POPWHITE: Percentage White of SMSA Population (%)
Stadium Capacity/ Metro Population TIME (0-8) CONSTANT Adj. R2
2.73* (1.87) –0.01*** (–2.72) 0.11** (2.26) .08
Burdekin et al. / NBA FANS INDIFFERENT TO RACE?
Models of Racial Composition of NBA Teams
Percentage of Team White(TWHITE) %
Coefficient (t statistic) Coefficient (t statistic) Coefficient (t statistic)
Percentage of Bench White(BWHITE) %
Percentage of Starters White(SWHITE) %
NOTE: The dependent variables in the models are (a) percentage of the team members who are White, (b) percentage of bench players who are White, and (c) percentage of the starters who are White. The t tests are based on robust standard errors. The results are based on 251 observations during nine seasons (1990-1999). Data on teams are from the Official NBA Register (Sporting News, various years); stadium capacity is from the NBA Web site (see www.nba.com), supplemented by telephone calls to teams to ver- ify capacity in various years; population is total SMSA population from the U.S. Bureau of the Census and County and City Extra: Annual Metro City and County Data Book (various years), and for Canadian cities, Statistics Canada (see www.statcan.ca) and the Canadian Ministry of Finance (see www.bcstats.gov.bc.ca). *t statistic significant at .10 level; **t statistic significant at .05 level; ***t statistic significant at .01 level.
Applying these possible interpretations to our cross-sectional analysis of team racial composition, window dressing implies that the coefficient on TWHITE will be positive if all markets have similar demand for White players, and that the inter- action effect between TWHITE and POPWHITE will be positive if demand for White players is higher in markets with larger White populations. Finally, if higher quality White players tend to play for teams with relatively high percentages of White players, then the cross-sectional coefficient on TWHITE will be positive due to the correlation between White player quality and TWHITE. To examine how these relationships have changed over the years of our study, we interact the mea- sure of the racial mix of the team with a time trend (TIME).11
Table 4 presents two alternative specifications. We measure home-game reve- nue as average ticket price times average attendance at home games. We include variables to control for the effect of team winning percentage, the number of com- peting major professional sport franchises (NFL, NHL, and MLB), stadium capac- ity, SMSA average income, and SMSA total population. The significance tests are based on robust standard errors. Except for the income variable (whose signifi- cance falls to the 90% level in one instance), all control variables are significant at the 95% confidence level or better, and all have the expected signs. In addition, the time trend is significant and positive, reflecting both rising attendance and rising ticket prices during the 1990s.12