The authors address a methodological issue in estimating pay penalties to women professionals—whether the dependent variable in the pay equation should be hourly wage or earnings measured over some other period such as weeks, months, or a year. The authors analyze data on college graduates working in professional specialty occupations and on physicians, and they report two main conclusions. First, the appropriate specification of the pay equation for (employed) professionals is with ln(earnings) as the dependent variable with ln(hours) included as an independent variable. Second, the implications of using this specification for estimates of gender pay differentials are serious. In the data on professionals, respecification of the pay equation changes the estimate of the pay penalty to women from-4 6 to-7 6 percent. In the data on physicians, it changes the estimate from no penalty to a very large one, from-1 8 to-13.0percent.
Key Words: quantitative methods • methodology (data analysis) • compensation • professional workers • gender inequality