In order to test whether the price of a standard good like soda varies across areas in ways that suggest
discrimination, we use ZIP code-level data on the price of soda in fast-food restaurants in New Jersey and
Pennsylvania, and estimate the following regressions:
ln(psoda) = -0.504 + 0.065prpblck + 0.044ln(income) + 0.075N J n = 401 R2 = .1678 (1)
ln(psoda) = -0.030 + 0.027prpblck + 0.083N J n = 401 I?2 = .1608 (2)
where prpblck is the proportion of the ZIP code population that is black, income the average income in the
ZIP code, and NJ a dummy variable for New Jersey.
1. Interpret the estimated coefficients on prpblck and on N J in model 1 (Remember to discuss sign and
size).
2. Comparing the results from the two estimations, is discrimination larger or smaller when you control
for the income variable? What do you infer from this on the correlation between the average income
and the racial characteristics of the zip code (justify your claim)?
