Consider Example 7.11, where, prior to computing the interaction between the race/ethnicity of a player and the city’s racial composition, we center the city composition variables about the sample averages, percblck and perchisp (which are, approximately, 16.55 and 10.82, respectively). The resulting estimated equation is

(i) Why are the coefficients on black and hispan now so much different than those reported in equation (7.19)? In particular, how can you interpret these coefficients?

(ii) What do you make of the fact that neither black nor hispan is statistically significant in the above equation?

(iii) In comparing the above equation to (7.19), has anything else changed? Why or why not?

Transcribed Image Text:

log(salary) = 10.23 + .0673 years + .0089 gamesyr + .00095 bavg + .0146 hrunsyr (.0034) %3D (2.18) (.0129) (.00151) (.0164) + .0045 rbisyr + .0072 runsyr + .0011 fldperc + .0075 allstar (.0021) (.0076) (0.0046) (.0029) + .0080 black + .0273 hispan + .0125 black · (percblck – percblck) (.0840) (.1084) (.0050) + .0201 hispan • (perchisp - perchisp) (.0098) n = 330, R = 0.638.