Measuring time at work: are self-reports accurate?

Jerry A. Jacobs is professor of sociology in the Department of Sociology, University of Pennsylvania. This research was supported by a grant from the Sloan Foundation. Are Americans working more than they have in decades? Does the answer depend upon how one measures an individual’s time at work? Since the publication of Juliet Schor’s best-selling book, The Overworked American, questions regarding the amount of time workers devote to their jobs have received considerable attention. Interest in the length of the workweek is related to important shifts in the demography of the labor force. In particular, the rise of dual earner families has left many individuals feeling pressed for time. The “time famine” faced by working parents has generated much research and public discussion. John P. Robinson and his colleagues have carefully collected and analyzed time diary data from nationally representative samples of respondents since the 1960s. This research effort has produced many interesting and important findings regarding how American’s use their time. Based on their analyses of time diary data, Robinson and Ann Bostrom raised questions about the accuracy of the standard self-reported measure of working time. They suggest that respondents who claim to work long hours exaggerate the amount of time they spend on the job, compared with the time-diary measure. This finding challenges claims that have been made about trends in the time Americans spend on the job. For example, Philip L. Rones and others conclude that the proportion of Americans who work more than 50 hours per week has increased since 1970, based on analysis of data from the Current Population Survey (CPS). If selfreported working time is exaggerated, then this conclusion becomes suspect. Claims of a general increase in working time would also be called into question if workers exaggerate their time on the job. Measures of working time are also important because they are instrumental in computing hourly wage rates. Due to the fact that many of the individuals reporting long hours on the job are workers with college degrees who earn high incomes, exaggeration of work hours among this group would increase estimates of the extent of inequality in the labor market. This article examines the accuracy of self-reported measures of working time raised by John Robinson and Ann Bostrom in four ways. First, it re-examines the issue of time inflation by offering a new interpretation of Robinson and Bostrom’s results. Second, it investigates a new measure of the workweek, derived from departure and return times, and compares it to the results obtained with conventional self-reports. Third, it searches for factors that might produce bias and error in self-reports. And, fourth, this article considers the effect of differences across reference periods (last week versus last year) based on data from the CPS.