doi:10.1080/02699930143000013 Netburn, D. (2016, May 3). Medical errors are No. 3 cause of death in the U.S., after heart disease and cancer. Los Angeles Times. Retrieved from http://www.latimes.com/science/sciencenow/la-sci-sn-medical-errors-20160503 -story.html Neuman, W. R., Just, M. R., & Crigler, A. N. (1992). Common knowledge. News and the construction of political meaning. Chicago: The University of Chicago Press. Riles, J. M., Sangalang, A., Hurley, R. J., & Tewksbury, D. (2015). Framing cancer for online news: Implications for popular perceptions of cancer. Journal of Communication, 65, 1018-1040. doi:10.1111/jcom.12183 O‘Donnell, J. (2016, May 3). Second study says medical errors third-leading cause of death in U.S. USA Today. Retrieved from https://www.usatoday.com/ story/news/ politics/2016/05/03/second-study-says-medical-errors -third-leading-cause-death-us/83874022/ O' Keefe, D. J. (2003). Message properties, mediating states, and manipulation checks: Claims, evidence, and data analysis in experimental persuasive message effects research. Communication Theory, 13, 251-274. O'Keefe, D. J., & Jensen, J. D. (2009). The Relative Persuasiveness of Gain‐Framed and Loss‐Framed Messages for Encouraging Disease Detection Behaviors: A Meta‐Analytic Review. Journal of Communication, 59, 296-316. O‘Keefe, D. J., & Jensen, J. D. (2007). The relative persuasiveness of gain-framed and loss-framed messages for encouraging disease prevention behaviors: A meta-analytic review. Journal of Health Communication, 12, 623-644. 193 Oatley, K., Keltner, D., & Jenkins, J. M. (2006). Understanding emotions. Malden, MA; Oxford: Blackwell Pub. Ofri, D. (2013, May 30). My near miss; meanwhile. The International Herald Tribune. Retrieved from LexisNexis Academic Database. Otieno, C., Spada, H., & Renkl, A. (2013). Effects of news frames on perceived risk, emotions, and learning. Plos ONE, 8, 1-12. Oyebode, F. (2013). Clinical errors and medical negligence. Medical Principal Practice, 22, 323-333. DOI:10.1159/000346296 Ozkan, S., Kocaman, G., Ozturk, C., & Seren, S. (2011). Frequency of pediatric medication administration errors and contributing factors. Journal of Nursing Care Quality, 26, 136-143. Pan, Z., & Kosicki, G. (2005). Framing and the understanding of citizenship. In S. Dunwoody, L. Backer, D. McLeod, & G. Kosicki (Eds.), The evolution of key mass communication concepts. Cresskill, NJ: Hampton Press. Pan, P., & Meng, J. (2016). Media frames across stages of health crisis: A crisis management approach to news coverage of flu pandemic. Journal of Contingencies & Crisis Management, 24, 95-106. doi:10.1111/1468-5973.12105 Perneger, T. V. (2005). The Swiss cheese model of safety incidents: Are there holes in the metaphor? BMC Health Services Research, 5, doi: 10.1186/1472-6963-5-71 Petrova, E. (2010). Nurses' perceptions of medication errors in Malta. Nursing Stand, 24, 4148. Pollard, W. E. (2003). Public perceptions of information sources concerning bioterrorism before and after anthrax attacks: An analysis of national survey data. Journal of Health Communication, 8, 93-103. Potter, W. J., & Riddle, K. (2007). A content analysis of the media effects literature. Journalism & Mass Communication Quarterly, 84, 90-104. Pribble, J. M., Goldstein, K. M., Fowler, E. F., Greenberg, M. J., Noel, S. K., & Howell, J. D. (2006). Medical news for the public to use? What's on local TV news. The American Journal of Managed Care, 12, 170-176. 194 Preacher, K. J. & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator analyses. Behavior Research Methods, 40, 879-891. Price, V., Tewksbury, D., & Powers, E. (1997). Switching trains of thought: The impact of news frames on readers‘ cognitive responses. Communication Research, 24, 481-506. Quigley, B. M., & Tedeschi, J. T. (1996). Automatic vigilance: The attention-grabbing power of negative social information. Journal of Personality and Social Psychology, 61, 380-390. Rabin, R. C. (2013, April 30). Health care‘s ‗Dirty little secret‘: No one may be coordinating care. Kaiser Heath News. Retrieved from http:// kaiserhealthnews.org/news/coordination-of-care/ Reader, T. W., Flin, R., & Cuthbertson, B. H. (2007). Communication skills and error in the intensive care unit. Current Opinion in Critical Care, 13, 732-736. Reason, J. (2000). Human error: Models and management. British Medical Journal, 320, 768–770. Reese, S. D. (2001). Framing public Life: A bridging model for media research. In S. D. Reese, O. H. Gandy, Jr., & A. E. Grant (Eds.), Framing public life: Perspectives on media and our understanding of the social world (pp. 7-31), Mahwah, NJ: Erlbaum. Rice, S. (2016, May 4). Medical errors may kill 250,000 a year, but problem not being tracked. Retrieved from http://www.modernhealthcare.com/ article/20160504/NEWS/160509962 Robinson, A. R., Hohmann, K. B., Rifkin, J. I., Topp, D., Gilroy, C. M., Pickard, J. A., & Anderson, R. J. (2002). Physician and public opinions on quality of health care and the problem of medical errors. Archives of Internal Medicine, 162, 2186-2190. doi:10.1001/archinte.162.19.2186. Rogers, R.W. (1975). A protection motivation theory of fear appeals and attitude change. Journal of Psychology, 91, 93-114. 195 Rogers, A. E., Hwang, W. T., Scott, L. D., Aiken, L. H., & Dinges, D. F. (2004). The working hours of hospital staff nurses and patient safety. Health Affairs, 23, 202–212. Rothman, A. J., Bartels, R. D., Wlaschin, J., & Salovey, P. (2006). The strategic use of gain- and loss-framed messages to promote healthy behavior: How theory can inform practice. Journal of Communication, 56, S202-S220. Roseman, I. J. (1984). Cognitive determinants of emotion: A structural theory. Review of Personality and Social Psychology, 5, 11-36. Roseman, I. J., Spindel, M. S., & Jose, P. E. (1990). Appraisals of emotion-eliciting events: Testing a theory of discrete emotions. Journal of Personality and Social Psychology, 59, 899-915. Roseman, I. J. (2011). Emotional behaviors, emotivational goals, emotion strategies: multiple levels of organization integrate variable and consistent responses. Emotion Review, 3, 434-443. Roseman, I. J., Wiest, C., & Swartz, T. S. (1994). Phenomenology, behaviors, and goals differentiate discrete emotions. Journal of Personality and Social Psychology, 67, 206-221. Shaver, P., Schwartz, J., Kirson, D., & O'Connor, C. (1987). Emotion knowledge: Further exploration of a prototype approach. Journal of Personality and Social Psychology, 52, 1061-1086. Scherer, K. R. (1984). Emotion as a multicomponent process: A model and some cross-cultural data. In P. Shaver (Ed.). Review of Personality and Social Psychology, 5, 37-63. Beverly Hills: Sage Scheufele, D. A. (1999). Framing as a theory of media effects. Journal of Communication, 49, 103-122. Schuck, A. R. T., & de Vreese, C. H. (2006). Framing the EU enlargement: News media content and effects. European Journal of Communication, 21, 5-32 Scheufele, D. A., & Tewksbury, D. (2007). Framing, agenda setting, and priming: The evolution of three media effects models. Journal of Communication, 57, 9-20. doi:10.1111/j.1460-2466.2006.00326.x 196 Semetko, H. A., & Valkenburg, P. M. (2000). Framing European politics: A content analysis of press and television news. Journal of Communication, 50, 93-109. Shanafelt, T. D., Balch, C. M., Dyrbye, L., Bechamps, G., Russell, T., Satele, D. (2011). Special report: suicidal ideation among American surgeons. Archives of Surgery, 146, 54-62. Shen, F., Lee, S. Y., Sipes, C., & Hu, F. (2012). Effects of media framing of obesit among adolescents. Communication Research Reports, 29, 26-33. Shen, L. J. (2005). The interplay of message framing, cognition and affect in persuasive health communication (Doctoral dissertation). Available from ProQuest Dissertations & Theses Global. (305389221). Shen, L., & Dillard, J. P. (2007). The influence of behavioral inhibition/approach systems and message framing on the processing of persuasive health messages. Communication Research, 34, 433-467. Segrin, C., & Dillard, J. (1991). (Non)depressed persons‘ cognitive and affective reactions to (un)successful interpersonal influence. Communication Monographs, 58, 115-134. Slooten, V. E., Friedman, D. B., & Tanner, A. (2013). Are we getting the health information we need from the mass media? An assessment of consumers‘ perceptions of health and medical news. Journal of Consumer Health on the Internet, 17, 35-53. Slothuus, R. (2008). More than weighting cognitive importance: A dual-process model of issue framing effects. Political Psychology, 29, 1-28. Small, D. A., & Lerner, J.S. (2008). Emotional policy: Personal sadness and anger shape judgments about a welfare case. Political Psychology, 29, 149-168. Small, D. A., Lerner, J. S., & Baruch, F. (2006). Emotion priming and attributions for terrorism: Americans' reactions in a national field experiment. Political Psychology, 27, 289-298. Smith, C. A. & Ellsworth, P. C. (1985). Patterns of cognitive appraisals in emotion. Journal of Personality and Social Psychology, 48, 813-838. 197 Smith, C. A., Haynes, K. N., Lazarus, R. S., & Pope, L. K. (1993). In search of the 'hot' cognitions: Attributions, appraisals, and their relation to emotion. Journal of Personality and Social Psychology, 65, 916-929. doi:10.1037/0022-3514.65.5.916 Smith, C. A., & Lazarus, R. S. (1990). Emotion and adaptation. In L. A. Pervin (Ed.), Handbook of personality: Theory and research (pp. 609–637). New York: Guildford Press. Smith, C. A., & Lazarus, R. S. (1993). Appraisal components, core relational themes, and the emotions. Cognition and Emotion, 7, 233–269. Smith, L. K., Tanner, A. H., & Duhe, S. F. (2007). Convergence concerns in local television: Conflicting views from the newsroom. Journal of Broadcasting & Electronic Media, 51, 555-574. Southwick, F. S. (2013, February 19). Losing my leg to a medical error. The New York Times. Retrieved from http://www.nytimes.com/2013/02/20/opinion/losing-my-leg-to-amedical-error.html?_r=0 Sox, H. C. Jr. & Woloshin, S. (2000). How many deaths are due to medical error? Getting the number right. Effective Clinical Practice, 3, 277-228. Spencer, R., Coiera, E., & Logan, P. (2004). Variation in communication loads on clinical staff in the emergency department. Annals of Emergency Medicine, 44, 268-273. Stebbing, C., Kaushal, R., & Bates, D.W. (2006). Pediatric mediation safety and the media: What does the public see? Pediatrics, 117, 1907-1914. Stratton, K. M., Blegen, M. A., Pepper, G., & Vaughn, T. (2004). Reporting of medication errors by pediatric nurses. Journal of Pediatric Nursing, 19, 385-392. Stripe, S. C., Best, L. G., Cole-Harding, S., Fifield, B., & Talebdoost, F. (2006). Aviation model cognitive risk factors applied to medical malpractice cases. Journal of the American Board of Family Medicine, 19, 627-632. Study suggests medical errors now third leading cause of death in the U.S. Physicians advocate for changes in how deaths are reported to better reflect reality. (2016, May 3). Retrieved from http://www.hopkinsmedicine.org/news/media/releases /study_suggests_medical_errors_now_third_leading_cause_of_death_in_the_us 198 Sun, Y., Krakow, M., John, K. K., Liu, M., & Weaver, J. (2016). Framing obesity: How news frames shape attributions and behavioral responses. Journal of Health Communication, 21, 139-147. doi:10.1080/10810730.2015.1039676 Suresh, G. (2006). If it‘s in the paper, it must be true: Newspaper reporting of pediatric medication errors. Pediatrics, 117, 2281-2282. Swift, E. K., Koepke, C. P., Ferrer, J. A., & Miranda, D. (2001). Preventing medical errors: Communicating a role for medicare beneficiaries. Health Care Financing Review, 23, 77. Tabachnic, B. G. & Fidell, L. S. (2007). Using multivariate statistics. Boston: Pearson/Allyn & Bacon Tanner, A., & Friedman, D. B. (2011). Authorship and information sourcing for health news on local TV Web sites: An exploratory analysis. Science Communication, 33, 3–27. Tanner, A., Friedman, D. B., Barr, D., & Koskan, A. (2008). Preparing for disaster: An examination of public health emergency information on local TV web sites. Electronic News, 2, 218-234. Tevlin, R., Doherty, E., & Traynor, O. (2013). Improving disclosure and management of medical error – An opportunity to transform the surgeons of tomorrow. The Surgeon, 11, 338-343. Thomas, E. F., McGarty, C., & Mavor, K. I. (2009). Transforming ―apathy into movement ‖: the role of prosocial emotions in motivating action for social change. Personality and Social Psychology, 13, 310-333. Tibby, S. M., Correa-West, J., Durward, A., Ferguson, L., & Murdoch, I. A. (2004). Adverse events in a paediatric intensive care unit: Relationship to workload, skill mix and staff supervision. Intensive Care Medicine, 30, 1160-1166. doi:10.1007/s00134-004-2256-y Tiedens, L. Z., & Linton, S. (2001). Judgment under emotional certainty and uncertainty: The effects of specific emotions on information processing. Journal of Personality and Social Psychology, 81, 973-988. 199 Utz, S., Schultz, F., & Glocka, S. (2013). Crisis communication online: How medium, crisis type and emotions affected public reactions in the Fukushima Daiichi nuclear disaster. Public Relations Review, 39, 40-46. Van Gorp, B. (2007). The constructionist approach to framing: Bringing culture back in. Journal of Communication, 57, 60-78. Van 'T Riet, J., Ruiter, R. A., Werrij, M. Q., & De Vries, H. (2008). The influence of self-efficacy on the effects of framed health messages. European Journal of Social Psychology, 38, 800-809. doi:10.1002/ejsp.496 Varjavand, N, Nair, S, & Gracely, E. (2012). A call to address the curricular provision of emotional support in the event of medical errors and adverse events. Medical Education, 46, 1149-1151. Wagner, M. (2014). Fear and anger in Great Britain: Blame assignment and emotional reactions to the financial crisis. Political Behavior, 36, 683-703. Wallington, S. F., Blake, K. D., Taylor-Clark, K., & Viswanath, K. (2010). Challenges in covering health disparities in local news media: An exploratory analysis assessing views of journalists. Journal of Community Health, 35, 487-494. Wang, T. (2012). Presentation and impact of market-driven journalism on sensationalism in global TV news. International Communication Gazette, 74, 711-727. doi:10.1177/1748048512459143 Wang, Z., & Gantz, W. (2007). Health content in local television news. Health Communication, 21, 213-221. Waring, J. J. (2005). Beyond blame: Cultural barriers to medical incident reporting. Social Science & Medicine, 60, 1927-1935. Watson, D., Wiese, D., Vaidya, J., & Tellegen, A. (1999). The two general activation systems of affect: Structural findings, evolutionary considerations, and psychophysiological evidence. Journal of Personality and Social Psychology, 76, 820-838. Weber, H. (2004). Explorations in the social construction of anger. Motivation & Emotion, 28, 197-219. 200 Weingart, S. N., Wilson, R. M., Gibberd, R. W., & Harrison, B. (2000). Epidemiology of medical error. BMJ: British Medical Journal, 320, 774-777. Weiss, R. (1999, November 30). Medical errors blamed for many deaths; as many as 98, 000 a year in U.S. linked to mistakes. The Washington Post. Retrieved from LexisNexis Academic Database. Weissman, J. S., Rothschild, J. M., Bendavid, E., Sprivulis, P., Cook, E. F., Evans, R. S. (2007). Hospital workload and adverse events. Medical Care, 45, 448-455. Winterhoff-Spurk, P. (1998). TV news and the cultivation of emotion. Communications, 23, 545-556. Wirth, W., & Schramm, H. (2005). Media and emotions. Communication Research Trends, 24, 3-39. Witte, K. (1992). Putting the fear back into fear appeals: The extended parallel process model. Communication Monographs, 59, 329-349. Woodward, H. I., Lemer, C., & Wu, A.W. (2009). An end to the witch hunts: Responding to the defenders of blame and shame. A commentary on Collins, Block, Arnold, and Christakis. Social Science & Medicine, 69, 1291-1293. Wu, A.W., Cavanaugh, T. A., Mcphee, S. J., Lo, B., & Micco, G. P. (1997). To tell the truth: Ethical and practical issues in disclosing medical mistakes to patients. Journal of General Internal Medicine, 12, 770-775. Wu, M. (2006). Framing AIDS in China: A comparative analysis of US and Chinese wire news coverage of HIV/AIDS in China. Asian Journal of Communication, 16, 251-272. Yan, C. M., & Dillard, J. P. (2010). Emotion inductions cause changes in activation levels of the behavioral inhibition and approach systems. Personality and Individual Differences, 48, 676-680. Yan, C., Dillard, J. P., & Shen, F. (2012). Emotion, motivation, and the persuasive effects of message framing. Journal of Communication, 62, 682-700. doi:10.1111/j.1460- 2466.2012.01655.x Yang, S. U., Kang, M., & Johnson, P. (2010). Effects of narratives, openness to dialogic communication, and credibility on engagement in crisis communication through organizational blogs. Communication Research, 37, 473-497. 201 Yeaton, W. H., Smith, D., & Rogers, K. (1990). Evaluating understanding of popular press reports of health research. Health Education Quarterly, 17, 223-234. Yoo, J. H., & Kim, J. (2012). Obesity in the new media: A content analysis of obesity videos on YouTube. Health Communication, 27, 86-97. doi:10.1080/10410236.2011.569003 Yun, D., Nah, S., & McLeod, D. M. (2008). Framing effects of news coverage of the embryonic stem cell controversy: Issue involvement as an effects moderator. Communication Research Reports, 25, 312-315. doi:10.1080/08824090802440279 Zaller, J. R. (1992). The nature and origins of mass opinion. New York: Cambridge University Press. Zeelenberg, M., Nelissen, R. M. A., Breugelmans, S. M., & Pieters, R. (2008). On emotion specificity in decision making: Why feeling is for doing. Judgment and Decision Making, 3, 18-27.