Article posted on April 27, 2012

Does home winning percentage or overall winning percentage more directly affect college baseball attendance? One can legitimately make a case for either side of this argument.

On the one hand, it’s logical to assume home winning percentage could be more strongly correlated with attendance for one patently obvious reason: fans want to see their team win ballgames at home. Such a dichotomy might suggest that fans are primarily interested in the game at hand, and are less invested in the program’s success as a whole. In contrast, college baseball fans could indeed be “true fans,” following their teams closely and electing to attend games based equally on home and overall winning percentages.

To analyze this question, we gathered attendance data from 44 of the top 212 Division I teams in attendance from the 2007 through 2011 college baseball seasons.

First, it is necessary to establish one important underlying assumption: teams do not perform equally on the road and at home. If this were not the case, our question would be rendered moot. From 2007 to 2011, the top 212 college baseball teams posted a .707 winning percentage at home, compared to a .499 winning percentage on the road. These figures appear very different, and statistical analysis corroborates the scope of the difference. Running a two-sample means test, we are given a staggering t-value of 16.46, indicating well less than a .001% likelihood that this distribution could have occurred by chance. 1 Thus, home and away records were indisputably distinct from one another.

The question remains, however, do fans care? What drives their decision to attend games: witnessing victories at home or knowing the overall strength of the program?

Let us further explore this topic. The average attendance for these 44 teams in 2011 was 3,252 and the correlation between home winning percentage and average attendance was r = .15. In other words, only 2.3% of the variance in attendance can be attributed to home winning percentage. 2 Moreover, the correlation between overall winning percentage and attendance was r = .16, indicating about 2.6% of the variance could be accounted for by changes in overall winning percentage. Regardless of which (if either) is more profound, we can conclude that there are certainly other more relevant factors driving attendance than winning percentages.

However, given the breadth of samples used in our analysis, these correlations are indeed significant. Granted, the strengths of the correlations are modest, but we can nonetheless say with confidence that overall winning percentage has a greater impact on attendance than does home winning percentage. 3 While we cannot fully gauge why fans choose to attend games, we know it is based less on home winning percentage than it is on overall winning percentage. Thus, our question has a clear answer: fans are most loyal to the overall achievement of their programs.

Other factors, such as social capital or respect for tradition, undeniably drive fans’ desires to attend college baseball games as well. This point is worth further consideration. If not winning percentage, what other factors drive fan attendance? For one, perhaps fan loyalty is not determined by the season at hand, but by the season in most recent memory. To analyze this possibility, we can run a correlation between winning percentage in the 2010 season, and attendance at home and overall in the 2011 season. In fact, the results are quite similar to those we see using all data from 2011. The correlation between home winning percentage and attendance in the following year is r = .20, and the correlation between overall winning percentage and subsequent attendance is r = .22. These figures, however, are not statistically significant and no conclusions can be drawn.

So, why might this be the case? Why might winning percentage—home or in general—have a relatively minimal impact on attendance in college baseball? Two reasons were mentioned prior. First, it is possible that fans simply attend out of school spirit and tradition for program, and will attend regardless of how the team performs. Second, attendance may be a matter of social capital, and people attend games to socialize or garner a new topic of conversation. It is also possible that, since college baseball’s popularity is less mainstream than other sports such as football or basketball, its audience is already inherently fragmented, and are perhaps zealous fans of baseball whose draw to games is isolated from team success.

Since a significant correlation still does exist, athletic directors (ADs) cannot entirely disregard their team’s on-field success when trying to fill their stadiums. However, given the statistically minimal impact of team performance on attendance, ADs should regularly explore how marketing and promotions can impact gate revenue.

Key Insights

It’s difficult to pinpoint the exact reason college baseball fans attend games. However, this analysis clearly demonstrates that overall winning percentage has a greater impact on attendance than does home winning percentage.

References:

  1. To run a two-sample means test, we divide the square of the standard deviations of each value (.113 for home and .144 for road) by the number of samples (212 for both). Then, we add these numbers together and take the square root of this resulting figure to arrive at the standard error. Subtracting the means from one another and then dividing by the standard error, we are given a t-value (16.46). Given this t-value and 422 degrees of freedom (given based on the t-value and sample size), the resulting p-value is .0001. The p-value represents the percentage that the distribution occurs by chance. Since it is so low, we can reject the null hypothesis that the distribution occurs randomly.
  2. To find the percentage of variance, we simply square the r-values.
  3. Given 212 subjects, the r-values of .15 and .16 garner statistical significance at the 1% level, as they give us t-values of 2.19 and 2.45, respectively. These t-values indicate that there is a 99% likelihood these values did not occur by chance.