In lieu of an abstract, here is a brief excerpt of the content:

Appendix 1 Radio and Sunday Ball’s Effect on Attendance In the cases of radio, Sunday ball, and night ball, the benefits were varied and enmeshed with other factors. To help sort out the effects for radio and Sunday, I ran regression equations covering the seasons between 1929 and 1941. There were 208 observations, one for each year for each of the sixteen teams. I looked at each team’s win-loss record, whether a team shared a city with another club, whether a team had night ball, whether a team had Sunday ball, estimated city population, and real per-capita Gross National Product (gnp). The estimated city population variable uses the 1930 and 1940 census figures. The population figures for the intervening years were derived by a simple adjustment of one-tenth of the decade’s change per year. Obviously this is a crude estimate of a city’s population for non-census years. I tried using real spectator sports spending as an indicator of the economic conditions but the real per capita gnp variable was a better explanatory variable. For the dependent variables I used attendance (or its natural-log transformation) and real profit/loss. Given the potential “noise” in the profit/loss figures from player sales/purchases (which were not amortized ) and from installing lights or other stadium improvements, the relatively weak explanatory power of the equations is not surprising. A dummy variable for radio broadcasts (1 if the team broadcast home games during a given season; 0 if not) was not definitive, given the ambiguity concerning Detroit and Cleveland’s broadcasting of home 308 Appendix 1 games before 1935. In equations where the natural-log transformed attendance was the dependent variable and where Detroit and Cleveland are assumed to have broadcast home games early during the 1930s (1929–34 for Detroit and 1930–32 for Cleveland), the dummy variable was statistically significant. The coefficient was positive. If Detroit did not broadcast games in 1934 or if Cleveland’s dummy variable is altered , the radio dummy variable was not significantly significant. For equations using real profit as the dependent variable, the radio broadcast dummy variable was never statistically significant. There is some evidence, therefore, that radio broadcasts of home games might have had a positive effect, but this result depends largely upon the status of broadcasting in Detroit and Cleveland, and I do not include the radio dummy in the following discussion. On a game-by-game basis, night games were statistically significant in the Philadelphia Phillies’ regression equations.1 However, equations using game-by-game data could not address the issue of whether night games merely rearranged attendance or augmented it. In the current cross-team seasonal equations, the night baseball dummy variable (1 if by night; 0 if no night) was frequently not statistically significant. These results may arise from the relatively high correlation between the night dummy variable and real per-capita gnp—.53. Owners installed lights after the economic conditions improved.2 In the equations using real profit/loss as the dependent variable, the Sunday and night dummy variables were not statistically significant . Although there was a win-loss percentage variable, including a dummy variable for pennant winner increased the equation’s explanatory power, but the pennant-winner variable had a relatively high correlation coefficient of .55 with win-loss percentage. The evidence that Sunday and night ball improved a typical team’s profit or loss is slender. MacPhail’s Reds and Dodgers appear to have been the prime beneficiaries of night ball. Sunday ball could not reverse the Athletics’ slide into mediocrity under Connie Mack. ...

Share