A Comparison of Propensity Score and Linear Regression Analyses of Gender and Racial Wage Gaps in Information Technology Careers

• Activity Number: 148
• Type: Contributed
• Date/Time: Monday, August 12, 2002 : 2:00 PM to 3:50 PM
• Sponsor: Social Statistics Section*
• Author(s): Elaine Zanutto*+ and Michael Larsen
• Affiliation(s): University of Pennsylvania and University of Chicago
• Keywords: observational studies ; SESTAT ; CPS

Abstract

Propensity score methods are an alternative to the linear regression analyses commonly used to assess gender and racial wage gaps. The usual linear regression analysis of a gender wage gap predicts an outcome such as weekly earnings from a linear regression model that includes an indicator variable for gender along with other relevant covariates such as age, education, experience, and recent training. A statistically significant gender wage gap is declared when gender is a significant predictor of weekly earnings even after controlling for other covariates. An alternative method of assessing the gender wage gap is to use propensity score analysis techniques to create groups of male and female Information Technology (IT) professionals that are balanced, in terms of background covariates, so that subsequent outcome comparisons, made within these balanced groups, are not confounded by differences in background covariates. Similar comparisons can be made for racial groups. Using data from the Current Population Survey and NSF's SESTAT database we compare the performance of these two statistical methods for assessing gender and racial wage gaps in IT careers.