Essays on Volunteering

Thumbnail Image


umi-umd-3755.pdf (1.12 MB)
No. of downloads: 2706

Publication or External Link






Volunteer activity is an important part of the lives of Americans. This dissertation uses economic analysis to study volunteering. The first essay examines the impact of mandated service on public school students in Maryland. Proponents of mandates note that individual volunteer activity is correlated over time, and therefore argue that mandates will create lifetime volunteers. Prior studies demonstrate that the observed characteristics of volunteers are different from nonvolunteers. Thus, it is possible that unobserved characteristics drive the correlation in service over time and the policy will not increase future service. Using restricted-access data from the Monitoring the Future project, I find mandates increased volunteering among eighth-grade students. However, by the twelfth grade, I find the law had at best no impact on volunteer activity, and in some specifications it reduced volunteering. In contrast to creating lifelong volunteers, my results suggest that the mandate changed the timing of volunteering, but did not alter overall volunteering among affected students.

The second essay examines the impact of survey nonresponse on inferences about volunteer behavior. Time use diaries are a key source of data on volunteering, though they typically have a high nonresponse rate. Since participation in surveys and volunteering are likely influenced by the same qualities, nonresponse bias may distort estimates of volunteering. A random subsample of individuals appears in both the Current Population Survey (CPS) September Volunteer Supplement and the American Time Use Survey (ATUS). As such, we can compare the reported volunteering (as found in the CPS) for ATUS respondents to that of nonrespondents in order to uncover the impact of unobservable differences. We find higher levels of volunteer activity among ATUS survey respondents than nonrespondents, differences that persist across narrowly-defined demographic groups. Using regression analysis, with annual hours spent volunteering as the dependent variable, we control for the observable characteristics available in the data and compare the results found using the full sample to the respondents-only sample. Although the signs on the coefficient estimates are generally consistent across the samples, the size of the estimates varies in magnitude, indicating that nonresponse bias continues to exist.