Chapter 13
Experiments and Quasi-Experiments
Solutions to Empirical Exercises
1. The following table provides answers to (a)–(c)
Dependent Variable === Call_Back
Regressor (1) (2) (3) (4)
Black −0.032**
(0.008)
−0.038**
(0.012)
−0.023*
(0.011)
Female × Black 0.008
(0.012)
High
0.014
(0.008)
0.023
(0.012)
High × Black
−0.018
(0.016)
Intercept 0.097**
(0.006)
0.097**
(0.006)
0.073**
(0.005)
0.084**
(0.008)
Significant at the 5% * and 1% ** level.
(a) From (1) in the table, the call-back rate for whites is 0.097 and the call-back for blacks is
0.097 − 0.032 = 0.065. The difference is −0.032 which is statistically significant at the 1% level.
These number implies that 9.7% of resumes with white-sounding names generated a call back.
Only 6.5% of resumes with black-sounding names generated a call back. The difference is large.
(b) From (2) in the table, the call-back rate for male blacks 0.097 − 0.038 = 0.059, and for female
blacks is 0.097 − 0.038 + 0.008 = 0.067. The difference is 0.008, which is not significant at the
5% level.
(c) From (3) in the table, the call-back rate for low-quality resumes is 0.073 and the call-back rate
for high-quality resumes is 0.073 + 0.014 = 0.087. The difference is 0.014, which is not
significant at the 5% level. From (4) the (high-quality)−(low-quality) difference for whites is
0.023 and for blacks is 0.023 − 0.018 = 0.005; the black-white difference is −0.018 which is not
statistically significant at the 5% level.
(d) The following table shows estimated means of other characteristics for black and white sounding
names. There are only two significant difference in the mean values: the call-back rate (the
variable of interest) and computer skills (for which black-named resumes had a slightly higher
fraction that white-named resumes). Thus, there is no evidence of non-random assignment.
