Shifting perceptions of higher education: analyzing possible drivers of declining confidence in American colleges and universities

This thesis analyzes over 37,000 rows of General Social Survey data from 1974 to 2022 to draw conclusions about what factors are driving the decline in American confidence in higher education. I find that those with stronger conservative views, people with college degrees or parents with college degrees, and White Americans are less likely to possess high confidence in education. High confidence is more likely among those who are satisfied with their financial situations, Black Americans and other nonwhite individuals, and more religious people. Understanding how these and other characteristics influence confidence in education may assist colleges and universities as they navigate demographic changes, political conflicts, and shifting demands of employers.

SHIFTING PERCEPTIONS OF HIGHER EDUCATION: ANALYZING POSSIBLE DRIVERS OF DECLINING &21),'(1&( IN $0(5,&$1 COLLEGES AND UNIVERSITIES by Ella Sjoblom A Senior Honors Thesis Submitted to the Faculty of The University of Utah In Partial Fulfillment of the Requirements for the Honors Degree in Bachelor of Science In Economics Approved: ________ ______ Thomas Maloney, PhD Chair, Department of Economics ________ Gabriel Lozada, PhD Departmental Honors Liaison _____________________________ Monisha Pasupathi, PhD Dean, Honors College Richard Fowles, PhD Thesis Faculty Supervisor April 2025 Copyright © 2025 All Rights Reserved ABSTRACT This thesis analyzes over 37,000 rows of General Social Survey data from 1974 to 2022 to draw conclusions about what factors are driving the decline in American confidence in higher education. I find that those with stronger conservative views, people with college degrees or parents with college degrees, and White Americans are less likely to possess high confidence in education. High confidence is more likely among those who are satisfied with their financial situations, Black Americans and other nonwhite individuals, and more religious people. Understanding how these and other characteristics influence confidence in education may assist colleges and universities as they navigate demographic changes, political conflicts, and shifting demands of employers. ii TABLE OF CONTENTS ABSTRACT ii INTRODUCTION 1 METHODOLOGY 9 RESULTS 18 DISCUSSION 23 CONCLUSION 27 REFERENCES 28 iii Introduction As college tuition prices and student debt levels rise and ideological battles play out between universities and the government, Americans are rethinking what higher education has to offer. A college degree was once viewed as a clear pathway to economic mobility, desirable for anyone with the aptitude and means to pursue higher education. Many Americans still share these sentiments, though over the past fifty years, confidence in American higher education has eroded (Miner, 2020). Despite demonstrated positive effects on individual earnings and national GDP growth, perceptions of colleges and universities are more informed by individual experiences and political ideology (Li et. al, 2024; U.S. Bureau of Labor Statistics, 2024). This thesis seeks to explore the factors that lead individuals to be less confident in higher education and consider trends among demographic groups. Literature Review Defining Institutional Confidence To lay the groundwork for understanding American confidence in education, it is necessary to define the concepts of trust and confidence. Previous research offers multiple definitions of these concepts and their relationships with each other. Some suggest that trust and confidence are synonymous with one another, with trust as a measurement of confidence (Rousseau et al., 1998). Others suggest that trust and confidence are separate concepts and should be measured as such (Zmerli & Newton, 2008). Yamagishi and Yamagishi (1994) posit that confidence is “the expectation of competence,” while trust is the “expectation of goodwill and benign intent.” These 1 insights assist in understanding how survey respondents may perceive questions about institutional confidence, allowing for better analyses of results. Trends in American Institutional Confidence Brady and Kent (2022) state that institutions must be viewed as legitimate in one or all of four areas to obtain public confidence: regulatory authority, adherence to accepted logic and practices, efficiency and performance, and morality. These categories may apply to certain kinds of institutions more than others. For instance, legal and medical institutions may be trusted due to their association with morality, while institutions like corporations and banks gain trust based on efficiency and performance (Brady & Kent, 2022). Analyses of survey data from the most robust American public opinion surveys including the General Social Survey (GSS), Harris poll, and Gallup poll all record the decline of public confidence in the nation’s institutions (Miner, 2020; Pew Research Center, 2024; Brady & Kent, 2022). Early research on these patterns began in the 20th century with studies on trust in government and business (Lipset & Schneider, 1983). Institutional trust research expanded into consistent polling on institutional confidence across branches of government, news and media organizations, education institutions, and various private industries. Confidence levels in institutions are closely related to political ideology and how each institution fits into political party agendas, especially the party in power (Gershtenson et. al, 2006). This context informs further research into specific American institutions, including colleges and universities. 2 Education institutions, particularly colleges and universities, are generally expected to excel in three of the four categories described by Brady and Kent (2022). The public expects higher education institutions to deliver credentials and skills to improve students’ future performance in the workforce, produce quality research using logical methodologies, and uphold ethical standards in research procedures and student admissions processes. Colleges and universities are uniquely positioned to serve many roles in society and thus face many expectations in order to retain public confidence. They must secure the trust of their students, research funders, and the public in order to operate and carry out their academic missions. With less public confidence in institutions including higher education, it is in their best interest to understand the factors driving this trend. College Costs and Student Borrowing Trends Cost is a crucial component in an individual’s choice whether to pursue a college education. A 2022 survey from the University of Chicago found that most Americans believe that people are deciding not to attend college because they either cannot afford the cost or must work immediately after high school (National Opinion Research Center, 2022). High “sticker prices,” or the price of college tuition before aid and tax benefits, concern many potential students during the college application process. However, while sticker prices have increased by 114% since 1993, the actual net price of attendance has not changed after adjusting for inflation (Looney, 2024). Still, student borrowing rates tripled over the same time frame (Looney, 2024). The availability of federal student loan options, both subsidized and unsubsidized, allows students to borrow money regardless of means. These loans may be used for tuition and fees as well as living expenses. Less 3 than 30 percent of undergraduate students borrowed in 1996 and 2000, but the share of borrowers surged and peaked at over 40 percent of undergraduates in 2012 (Looney & Yannelis, 2024). Perceptions of the cost and payment structures in the American higher education system may play a role in declining confidence in the institution. Additionally, family wealth may impact college choice, graduation rates, and quality of experience with higher education. Students with higher family incomes may have more support in staying on track through their college programs (Pfeffer, 2018). Historical data shows increasing gaps between income groups in four-year college graduation rates among some generational cohorts (Ziol-Guest & Lee, 2016). These patterns may or may not directly translate into higher confidence in education, but they likely influence students’ perceptions of their college experiences. Higher Education and Economic Mobility Despite the rise in student borrowing, having a college degree still increases median weekly earnings and decreases unemployment compared to those who do not possess degrees (U.S. Bureau of Labor Statistics, 2024). Figure 1 below displays a comparison of weekly earnings by educational attainment compiled by the U.S. Bureau of Labor Statistics in 2024. Figure 1 (U.S. Bureau of Labor Statistics, 2024) 4 While increases in wages based on educational attainment are easy to observe, patterns across socioeconomic groups are less clear. Socioeconomic class and family wealth are strong predictors of college enrollment (Glater, 2016; Pfeffer, 2018). Further, Chetty et. al (2013) found that rates of upward economic mobility vary between high and low-income families. Overall economic mobility rates were highest at mid-tier public universities, with mobility from the bottom income tier to the top one percent more likely at elite colleges (Chetty et al., 2013). Still, the fraction of low-income students at mid-tier public universities associated with high economic mobility rates fell between 2000 and 2011 (Chetty et. al, 2013). The distribution of student debt is also disparate between racial and socioeconomic groups, with higher default rates on loans at institutions where more low-income and marginalized students attend (Glater, 2016). Examining the magnitude of economic mobility and potential burden of debt related to postsecondary education adds more context to wage data alone. 5 Race Much prior research explores the relationship between educational attainment and race. High school graduation rates increased across all racial groups in the U.S. between 2010 and 2022, and the high school completion gap narrowed between White and Black young adults to no measurable difference between the two groups (National Center for Education Statistics, 2023b). The percentage of young adults 25 through 29 with a bachelor's degree also increased across groups in the same time period, but the gap in attainment between White and Black young adults was 17 percent; similarly, the gap between White and Hispanic young adults stood at 20 percent (National Center for Education Statistics, 2023b). The difference in participation in higher education limits access to the benefits of a college education, such as wage increases and expanded networks. Policy actions by the federal government and institutions themselves have aimed to increase access to their programs across racial groups, expanding diversity among college student populations. In the 1990s and 2000s, many higher education institutions adopted race-conscious admissions policies to diversify their student populations and attempt to correct racial disparities due to past social and legal barriers to education (Glater, 2016). These policies generated controversy, and in 2019, half of American adults disproved of them, compared to a third who approved (Gramlich, 2023). Kravitz and Platania (1993) discovered that opposition to various affirmative action plan alternatives was directly correlated with the weight given to demographics in the analysis. Confidence in American higher education institutions to provide quality results and fair admissions wavered among White Americans, causing the legal battles that led to 6 the Supreme Court declaring affirmative action policies in college admissions unconstitutional in 2023 (Students for Fair Admissions Inc. v. President and Fellows of Harvard College, 2023). The rhetoric and media attention over the years surrounding this issue may impact confidence levels across racial groups in the U.S. Sex More American women aged 25 through 34 possess bachelor’s degrees than men in the same age category in every racial group (Hurst, 2024). This difference has existed since 1996, increasing to a 10-point gap between men and women in bachelor’s degree attainment in 2024 (Hurst, 2024). Lundberg (2020) discovered that much of this gap can be attributed to differing educational aspirations between girls and boys. Likewise, over 30 percent of men reported that they did not pursue a four-year degree because they “just didn’t want to” (Parker, 2021). Social expectations and the differences in career aspirations between men and women contribute to educational ambitions in each group. Miner (2020) found that American women were slightly more likely to possess high confidence in education compared to men. Political Views Conservative ideology is an established predictor of lower trust in education (Parker, 2019; Miner, 2020). Higher education institutions often receive critiques from the American right due to the dominance of liberal ideas on college campuses. A majority of college faculty identify as Democrats, which is not representative of the broader public (Cholbi, 2014). Republicans are more likely than Democrats to say that colleges have a negative impact on the U.S. (Parker, 2019). Both Gallup and Pew Research Center report 7 that the growing share of Americans with negative views of colleges and universities has mostly come from a loss of conservative confidence (Jones, 2024; Parker, 2019). This established pattern of distrust in higher education from conservative Americans is also apparent in analyses of the General Social Survey (Miner, 2020). Importance of Understanding Confidence in Higher Education Declining confidence in higher education is just one risk that American colleges and universities face as they adapt to meet the changing demands of students, employers, and governments. For years, American higher education institutions have anticipated an “enrollment cliff” in 2026 due to changing demographics (Campion, 2020). Because of the 2008 recession, there will be fewer 18-year-old prospective college students in 2026, which will later affect the size of the potential employee pool for firms in 2030 (Campion, 2020). College enrollment rates also peaked in 2011, with 42 percent of 18- to 24-year-olds enrolled in postsecondary education programs. Figure 2 shows enrollment rates among this age cohort since 1974. The COVID-19 pandemic impacted enrollment rates in 2020 and 2021, which accounts for the large dip at that time. If enrollment rates are already trending lower regardless of the decline of the college-aged population, colleges and universities will need to obtain more interest among this population to stay afloat. Figure 2 (National Center for Education Statistics, 2023a). 8 Methodology This paper utilizes data from the General Social Survey (GSS), one of the most prominent public opinion surveys of Americans since 1972, orchestrated by the National Opinion Research Center (NORC) at the University of Chicago. While aspects of the survey have changed over time, the GSS still retains the same core questions on demographics, attitudes, and behaviors of respondents. Specifically, the GSS has asked respondents about their confidence in a variety of American institutions since the 1970s including education. Among other institutions included in the segment are science, Congress, and the nation’s court system. Using the institutional confidence variable for education and several demographic and attitudinal questions, I develop two models to examine confidence in higher education over time. Dependent Variable For my dependent variable, I selected the education question of the institutional confidence segment of the GSS. This section of the survey highlights 13 American 9 institutions and asks respondents whether they possess “a great deal of confidence, only some confidence, or hardly any confidence” in the people running said institutions. The question wording specifically asks about the people involved with institutions, which may affect responses compared to if they had been asked about the institutions themselves. These questions have appeared in every GSS questionnaire since 1973, apart from 1985, which I omitted in my data set for analysis due to its absence that year. Otherwise, the question is present for respondents every year the survey was conducted. Furthermore, the question about education in this segment does not distinguish between kindergarten through twelfth grade and higher education. As a result, responses to this question may encapsulate a respondent’s sentiment towards one or both of those institutions. Since recent studies indicate nearly identical trends in average confidence in K-12 education as well as higher education (see Figure 3), it seems appropriate to utilize this variable as a proxy for confidence in either institution (Pew Research Center, 2024). Since my research focuses on confidence in higher education and its implications for colleges and universities, from this point forward the word “education” can refer to either or both higher education and education in general throughout the remainder of this paper. It also is not possible to draw conclusions about confidence levels in different kinds of higher education institutions, such as private versus public colleges, or elite universities compared to other schools. Figure 3 (Pew Research Center, 2024) 10 When building my regression models, I coded responses differently depending on the regression type. For the ordinary least-squares model, a “1” value corresponds to having hardly any confidence in education, a “2” to only some confidence, and a “3” to great deal of confidence. This method transforms the dependent categorical variable to a continuous variable. When interpreting the results, it is essential to remember that the original variable was ordinal and discrete. For example, the true ideological difference between “only some confidence” and “hardly any confidence” may not necessarily be the same as the distance between “a great deal of confidence” and “only some confidence,” despite the variable being scaled that way for this model. For the mixed effects logistic regression model, I combined the “only some confidence” and “hardly any confidence” categories into one. Thus, the model returns the log odds of whether an individual possesses “a great deal of confidence.” This model 11 builds upon methodology from previous research using this data (Miner, 2020). The combination of both linear and logistic approaches provides multiple investigations into confidence in education and its relationship to Americans’ demographic and behavioral characteristics. Independent Variables This section describes how each independent variable is captured by the GSS to provide context for interpreting the results. I selected independent variables based on previous research and current ongoing conversations about the value of higher education. While not comprehensive, these variables represent inquiries into characteristics that may affect the decline of American confidence in higher education. For the linear regression, each variable retains its unit measure defined in this section. In the mixed effects model, numeric independent variables were scaled to better compare their effects on the odds of having high confidence in education. Scaling did not apply to categorical variables in the mixed effects binary model. Individual and Parental College Attendance The GSS asks respondents their education status in terms of both years of schooling and highest level of educational attainment. I selected the educational attainment variable, as I was more concerned with whether respondents had attended college versus the linear effect of years of education on confidence in education. Since some individuals finish education programs in shorter or longer timelines than average, the total years of education is less relevant compared to whether they received a degree. I combined education levels “High school” and “Less than high school” into the “No 12 college” category, while respondents with a bachelor’s degree, graduate degree, or an associate or junior college degree were placed into the “College” category. Similarly, the GSS asks respondents about their parents’ level of education using the same question structures as previously described. The survey separates the individual’s mother and father into two different questions. I used the same coding structure and created a new variable that states whether the respondent had at least one parent who attended college in order to estimate the effect that parents’ education levels has on an individual’s perception of education. Sex The composition of the college educated population by sex has changed over time. In 2024, 47 percent of American women between ages 25 and 34 possessed a bachelor’s degree, compared to 37 percent of American men (Hurst, 2024). With a significant difference between men and women’s college attendance over time, it is essential to explore the relationship between sex and confidence in education. Figure 4 shows the proportion of GSS respondents with high confidence in education by year between 1974 and 2022 sorted by sex. Including the entire data set, the percentage of women with high confidence in education is one percent higher than the percentage of men. Figure 4 (General Social Survey, 2024). 13 Age The GSS age variable is straightforward, a continuous variable calculated using the respondent’s reported date of birth. The maximum age in the data set is 89, as the GSS does not record the precise age of respondents 89 or older. The mean age of respondents in this data set is 46.7. GSS samples do not include children, so this analysis is limited to the effect of age of adults over 18 years old on confidence in education. Political Views Given the pattern of higher distrust in higher education among conservatives established in prior research, political views are of great importance to these models (Jones, 2024; Miner, 2020; Parker, 2019). The GSS records political views of respondents 14 on a one through seven scale, with one being “extremely liberal” and seven being “extremely conservative.” Four, the middle of the scale, describes respondents whose views are “moderate” or “middle of the road.” Thus, when interpreting model results, an increase on the political view scale is equivalent to an increase in conservatism, while a decrease indicates an increase in liberalism. Figure 5 illustrates the percentage of high confidence in education among ideological groups over time, including liberals, conservatives, and moderates. This variable is treated as numeric in the model. Figure 5 (General Social Survey, 2024). Religiosity 15 The religiosity variable is based upon a GSS question that measures religious service attendance. A zero corresponds to “never” attending religious services, while an eight means that the respondent attends religious services “several times a week.” Like political ideology, self-reported religiosity is treated as a numeric scale variable for regression purposes. The religiosity variable does not account for respondents’ particular religious identities or denominations and measures only one component of religiosity. These considerations should be taken into account when interpreting results. Race Considering past policies and attempts by American higher education institutions and governments to expand access across racial groups, investigating race as a driver of confidence is of great interest. Until 2000, the GSS recorded respondents’ races by appearance during interviews, placing respondents as “White,” “Black,” or “Other” race categories. Beyond 2000, the survey adopted a self-identifying race question, allowing respondents to list up to three races. Because this research uses responses from 1974 through 2022, it was not possible to use the updated race variable without introducing inconsistencies into statistical models. This unfortunate limitation of the GSS data prevents a deeper analysis of confidence in education among racial groups beyond the three previously listed categories. Income and Financial Situation The income variable used in these models is respondent household income in thousands of dollars. When recording income information, the GSS asks respondents to place their income within a range (for example, $10,000-$14,999) for ease of response. 16 Then, their income in dollars is imputed using midpoint averaging for each range (Hout, 2004). Available income categories have shifted over time due to inflation, which is why it is more sensible to use the imputed income variable for this analysis rather than income categories. Along with income, this research explores how an individual’s satisfaction with their financial situation impacts their confidence in education. The GSS asks respondents to rate their level of satisfaction as “not satisfied at all,” “more or less satisfied,” or “pretty well satisfied.” In the model, these responses are coded as one, two, or three as a numeric variable on a one through three scale. Like the dependent variable, it is essential to consider the original question phrasing and ordinal nature of the one through three scale when interpreting results. Time This selection includes data from 1974 through 2022, without 1985 since the confidence in education question was not asked that year. Additionally, the GSS was conducted every year until 2000, when it switched to every other year. The 2020 GSS was delayed to 2021 due to the pandemic, and the survey resumed its regular schedule in 2022. There were also adjustments in survey administration and methodology as a result of pandemic precautions, so examining trends is less reliable between 2021 data and previous years (National Opinion Resource Center, 2021). Because the time frames between samples differ, it is not possible to observe yearly shifts in confidence in education. In the models, time is broken into six periods, which group survey years by decade. The year the survey was administered is used as a random effect in the mixed effects model, generating a random intercept to account for variation across 31 survey 17 years. Table 1 below shows the proportion of respondents in each categorical variable, plus means and standard deviations for the continuous variables. Table 1: Descriptive Statistics for Independent Variables All Years 1970s N= N= 37,695 6,142 1980s N= 8,512 1990s N= 6,942 2000s N= 5,528 2010s N= 6,694 2020s N= 3,877 College Degree 32% 18% 23% 32% 37% 39% 52% Female 54% 54% 54% 55% 54% 54% 55% Black 13% 9.3% 15% 12% 13% 15% 12% Other Race 5.5% 0.6% 2.3% 4.7% 8.7% 9.6% 9.7% Parent Degree 22% 10% 15% 22% 27% 30% 36% Age 45.47 43.00 43.57 44.20 45.81 47.91 51.12 (17.01) (16.87) (17.02) (16.23) (16.56) (17.13) (17.18) Income (thousands) 46.65 41.77 41.68 46.23 52.06 49.35 53.71 (37.50) (29.90) (31.02) (34.67) (43.41) (42.79) (44.13) Political Ideology 4.07 (1.39) 4.00 (1.30) 4.09 (1.32) 4.15 (1.36) 4.13 (1.41) 4.04 (1.47) 3.97 (1.54) Religiosity 3.65 (2.71) 3.96 (2.65) 4.01 (2.62) 3.79 (2.66) 3.59 (2.72) 3.32 (2.79) 2.75 (2.70) Financial Satisfaction 2.03 (0.75) 2.09 (0.74) 2.01 (0.75) 2.01 (0.74) 2.04 (0.75) 2.01 (0.75) 2.03 (0.76) Results The results of these two models illustrate some surprising associations between certain characteristics and lower confidence in education. Consistent with the findings of 18 Miner (2020), both models reveal that possessing a college degree is a significant predictor of lower levels of confidence in education. This result is counterintuitive to what many would expect, given the known association between a college education and higher income levels (U.S. Bureau of Labor Statistics, 2024). Similarly, having at least one parent with a college degree is associated with lower confidence in education, though it is a less significant predictor than being college educated. The proportion of respondents in this data set with at least one college-educated parent or with a degree themselves increases over time, which could influence this result. Still, it is interesting that those with no direct experience with higher education may trust it more compared to those who have experienced it. In these models, sex is not statistically significant compared to other predictors. Considering the growing gender gap in postsecondary educational attainment between men and women, this is an interesting result. Miner (2020) utilized GSS data through 2018 to study confidence in education, and women were slightly more likely to have high confidence than men in that analysis. However, with the inclusion of the 2021 and 2022 GSS rounds and other variables, the results shifted in these models. This shift in the GSS data could signal that the overall decline of confidence in education appears in women and men more equally now than in the past. According to these regressions, an increase in one’s household income does not predict an increase in confidence in education. In fact, it may predict lower trust. However, an increase in satisfaction with one’s personal financial situation is associated with a higher likelihood of confidence in education. From this result, it seems that one’s perception of their situation matters more than the situation itself. The interaction of 19 income and financial satisfaction reveals that an increase in income has a more positive association with trust in education in the mixed effects logistic regression. The combined effect of college degree status and financial satisfaction was not statistically significant in either model. Aligning with expectations based on past research, being more conservative on the political spectrum predicts a lower likelihood of high confidence in education. This result also implies the reverse, which is that more liberal individuals are more likely to possess high confidence. More religious service attendance is associated with higher confidence in education. Belonging to a race other than white was also a significant predictor of higher confidence. This result cannot account for nuance in race beyond the racial categories of “White,” “Black,” and “Other” due to GSS survey methodology. Finally, an increase in respondent age is associated with lower confidence in education, a statistically significant result in the linear model but not in the mixed effects logistic regression. Time has a statistically significant negative effect on confidence in education in both models. Since the proportion of college-educated adults in the U.S. has increased over the time periods this research highlights, this is an important finding (Glater, 2016). Table 2: Linear Model Results Confidence in Education Predictors Estimates CI p (Intercept) 2.26 2.21 – 2.31 <0.001 20 Age -0.00 -0.00 – -0.00 <0.001 College Degree -0.07 -0.13 – -0.02 0.009 Income -0.00 -0.00 – -0.00 0.004 Female -0.00 -0.02 – 0.01 0.869 Black 0.13 0.10 – 0.15 <0.001 Other Race 0.14 0.10 – 0.17 <0.001 Political Ideology -0.03 -0.04 – -0.03 <0.001 Parent Degree -0.03 -0.05 – -0.01 0.005 Religiosity 0.02 0.01 – 0.02 <0.001 Financial Satisfaction 0.09 0.08 – 0.11 <0.001 Time -0.04 -0.05 – -0.04 <0.001 Income × Financial Satisfaction -0.00 -0.00 – 0.00 0.341 College Degree × Income 0.00 0.00 – 0.00 0.002 College Degree × Financial Satisfaction 0.00 -0.02 – 0.03 0.709 Observations 37695 R2 / R2 adjusted 0.041 / 0.041 Table 3: Mixed Effects Logistic Regression Confidence in Education 21 Predictors Odds Ratios (Intercept) 0.55 0.47 – 0.64 <0.001 Age 0.99 0.97 – 1.02 College Degree 0.85 0.81 – 0.91 <0.001 Income 0.77 0.74 – 0.80 <0.001 Female 0.96 0.92 – 1.01 Black 1.53 1.43 – 1.63 <0.001 Other Race 1.57 1.42 – 1.73 <0.001 Political Ideology 0.94 0.92 – 0.96 <0.001 Parent Degree 0.93 0.87 – 0.99 Religiosity 1.14 1.12 – 1.17 <0.001 Financial Satisfaction 1.22 1.19 – 1.26 <0.001 Time 0.88 0.84 – 0.92 <0.001 Income × Financial Satisfaction 1.03 1.00 – 1.06 0.027 College Degree × Income 1.11 1.04 – 1.17 0.001 College Degree × Financial Satisfaction 1.03 0.97 – 1.09 0.343 Random Effects σ2 3.29 22 CI p 0.654 0.092 0.025 τ00 year 0.03 ICC 0.01 N year 31 Observations 37695 Marginal R2 / Conditional R2 0.054 / 0.064 Discussion The results of my regressions offer several interesting points for discussion. The fact that an increase in household income appears to have a negative association with high confidence in education seems peculiar given the established effect of higher education on income (U.S. Bureau of Labor Statistics, 2024). Financial satisfaction and a higher household income combined may predict higher confidence in education, but higher income alone does not. There are a few potential explanations for these relationships. First, these regressions do not account for the city or region where respondents live. A respondent’s income may be high compared to others in the data set, but low or average based on cost of living in their location. Second, this result parallels sentiment patterns that mismatch macroeconomic indicators since the COVID-19 pandemic (Cummings et. al, 2024). Between 2020 and 2024, there was immense growth in the stock market, and wage increases outpaced price increases since the end of 2019 despite the inflationary period of 2021 through 2023 (Cummings et. al, 2024). American monthly consumer sentiment dipped to lower levels in 2022 than during the Great Recession of 2008 (University of Michigan Survey of Consumers, 2025). Since this data ends with GSS responses from 2022, perhaps low sentiment towards the economy bled into sentiment towards higher education and its expected effects on economic status. 23 These parallels suggest an avenue for further research, comparing trust in higher education over time to general feelings about the economy over time. Finally, the GSS income variable is imperfect due to the structure of the survey question and the difficulty of reporting income in real time during a survey, which certainly affects the reliability of income as a regression variable. Miner (2020) also used GSS data to explore the pattern of declining confidence in education among college-educated individuals. These regressions also demonstrate the same pattern among the children of college-educated individuals. Based on both analyses, individuals with college degrees are distrustful in the system as it stands. More research exploring these individuals and their experiences would help institutions understand why people are dissatisfied with higher education. The obvious explanation for this trend would be the tripling in student borrowing since 1993 and increasing sticker prices for college tuition (Looney, 2024). However, much of this borrowing is concentrated among graduate students, and net tuition prices have not increased since the 1990s due to nearly equivalent increases in financial aid and student tax benefits not including loans (Looney, 2024). Graduate students pursuing professional degrees, such as law and medical students, may borrow more as they anticipate higher future earnings. Understanding how perceptions of professional programs differ from undergraduate and other graduate programs could be an area to further investigate. Additionally, a Gallup poll found that Americans possess more trust in community colleges and two-year schools compared to four-year schools (Marken, 2024). 58 percent of Americans said they have high confidence in two-year schools “providing an affordable education” compared to 11 percent who believe the same about four-year schools (Marken, 2024). Understanding 24 public opinion surrounding college affordability and providing more transparency about available financial aid will be essential for higher education institutions going forward, especially with upcoming declines in the college-aged population (Campion, 2020). The models also highlight higher confidence in education among nonwhite Americans. However, the end range of this data was collected prior to the Supreme Court declaring affirmative action policies in college admissions unconstitutional in 2023 (Students for Fair Admissions Inc. V. Presidents and Fellows of Harvard College, 2023). A New York Times analysis of National Center for Education Statistics data showed an average one percent decrease in the shares of Black and Hispanic students in incoming classes at selective American universities the year after the ruling (Bhatia et al., 2025). Most initial analyses of the decision’s effects appear to focus on elite college enrollment, which is only one segment of the U.S. higher education system. It remains to be seen whether this decision causes additional decreases in enrollment or further declines in confidence in higher education among nonwhite Americans. Survey results from future years could reveal whether affirmative action in college admissions was a driver of decline in confidence in higher education among those groups. Finally, the sudden revocation of hundreds of international student visas on American college campuses around the country under the second Trump administration could have further effects on confidence in higher education in nonwhite racial groups (Ma et. al, 2025). The lack of due process for these students and the inability of most higher education institutions to help them creates tension on college campuses, which could diminish trust in those institutions going forward. 25 Current events also call into question whether this model’s results regarding political ideology and confidence in education will be unchanged going forward. Currently, political ideology is a statistically significant predictor of confidence, with conservative views predicting lower confidence and liberal views predicting higher confidence. However, with a new Republican administration attempting to disband the U.S. Department of Education and cut federal funding for university research, there could be an increase in liberal distrust in education institutions as well (Exec. Order 14242, 2025; U.S. Department of Education, 2025). The Trump administration is using federal funding as leverage to compel universities like Columbia and Harvard to comply with their demands regarding student protesters and ideological agendas. As elite private colleges with large endowments, these institutions have some resources to push back against the federal government, but losing federal funding would be a significant blow to their research budgets. For example, about 46 percent of research funding at Harvard’s School of Public Health comes from federal funds, and the Trump administration froze over $2 billion in federal funding after the university refused to comply with their demands (Powell, 2025). Columbia University yielded to the Trump administration in March 2025, agreeing to discipline student protesters and order a mask ban on campus to recover $400 million in federal funding (Columbia University, 2025; U.S. Department of Education, 2025). At the state level, many Republican legislatures and governors seek to reform public colleges through defunding programs they view as unproductive and eliminating diversity initiatives (Mazzei, 2023; H.B. 265, 2025). State appropriations to colleges have fallen since 2001, and this pattern may continue going forward (Archibald & Feldman, 2018). As these conflicts shock American colleges and universities, trust in 26 education could decline among those who disagree with government actions and lose faith in universities’ independence from partisan political whims. Conclusion Continuing to monitor public confidence in education will reveal whether the patterns observed in my research hold true in the future or transform due to events that occurred since the 2022 GSS was administered. Declining confidence in education is already a risk for enrollment numbers at American higher education institutions throughout the country. The results of ongoing debates involving colleges and universities may have widespread implications for public confidence in them. To prepare for what may be ahead, institutions must prioritize open communication with current and prospective students, monitor policy changes at both the state and federal levels, and analyze risks when developing budgets and strategic plans. 27 References Archibald, R., & Feldman, D. (2018). Drivers of the Rising Price of a College Education. Midwestern Higher Education Compact. https://files.eric.ed.gov/fulltext/ED588510.pdf Bhatia, A., Blatt, B., Paris, F., Parlapiano, A., & Washington, E. (2025, January 15). What Happened at Top Colleges After Affirmative Action Ended. The New York Times. https://www.nytimes.com/interactive/2025/01/15/upshot/college-enrollmentrace.html Brady, H.E., & Kent, T.B. (2022). Fifty Years of Declining Confidence & Increasing Polarization in Trust in American Institutions. Daedalus, 151(4), 43-66. https://doi.org/10.1162/daed_a_01943 Brooks, M.E., Kristensen, K., van Benthem, K.J., Magnusson, A., Berg, C.W., Nielsen, A., Skaug, H.J., Maechler, M., & Bolker, B.M. (2017). “glmmTMB Balances Speed and Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling.” The R Journal, 9(2), 378–400. doi:10.32614/RJ-2017-066. Campion, L.L. (2020). Leading Through the Enrollment Cliff of 2026. TechTrends, 64(3), 542-544. https://www.proquest.com/docview/2407554567/citation/5E73F905241F4545PQ/ 1 Chetty, R., Friedman, J.N., Saez, E., Turner, N., & Yagan, D. (2013). Mobility Report Cards: The Role of Colleges in Intergenerational Mobility (Working Paper No. 28 23618). National Bureau of Economic Research. https://www.nber.org/papers/w23618 Cholbi, M. (2014). Anti-Conservative Bias in Education is Real—But Not Unjust. Social Philosophy and Policy, 31(1), 176-203. https://doi.org/10.1017/S0265052514000181 Columbia University (2025, March 21). Advancing Our Work to Combat Discrimination, Harassment, and Antisemitism at Columbia. https://president.columbia.edu/sites/default/files/content/03.21.2025%20Columbia %20-%20FINAL.pdf Cummings, R., Harris, B., & Mahoney, N. (2024). The paradox between the macroeconomy and household sentiment. The Brookings Institution. https://www.brookings.edu/articles/the-paradox-between-the-macroeconomy-andhousehold-sentiment/ Davern, M., Bautista, R., Freese, J., Herd, P., & Morgan, S.L. (2024). General Social Survey, 1972-2022 cumulative file. [Machine-readable data file]. 1 datafile and 1 codebook (2022 Release 3a). Chicago, IL: National Opinion Research Center. https://gss.norc.org/us/en/gss/get-the-data.html Principal Investigator, Michael Davern; Co-Principal Investigators, Rene Bautista, Jeremy Freese, Pamela Herd, and Stephen L. Morgan. NORC ed. Chicago, 2024. Exec. Order 14242, 3 C.F.R., 13679-13680 (2025). https://www.govinfo.gov/content/pkg/FR-2025-03-25/pdf/2025-05213.pdf 29 Gershtenson, J., Ladewig, J., & Plane, D. L. (2006). Parties, Institutional Control, and Trust in Government. Social Science Quarterly, 87(4), 882–902. http://www.jstor.org/stable/42956164 Glater, J.D. (2016). Debt, Merit, and Equity in Higher Education Access. Law and Contemporary Problems, 79(3), 89-113. https://www.jstor.org/stable/43920684 Gramlich, J. (2023). Americans and affirmative action: How the public sees the consideration of race in college admissions, hiring. Pew Research Center. https://www.pewresearch.org/short-reads/2023/06/16/americans-and-affirmativeaction-how-the-public-sees-the-consideration-of-race-in-college-admissionshiring/ H.B. 265, 66th Legislature, 2025 Reg. Sess. (Utah 2025) https://le.utah.gov/~2025/bills/static/HB0265.html Hout, M. (2004). Getting the Most Out of the GSS Income Measures. University of California Berkeley Survey Research Center. https://gss.norc.org/content/dam/gss/getdocumentation/pdf/reports/methodologicalreports/MR101%20Getting%20the%20Most%20Out%20of%20the%20GSS%20I ncome%20Measures.pdf Hurst, K. (2024). U.S. women are outpacing men in college completion, including in every racial and ethnic group. Pew Research Center. https://www.pewresearch.org/short-reads/2024/11/18/us-women-are-outpacingmen-in-college-completion-including-in-every-major-racial-and-ethnic-group/ 30 Jones, J.M. (2024). U.S. Confidence in Higher Education Now Closely Divided. Gallup. https://news.gallup.com/poll/646880/confidence-higher-education-closelydivided.aspx Kravitz, D.A., & Platania, J. (1993). Attitudes and Beliefs About Affirmative Action: Effects of Target and of Respondent Sex and Ethnicity. Journal of Applied Psychology, 78(6). https://doi.org/10.1037/0021-9010.78.6.928 Li, J. Xue, E., Wei, Y., & He, Y. (2024). How popularizing higher education affects economic growth and poverty alleviation: empirical evidence from 38 countries. Humanities and Social Sciences Communications, 11(520), https://doi.org/10.1057/s41599-024-03013-5 Lipset, S. M., & Schneider, W. (1983). The Decline of Confidence in American Institutions. Political Science Quarterly, 98(3), 379–402. https://doi.org/10.2307/2150494 Looney, A. (2024). How Much Does College Cost and How Does it Relate to Student Borrowing? Tuition Growth and Borrowing Over the Past 30 Years. National Tax Journal, 77(3). https://download.ssrn.com/frb_philadelphia/fedpwp98755.pdf Looney, A., & Yannelis, C. (2024). What Went Wrong with Federal Student Loans? (Working Paper No. 34369). National Bureau of Economic Research. https://www.nber.org/papers/w32469 Lüdecke, D (2024). sjPlot: Data Visualization for Statistics in Social Science. R package version 2.8.17, https://CRAN.R-project.org/package=sjPlot. 31 Lundberg, S. (2020). Educational Gender Gaps. Southern Economic Journal, 87(2). 416439. https://pmc.ncbi.nlm.nih.gov/articles/PMC7842519/ Ma, A., Seminera, M., & Keller, C.L. (2025, April 17). Visa cancellations sow panic for international students, with more than 1,000 fearing deportation. The Associated Press. https://apnews.com/article/international-student-f1-visa-revoked-collegef12320b435b6bf9cf723f1e8eb8c67ae Marken, S. (2024). Americans More Confident in Two-Year Schools. Gallup. https://news.gallup.com/poll/646841/americans-confident-two-year-schools.aspx Mazzei, P. (2023). Sports Are In, Gender Studies Are Out at College Targeted by DeSantis. The New York Times. https://www.nytimes.com/2023/09/22/us/newcollege-florida-desantis.html Miner, M.A. (2020). Unmet Promises: Diminishing Confidence in Education Among College-Educated Adults from 1973 to 2018. Social Science Quarterly, 101(6), 2312-2231. https://doi.org/10.1111/ssqu.12873 National Center for Education Statistics. (2023a). Digest of Education Statistics: List of Current Digest Tables. https://nces.ed.gov/programs/digest/current_tables.asp National Center for Education Statistics. (2023b). Educational Attainment of Young Adults. https://nces.ed.gov/programs/coe/indicator/caa/young-adult-attainment National Opinion Research Center. (2021). The General Social Survey: Documentation and Public Use File Codebook. https://gss.norc.org/content/dam/gss/getdocumentation/pdf/codebook/GSS%202021%20Codebook%20R1.pdf 32 National Opinion Research Center. (2022). Survey: Americans See Cost as the Biggest Barrier to Higher Education. https://www.norc.org/research/library/survey-americans-see-cost-as-the-biggest-barrier-to-higher-educ.html Parker, K. (2019). The Growing Partisan Divide in Views of Higher Education. Pew Research Center. https://www.pewresearch.org/social-trends/2019/08/19/thegrowing-partisan-divide-in-views-of-higher-education-2/ Parker, K. (2021). What’s behind the growing gap between men and women in college completion? Pew Research Center. https://www.pewresearch.org/shortreads/2021/11/08/whats-behind-the-growing-gap-between-men-and-women-incollege-completion/ Pfeffer, F.T. (2018). Growing Wealth Gaps in Education. Demography, 55(3), 1033-1068. https://doi.org/10.1007/s13524-018-0666-7 Pew Research Center. (2024). From Businesses and Banks to Colleges and Churches: American Views of U.S. Institutions. https://www.pewresearch.org/wpcontent/uploads/sites/20/2024/02/PP_2024.02.01_institutions_REPORT.pdf Powell, A. (2025, April 15). Trump administration freezes $2.2 billion in grants to Harvard. Harvard T.H. Chan School of Public Health. https://hsph.harvard.edu/news/trump-administration-freezes-2-2-billion-in-grantsto-harvard/ Rousseau, D.M., Sitkin, S.B., Burt, R.S., & Camerer, C. (1998). Introduction to Special Topic Forum: Not so Different after All: A Cross-Discipline View of Trust. The 33 Academy of Management Review, 23(3), 393-404. https://www.jstor.org/stable/259285 Students for Fair Admissions v. Presidents and Fellows of Harvard College, No. 20-1199 U.S. 600 (2023). https://www.oyez.org/cases/2022/20-1199 University of Michigan Survey of Consumers (2025). Time Series Data. The University of Michigan. https://data.sca.isr.umich.edu/data-archive/mine.php U.S. Bureau of Labor Statistics. (2024). Education pays. https://www.bls.gov/emp/chartunemployment-earnings-education.htm U.S. Department of Education. (2025, March 7). DOJ, HHS, ED, and GSA Announce Initial Cancelation of Grants and Contracts to Columbia University Worth $400 Million. https://www.ed.gov/about/news/press-release/doj-hhs-ed-and-gsaannounce-initial-cancelation-of-grants-and-contracts-columbia-university-worth400-million Yamagishi, T., & Yamagishi, M. (1994). Trust and Commitment in the United States and Japan. Motivation and Emotion, 18(2), 129-166. https://doi.org/10.1007/BF02249397 Ziol-Guest, K.M., & Lee, K.T.H. (2016). Parent Income-Based Gaps in Schooling: CrossCohort Trends in the NLSYs and PSID. AERA Open, 2(2). https://doi.org/10.1177/2332858416645834 Zmerli, S., & Newton, K. (2008). Social Trust and Attitudes toward Democracy. The Public Opinion Quarterly, 72(4), 706–724. http://www.jstor.org/stable/25167660 34 Name of Candidate: Ella Sjoblom Date of Submission: April 24, 2025 35