Factors that drive daily attendance at minor league baseball games

This paper examines different factors that are thought to drive attendance at Minor League Baseball games to test what the actual effect of those factors is. It takes into account factors such as day of the week the game is played, month the game is played, whether the game is played on or around a holiday, and team performance to date. I also examined how each of these factors vary across different levels of Minor League Baseball. My initial hypothesis was that each of these would have a relatively strong effect on the game's attendance. As expected, the results showed that weekend games in summer tended to have much higher average attendance than other games and holiday games had the highest attendance overall. These effects were clear across all levels of Minor League Baseball, though they were more pronounced in the higher levels where attendance is higher with more variability. One unexpected effect that I concluded from the study is that attendance of games just outside of holiday weekends was negatively impacted by the proximity of holiday games.

FACTORS THAT DRIVE DAILY ATTENDANCE AT MINOR LEAGUE BASEBALL GAMES By Dawson Hughes 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 Quantitative Analysis of Markets and Organizations Approved: ______________________________ Scott Schaefer Thesis Faculty Supervisor _____________________________ Mike Cooper Chair, Department of Finance _______________________________ Scott Schaefer Honors Faculty Advisor _____________________________ Sylvia D. Torti, PhD Dean, Honors College May 2019 Copyright © 2019 All Rights Reserved ACKNOWLEDGEMENTS I would like to thank Baseball Prospectus for supplying me with the data necessary to make this project possible. I also thank Minor League Baseball, MiLB, for its willingness to support this project. Lastly, I thank the Salt Lake Bees organization for supplying me with the supplemental data I used to further my studies beyond the initial scope of the project. ii ABSTRACT This paper examines different factors that are thought to drive attendance at Minor League Baseball games to test what the actual effect of those factors is. It takes into account factors such as day of the week the game is played, month the game is played, whether the game is played on or around a holiday, and team performance to date. I also examined how each of these factors vary across different levels of Minor League Baseball. My initial hypothesis was that each of these would have a relatively strong effect on the game’s attendance. As expected, the results showed that weekend games in summer tended to have much higher average attendance than other games and holiday games had the highest attendance overall. These effects were clear across all levels of Minor League Baseball, though they were more pronounced in the higher levels where attendance is higher with more variability. One unexpected effect that I concluded from the study is that attendance of games just outside of holiday weekends was negatively impacted by the proximity of holiday games. iii TABLE OF CONTENTS ACKNOWLEDGEMENTS ii ABSTRACT iii TABLE OF CONTENTS iv INTRODUCTION 1 LITERATURE REVIEW 3 METHODS 6 LIST OF VARIABLES AND SUMMARY STATISTICS 10 RESULTS AND DISCUSSION 12 FURTHER RESEARCH 25 REFERENCES 27 iv 1 INTRODUCTION In the past, economists have completed numerous studies on the economics of sports. Perhaps the most famous sports economics studies of all time have been those studying the economic impacts of the Olympics on the host city or country. These studies proved to be very popular and successful, which paved the way for further studies in the field of sports economics. Sports economists began studying similar topics, such as the economic impacts of professional sports teams and leagues on their local municipalities. Even as the amount of research done on the economics of sports grew, it wasn’t until fairly recently that sports economists took an interest in the economics of the minor leagues. Even with the new studies, there is still only a fraction of published research on the minor leagues of sports when compared to their professional counterparts. My interest in this topic began by studying the impacts of Minor League Baseball teams on their local economies, however, there is already quite a bit of established research on that topic at this point. I decided to focus my research solely on the factors that drive attendance at those games. While there is obviously a plethora of factors that drive attendance—team performance, local fan base, game day promotions to name a few—this paper will specifically focus on how attendance is affected by calendar effects such as day of week, month of year, and proximity to a holiday. While the other factors are equally relevant to finding the true drivers of attendance, I’ve narrowed the scope of study to calendar effects and spread it across each level of Minor League Baseball. In this paper, I will discuss those effects in detail and how they vary across each level of play. The unique aspect of this research is that it examines games day-by-day. Most research of this type examines a season as a whole rather than on a micro level. For example, studies 2 in the past have concluded what the expected attendance of a given team or league is for a season. Instead, I have calculated the expected attendance of a given team or league on a specific date, day of week, and factored in how close that game is to a holiday. From what I could find, this game-by-game approach has not been studied extensively for Minor League Baseball. Coming into this study, I had several initial hypotheses. The first hypothesis was that weekend games would post higher average attendance than weekday games. This seems fairly intuitive—more people work on weekdays than weekends so it is more difficult for them to attend a baseball game. Saturdays in particular should have the highest attendance, since on Fridays people are just getting off work and on Sunday many people may not want to go to a later game if they have work the next day. My next hypothesis was that games in the mid-summer months would have higher attendance than those early or late in the season. In many places that have MiLB teams, it can still be very cold out throughout May and many April games are rainy. However, by the middle of summer when the weather is nice in most places, more people will be interested in going to a baseball game as their form of entertainment instead of indoor entertainment during the other months. The last hypothesis was that overall, games on or around holidays would have the highest attendance. Most Minor League Baseball teams have promotions on holidays, such as fireworks after the games, in an attempt to draw some of their biggest crowds. Due to the amount of resources they put into holiday games and the fact that most people have those days off work, they should be able to bring in some of their biggest crowds for these games. 3 LITERATURE REVIEW As mentioned before, there is very little published research of the same nature as what I have done in this study. The most extensive study of drivers of attendance at Minor League Baseball games was done by Richard J. Cebula in his paper “The Potential Role of Marketing in Promoting Free Enterprise in the U.S.: A Study Involving Minor League Baseball and Ticket- Sales Revenue Maximization.” In this work, Cebula uses a large number of driving factors including dummy variables for if there is a discount on beer, if the game features live music, fireworks, and group discounts. Cebula uses a couple of the same variables that I used, including weekend games or summer games, but does not break it out by every day and every month. His work uses more variables but a smaller sample size than mine; his dataset is made up only of the games played in the Carolina League in 2006. While he comes up with many statistically significant results, it only examines results of the eight teams in the Carolina League for one season. There are 256 Minor League Baseball teams with a Major League affiliate, so it is difficult to draw any conclusions about all of Minor League Baseball based on the one league for one season. One last key difference in our papers is that while his is calculating effects on attendance, the goal of his paper is maximize revenue at games given that ticket prices in the Carolina League are all about the same price, whereas mine is focused completely on attendance and has no pricing data. Nola Agha has published several papers examining the economics of Minor League Baseball. Unlike what most research has found on the subject of professional sports, Agha found in “The Economic Impact of Stadiums and Teams: The Case of Minor League Baseball” that Minor League baseball teams often provide positive 4 economic benefits to the cities in which they play. In this paper, he argues the pros for building stadiums specifically for Minor League Baseball teams. Building stadiums for cities to bring in professional sports franchises has long been the subject of hot political debate, but Agha claims that Minor League Baseball teams have much higher potential to bring positive economic benefits than their professional counterparts. While there are collinearity issues with professional sports since most pro sports teams are located in large cities, Minor League Baseball does not face this same issue due to the large variation in location of teams. Agha did not find many significant positive effects of Minor League Baseball teams, but there were a few positive effects, unlike professional sports where the effects are regarded as being statistically significant overwhelmingly negative overall. Another one of Agha’s examined specifically how Minor League Baseball teams could create economic benefits. In “An Explanation of Economic Impact: Why Positive Impacts Can Exist for Smaller Sports,” Agha lays out nine conditions that must be present in order for there to be positive economic impacts. The nine conditions are as follows: new visitors, geographic isolation, locals change spending, locals stay locally, government spending, new stadiums, venue utilization, and crowding out. Agha claims that when these conditions are present, there is the potential for positive economic impacts, even for professional sports teams. According to Agha, professional sports teams are much less likely to meet the conditions, thus explaining why they are less likely to bring economic benefits to their cities than Minor League teams, which are far more likely to meet the conditions. As a result, Minor League Baseball teams have fairly high potential to bring positive economic impacts. 5 Yeliz Yalcin, David Denaux, and Zulal Denaux conducted a study on factors that drive attendance at Major League Baseball games. Although I expect the factors to be quite different at the Major League and Minor League levels, this can still help to understand some factors that get people to go to baseball games in the first place. They examined day of week, month of year, day or night, whether or not it was an interleague game, and ticket price. Additionally, the study examines how team performance can affect attendance. As expected, many of these factors turned out to be statistically significant. Similar to Minor League Baseball, weekend games and those in mid-summer tend to have higher attendance than weekday games and those early in the season. Some of the factors, such as interleague play, don’t translate to Minor League Baseball due to the different structure of the leagues and demand to see specific teams. Although it is uncertain if team performance has an effect on attendance at Minor League games, it definitely has an effect on attendance at Major League games. The study found that a home team with a win percentage of greater than .500 draws on average 4,175 more fans than a home team with a win percentage of lower than .500 on the day of the game. Economic factors were taken into consideration as well, such as home city unemployment rate, though it wasn’t found to be statistically significant in determining attendance. 6 METHODS Baseball Prospectus supplied me with data for every Minor League Baseball game played between 2005 and 2017 for the Rookie, Short Season A, Single A, Advanced A, Double A, and Triple A leagues. The initial dataset included data on 171,965 games played with 13 variables. In order to make the data meet the requirements to pursue this study, I implemented an extensive data cleaning process. There were several pieces of the data that were missing important values, contained values that could be misleading, or had a value of 0 for attendance, the most important variable of the study. I completed all calculations and processes using Python in Jupyter Notebook. The first step of the data cleaning process was finding a solution to the issue of games with the attendance listed as 0. It is unlikely, if not impossible, that there were actually any Minor League Baseball games played at any level during this time period where not a single fan showed up. Given this, I hypothesized that all zero-attendance games were caused by one of two potential scenarios: the game played was part of a double-header in which the attendance was only listed for one of the games, or there was a human error where the scorekeeper simply forgot to list the attendance for a given game. In order to find what was the case more often, I created dummy variables to indicate whether or not the game was a double-header. To do this, any time the “GAME_ID” variable ended in a 2 or 1 where there was a 2 on the same day, both games would be marked as a double-header. After filtering all the games to only include those with zero attendance, it was an obvious trend. Any time a game was marked as having zero attendance, it was simply a case of a double-header in which it would not make sense to list attendance twice. When there is a double-header, teams typically sell tickets 7 to both games as a pair without changing the price of the ticket. If there is a game played at 1:00 PM and another game played at 5:00 PM, fans buy one ticket and are allowed to stay for both of the games. Teams don’t typically count how many fans stay or leave for the second game, so the whole day of games is just listed as one event. To avoid doublecounting fans, one game of each double-header is listed as having zero attendance and the true attendance is listed for the other game. In order to alleviate the issue of zeroattendance games skewing the outcomes, I dropped them from the dataset as soon as I was confident they could all be explained by double-headers that only listed attendance once. Another issue I faced early on was teams that changed names throughout the years or even teams whose abbreviation switched. For example, throughout the dataset the Orem Owlz were sometimes listed as “ORE” and sometimes listed as “ORM”. Throughout the years there were also many teams that switched locations or changed their name. While I initially saw this as a problem, it turned out to be a non-factor since we decided just to focus the study on each season within a calendar year, rather than comparing team effects across multiple calendar years. For any future studies that examine attendance effects for each specific team across multiple calendar years, this would be an important consideration. Since team names remained constant over the course of one season at a time, I didn’t include this in the scope of the study. On occasion, teams switched stadiums in the middle of the year. This happened for a few reasons—mostly due to stadium construction and extended periods of poor weather such as hurricanes. These cases have the potential to have a strong effect on attendance that isn’t captured by the other explanatory variables in the dataset. One case 8 that happened a few times was that while a team’s stadium was under construction, they had to play all of their remaining home games on the road. If the would-be home team otherwise consistently draws large crowds in their regular stadium, playing all of their games on the road could hurt their attendance figures. It is unlikely that the fan base will follow the team close enough to travel to all of the away games that they would have otherwise gone to had they been playing at their normal home stadium. Alternatively, the reverse scenario may hold true as well. If there is a team that consistently draws small crowds and underperforms at home, their season attendance figures may benefit from exclusively playing road games, where the home teams are able to consistently draw larger crowds than the would-be home team would have otherwise been able to draw. While we expect that this has the potential to have a strong effect on the attendance of games, I did not remove these cases from the dataset. Both scenarios are equally as likely, though it is difficult to actually calculate the counterfactual in these scenarios to know what the difference would have been if the teams had played all their games at home. At one point, I created a separate dataset including only those cases where a team played fewer games at their home stadium than expected. Almost all of these cases could be explained by the aforementioned reasons, though there was not a significant amount of them. Out of nearly 172,000 observations, these unusual cases made up fewer than 100 that I was able to find and did not seem to hold a significant effect. For that reason and since they weren’t likely to make a large difference in the results either way, I decided not to drop these observations from the dataset. The next step in preparing the dataset for use was removing the teams from the Mexican Leagues. Although there is plenty of data on the Mexican League teams, this 9 was beyond the scope of what I wanted to include in the study. Additionally, there are many other factors that could affect attendance at games in the Mexican League than in the United States. Many of the teams in the Mexican League bring in huge attendance figures fairly often that would skew the rest of the data. Furthermore, the Mexican League typically follows a different calendar from the rest of the Minor Leagues in the United States, so since this study focuses primarily on attendance effects based on the calendar, things such as different holidays in Mexico and the United States could have a large effect on the data that would not be helpful to this study. After cleaning the data, I ran regressions on a variety of things related to the calendar effects, beginning with attendance by month. After starting with attendance by month, I added in attendance by day of week, date of year, and whether or not the game falls on or around a holiday. To find the marginal effect of each, I created interaction variables between each of the different types of variable as well as the level of play. I was able to use each of these to find the marginal effect of each variable as well as the difference in effect of each variable across the different levels. Lastly, to further the scope of the project, I included data given by the Salt Lake Bees on when they had games that featured fireworks in 2018. This was not included in the original dataset at all and was supplied directly by the Bees franchise without other additional data, but I was able to use it to calculate the marginal effect of fireworks on attendance for otherwise similar Bees game for the 2018 season. However, this data is likely not enough to be extrapolated to reflect any type of effect of fireworks at any other teams’ games for any other year. 10 LIST OF VARIABLES AND SUMMARY STATISTICS BY LEVEL YEAR_ID == Year the game was played ¾ Values between 2005 and 2017 LVL == Level of Minor League Baseball ¾ Values include ROK for Rookie League, ASX for Short Season A, AFX for A, AFA for Advanced A, AAX for Double A, and AAA for Triple A GAME_ID == An indicator variable so that every game has a unique identifier ¾ Composed of home team abbreviation followed by year, month, day, and number of game that day ¾ Example: OGD200506251 Indicates a game played at home by the Ogden Raptors on June 25, 2005 that was the only game played that day (not a doubleheader) HOME_TEAM_ID == A three letter abbreviation for which team is considered the home team for each game ¾ Example: OGD indicates the Ogden Raptors AWAY_TEAM_ID == A three letter abbreviation for which team is considered the away team for each game PARK_ID == An indicator variable so every stadium has a unique identifier ¾ Example: 2802 indicates Lindquist Field, home of the Ogden Raptors venue == The name of the stadium where the game is played ATTEND_PARK_CT == The total number of attendees at the given game 11 AAA Summary Statistics Count 28841 Mean 6676 A Summary Statistics Standard 2914 Count 27843 Mean 3603 Standard 2310 Deviation Deviation AA Summary Statistics Count 28023 Mean 4433 Standard 2088 Deviation Advanced A Summary Statistics Count 25866 Mean 2270 Standard 1612 Deviation Rookie Summary Statistics Count 11109 Mean 1618 Standard 1054 Deviation 12 RESULTS AND DISCUSSION Chart 1 AAA stadiums range in capacity from 6,500 to 16,600 Chart 1 above and each of the following charts all represent the same thing. Each of them graphically displays the attendance count for every game in each level through a group of bins. For example, the chart above displays the frequency of games by given attendance levels in segments. It can be interpreted by matching the attendance with the number of times a game fell in that attendance bucket. There were fewer than 1000 games in AAA where the attendance was 0-2,000 and there were more than 5,000 games each where attendance was in the buckets between 4,000 and 8,000. Given this, the chart shows that most frequently AAA games have attendance in the 4,000-6,000 range. From 10,0000-12,000 attendance games, there is a huge drop off in frequency. At 12,000 and above, there are very few observations overall and those games likely had other factors affecting them, such as a star Major League player playing or a big promotion night. 13 The following charts all show similar trends as well. Chart 2 suggests that most AA games have attendance counts in the 3,000-5,000 range before it starts to drop off. Much like the AAA histogram, AA also includes a few observations of games with substantially higher attendance counts than most of the other games. It’s highly unlikely to have a AA game with attendance greater than 10,000, so these games strongly suggest outside factors at play such as a top player or human error as mentioned above. Though most of the attendance histograms are similarly shaped, Chart 5 for Rookie League attendance stands out. It has a unimodal distribution, with a far higher concentration of games in the bin just below 1,000 attendees than any other of the charts have in any single bin. Chart 2 AA Stadiums range in capacity from 5,038 to 11,000 14 Chart 3 Chart 4 15 Chart 5: Rookie Attendance Histogram Chart 6: Short Season A Attendance Histogram 16 Regression 1: ATTEND_PARK_CT = B0 + B1MAY + B2JUNE + B3JULY + B4AUGUST + B5SEPTEMBER Table 1: Attendance by month across all levels R-squared: 0.005 Observations: 121682 Table 1 above displays the total effect on attendance of only the month of the year. It is not broken out by level of play, so the numbers reflect the average across all levels, Rookie League-AAA. As the omitted variable, the average attendance for the month of April is 3,624, the intercept. The coefficient on each of the following variables represents the additional expected attendance per month when compared to a game in April. For example, the month of June has a coefficient of 639, so the expected attendance for a game in the month of June across all levels is 4,263 (3,624 + 639). As displayed by the table, June has the highest average attendance across all levels, which is to be expected given that it is right in the middle of the summer months when more people would likely want to go to a baseball game. April, however, is the lowest attended month on average across all levels. This is also not surprising, given that it is often still cold and rainy in many places in April. Early in the season, teams can expect lower 17 attendance that will spike in the middle of the season before it declines again towards the end. Attendance by month is parabolic, though with a spike again at the end of the season for the last few games in September. The large t-statistic for each coefficient suggests that they are all statistically significant. There is one key problem that arises with pooling all of the levels together for attendance: not all levels play the same amount of games per month. Specifically, the Rookie League never has any games until June. The lower average attendance of Rookie League games drags down the overall average for June-September so April and May appear higher than they likely otherwise would be. Table 6 shows the amount of games played per month by level. Regression 2: ATTEND_PARK_CT = B0 + B1MONDAY + B2SATURDAY + B3SUNDAY + B4THUSDAY + B5TUESDAY + B6WEDNESDAY Table 2: Attendance by day of week across all levels R-squared: 0.066 Observations: 121682 18 Table 2 also includes all levels, so it is the expected attendance on a given day for any Minor League Baseball game. Like Table 1, all values on this table are statistically significant due to their high t-statistics. Friday games are the omitted variable, so they are represented by the intercept of 4,914. This means that the average game on a Friday over the course of the entire season for all teams has an expected attendance of 4,913. Each of the coefficients show the expected difference between the attendance of the average Friday game and the expected attendance of a game on any other given day of the week. Saturday, the only day with a positive coefficient, has the highest average attendance of all days at 5,203. Since every other day has a negative coefficient, Saturday has the highest average attendance followed by Friday. Monday through Wednesday all have similar coefficients of between -1,574 and -1,553 for expected attendance of between 3,340 for a Monday game and 3,361 for a Tuesday game. While Sunday also has a negative coefficient, at -1,042 it is not nearly as large as the Monday-Wednesday games, confirming my initial hypothesis that weekend games would have much higher attendance on average than weekday games. Interestingly, Thursday games actually have higher expected attendance than Sunday games. This means that although Thursday is a weekday, it draws attendance closer to weekend games than weekday games. This is likely due to long weekends where people have Friday off work so they decide to go to a baseball game on Thursday. Additionally, there is the possibility that Thursday is a popular day for promotions that affect attendance. Many teams host weekly promotions on Thursday such as Thirsty Thursday discount beer days. It is uncertain whether or not these promotions have a statistically significant effect on attendance, but it would not be 19 surprising if that is one of the key factors in Thursday having such high attendance when compared to other weekday games that have much lower expected attendance. Regression 3: ATTEND_PARK_CT = B0 + B1MAY*LVL + B2JUNE*LVL + B3JULY*LVL + B4AUGUST*LVL + B5SEPTEMBER*LVL Table 3: Attendance by month and level R-squared: 0.394 Observations: 121682 Intercept: Triple-A aax: Double-A afa: Advanced-A afx: Single-A 20 Table 3 has the same information as Table 1 but broken out by level of play. The omitted variable is a AAA game in April and each of the following month variables without a level attached are also from AAA. With an intercept of 5,482, a AAA game in April has on average 1,858 more people in attendance than the 3,624 expected attendees at the average April game across all levels. Attendance at the AA level is the closest to the average of all levels, suggesting that the much higher attendance in AAA pulls the average up quite a bit since the several levels below don’t have as wide of attendance ranges. AAA also follows the same month-by-month trend as the overall trend, with attendance peaking in June, declining through July and August and spiking again in September. This is a similar trend across all levels, with every single level reaching its peak attendance in July and having the lowest attendance in April. This again confirms my initial hypothesis that summer months would have the highest attendance, regardless of level. This also suggests that it is one of the strongest overall driving factors of attendance at Minor League Baseball games since it is so consistent and such a high correlation across all levels. Across all levels, September has a much higher standard error level than the other months. This likely has two causes. First, September games are not common at the Minor League levels, so there are far fewer data points of September games than there are of any other month. The fewer amount of observations can lead to higher variability of the recorded attendance for the month of September. Additionally, this is not controlling for the fact that Labor Day falls right at the beginning of September, roughly the last week of play for most Minor League tames. If a team plays four games in September and one of 21 those is a holiday game, the standard error will likely be much higher than another month since a quarter of the games played were affected by the holiday. For other months with a holiday, such as July, this is not nearly as big of a factor since games are played on all 31 days, the other 30 days can wash out the effect of the holiday game. Regression 4: ATTEND_PARK_CT[LVL] = B0 + B1MAY + B2JUNE + B3JULY + B4AUGUST + B5SEPTEMBER + B6SATURDAY + B7SUNDAY + B8MONDAY + B9TUESDAY + B10WEDNESDAY +B11THURSDAY Table 4: Month and day attendance broken out by level Observations: 121682 22 Table 4 combines the information from each of the previous tables into one table than includes level of play as well as month of year and day of week. All values again are statistically significant. The intercept for each level is given by the expected attendance of a Friday game in April in each level. This table shows how pronounced the differences are in attendance across levels. Even among Friday games in April, there is a drop-off of nearly 2,000 attendees from AAA to AA and 2,000 more to Advanced-A. All levels follow all the same trends across months and days. Saturdays always have the highest attendance followed by Fridays, and all levels aside from Rookie have the lowest attendance on Monday. Rookie League has the lowest attendance on Wednesday. Also, all leagues follow the same trend of having higher attendance on Thursdays than on Sundays. 23 Regression 5: ATTEND_PARK_CT= B0 + B1MAY + B2JUNE + B3JULY + B4AUGUST + B5SEPTEMBER + B6SATURDAY + B7SUNDAY + B8MONDAY + B9TUESDAY + B10WEDNESDAY + B11THURSDAY + B12MEMORIAL DAY + B13JULY4TH + B14LABOR DAY Table 5: Month, day of week, and holidays R-squared: 0.081 Observations: 121682 Table 4 includes month data, day of week, and whether or not the game was played on or around a holiday. One issue with this is that while Memorial Day and Labor Day fall on different dates each year, only one date was used for each of them. Memorial Day is assigned to May 31 and Labor Day is assigned to September 1 since it varies yearby-year. As a likely result of this, the R-Squared for this is much lower than I would expect otherwise taking holidays into account, at .081. Unsurprisingly, however, each holiday has a statistically significant effect on attendance, with the 4th of July having overall the highest attendance of any day in the calendar for the year. 24 Table 6: Number of games per month by level April May June July August September AAA 4585 5955 5735 5621 8015 930 AA 4520 5780 5349 5672 5749 953 Advanced A 4381 5404 4743 5321 5191 826 Single A 4467 5800 5148 5693 5799 936 Rookie 0 0 1774 4416 4303 616 As mentioned before, not all months have the same amount of games played per level. While Single-A through Triple-A are fairly uniform April-September, the rookie league has far fewer games in every month and has 0 games played in April and May. The likely effect of this is that any time an average is given that isn’t broken out into different levels, the pooled averages are skewed due to lower attendance in the Rookie League. 25 FURTHER RESEARCH There is still plenty of additional research that could be done on this topic. This paper has covered the various calendar effects on attendance fairly extensively, however, there is still lots of work that could be done. Just in the dataset that I have, there were several variables that I didn’t use that could have key effects on attendance at Minor League Baseball games. For example, the home runs and runs scored variables could have large effects. If a team has been scoring lots of runs lately, I would expect to see that reflected in their game-by-game attendance figures. Even among these games, I would expect that teams that hit more home runs average higher attendance than teams scoring the same amount of runs but hit fewer home runs. Fans are presumably more likely to go to games they deem more exciting and these are just a couple of the “excitement factors.” Possibly even more important excitement factors are teams’ win-loss record. If the current number one team is hosting the current number two team, fans will likely be much more excited about the game than if it is two teams towards the bottom of the standings. Similarly, if either the home team or visiting team is coming into the game on a long win streak, I would also expect that to have a positive effect on attendance while a long losing streak would have a negative effect. Effects such as these have been measured and hold true at the major league level of sports, however there is less established research on what the effect is at the minor league level, where fans are more likely there just for fun or to see a top player rehabbing than they are to see their favorite team be competitive. That leads into another opportunity for further research that wasn’t included in my dataset: effects of top Major League Baseball players who have been temporarily 26 assigned to an MiLB team for rehab. One of the most exciting things that can happen for fans of a local Minor League Baseball team is when one of their favorite star MLB players is assigned for rehab with their local team. If Mike Trout of the Angels ever did a rehab stint with the Salt Lake Bees, attendance numbers would almost certainly reflect that a star is in town for the time being. My dataset didn’t include anything of that sort, but some digging for top players’ time in the Minor Leagues could yield strong results when measured against attendance. Another likely major factor in determining attendance at Minor League Baseball games is promotions. It is a well-known strategy by teams in an attempt to boost attendance. Frequently, teams will run regular promotions throughout the season such as “Thirsty Thursday” beer discounts, post-game fireworks, or giveaways at the gate. Given the right data, one could calculate the marginal effect of each type of promotion. 27 REFERENCES Cebula, Richard J. “The Potential Role of Marketing in Promoting Free Enterprise in the U.S.: A Study Involving Minor League Baseball and Ticket- Sales Revenue Maximization.” Southeastern.edu, June 2009, www2.southeastern.edu/orgs/econjournal/index_files/JIGES%20JUNE%202009%20CE BULA%208-8-09%20Private_Enterprise,_Marketing%20para.pdf. Agha, Nola. “ The Economic Impact of Stadiums and Teams: The Case of Minor League Baseball.” Sagepub, 2013, journals.sagepub.com/doi/pdf/10.1177/1527002511422939. Agha, Nola. “An Explanation of Economic Impact: Why Positive Impacts Can Exist for Smaller Sports.” Emerald Insight, 2016, www.emeraldinsight.com/doi/pdfplus/10.1108/SBM-07-2013-0020. Lahman, Sean. “Baseball in the Age of Big Data.” Seanlahman.com, 4 Aug. 2013, www.seanlahman.com/2013/08/04/baseball-in-the-age-of-big-data/. Michael, Steven William. “Evaluations of Minor League Baseball Stadiums.” Iowa State Digital Repository, 2010, lib.dr.iastate.edu/cgi/viewcontent.cgi?referer=https://www.google.com/&httpsredir=1&ar ticle=2599&context=etd. 28 Baade, Robert A, and Richard F Dye. “Sports Stadiums and Area Development: A Critical Review.” Sagepub, Aug. 1988, journals.sagepub.com/doi/pdf/10.1177/089124248800200306. Baade, Robert A, and Richard F Dye. “ AN ANALYSIS OF THE ECONOMIC RATIONALE FOR PUBLIC SUBSIDIZATION OF SPORTS STADIUMS.” Heartland.org, www.heartland.org/_templateassets/documents/publications/baade1988.pdf. Denaux, David, et al. “Factors Affecting Attendance of Major LeagueBaseball: Revisited.” Researchgate.net, 26 May 2011, www.researchgate.net/publication/225435375_Factors_Affecting_Attendance_of_ Major_League_Baseball_Revisited.