Premium
MEASUREMENT ERROR IN JOB EVALUATION AND THE GENDER WAGE GAP
Author(s) -
Chenf ShihNeng,
Orazem Peter F.,
Mattila J. Peter,
Greig Jeffrey J.
Publication year - 1999
Publication title -
economic inquiry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.823
H-Index - 72
eISSN - 1465-7295
pISSN - 0095-2583
DOI - 10.1111/j.1465-7295.1999.tb01424.x
Subject(s) - phone , library science , wage , computer science , law , political science , philosophy , linguistics
Job evaluation is used to establish pay for more than half of the workers in the United States. It is also the keytool used in establishing the extent of pay bias in firm pay systems. However, job evaluations are subject to measurement error that can bias estimates of the magnitude of paydiscrimination. A practical procedure for makingthese corrections is outlined. Using computed,reliabilityratios to adjust for measurement,error in a studyof state government jobs, we find that measurement errors exaggerate the implied extent of discrimination against predominantly,female jobs by 34% to 44%. Measur ement errors also exagger ate the number,of independent job factors which affect pay. 1 Wage equations are widelyused to estimate the size of the pay gap between male- and female-dominatedjobs. When jobs are the unit of observation, job wagerates are regressed on job attributes (such as skill, effort, responsibility, and working conditions) and on the percent of female incumbents inthe job. Thecoef ficient on theperc ent female variableestim atesthe size of the gender paygap, which is often taken as a measureof pay discrimination against women., These estimates frequentlybecome,the metric for upward,adjustment of female-dominated job pay to redress the discrimination. Summarizing,the research which uses jobs as the unit of observation, Ehrenberg (1989, p. 93) concluded that female-dominatedjobs are underpaid by 15% to 34% relative to male-dominated jobs. Similar findings have been reported in studies which use individuals instead of jobs as the unit of observation. Manycriticis ms have been raised of thesetwo types of studies. One common,criticism is that not all relevant control variables for job attributes or individual productivitymay,be measured,(or even measurable)so that specification error arises. I f these omitted variables are correlated withthe gender variable, then the estimated gender pay gap incorporateslegitima te differences in productivityand/or job attributes. However, few studies have analyzed the impact