|
Activity Number:
|
431
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 12, 2004 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Business and Economics Statistics Section
|
| Abstract - #301974 |
|
Title:
|
The Use of Interactions in Analyzing Gender/Race Differences in Compensation
|
|
Author(s):
|
Gholam H. Borhani*+ and Chris Haan
|
|
Companies:
|
Economic Research Service and Economic Research Service
|
|
Address:
|
9009 Eagles Ridge Dr., Tallahassee, FL, 32312,
|
|
Keywords:
|
interaction ; compensation ; gender ; race
|
|
Abstract:
|
Statistical techniques are often used in litigation and regulatory settings to determine whether there are statistical differences in compensation between protected group and unprotected groups of employees. Multiple regression analysis commonly used to assess differences in compensations of protected and unprotected groups after other factors that might affect compensations are taken into account. Due to the complexities involved in compensation decision-making process and the myriad factors that may affect the compensation decisions, it becomes very tempting to include many different factors in a regression model. With more than one independent variable in the model, however, it is quite possible that there may be interactions between these variables. Therefore, when the regression analysis is used to decide whether there are compensation differences the interactions between the independent variables has to be also assessed. We present a simple hypothetical example, in which we show how ignoring interactions between independent variables in the regression analysis may lead the researcher to erroneous conclusions.
|
