The Effect of Incarceration on Marriage and Work Over the Life Course

Abstract The current study adopts the life course framework to examine the effect of incarceration on the likelihood of becoming married and attaining full‐time employment. It is hypothesized that men who have been incarcerated will be less likely to marry and to gain full‐time employment. Data from the National Longitudinal Survey of Youth are used to test the hypothesis. Results from the growth‐curve models support the life‐course theoretical model. Across all models estimated, incarceration is negatively associated with marriage and employment. In addition, positive milestones (e.g., education) are associated with improved chances of employment and marriage. The findings reinforce the importance of considering a multitude of life events when estimating life trajectories. Keywords: incarcerationmarriageworkemployment Acknowledgments An earlier version of this article was presented at the annual meeting of the American Society of Criminology, Denver, 2003. The author is grateful to Eric Baumer, Bob Bursik, Timothy Bynum, Christina DeJong, Ken Frank, and Merry Morash for their assistance on earlier drafts of this article. The author would also like to thank Donna Bishop and the anonymous reviewers for their helpful comments. Notes 1. An appropriate level of sample retention was maintained throughout the study. A retention rate of nearly 90% was sustained for the first 16 waves of the survey. The retention rate dropped to 86% in 1996 and 80% in 2000. Excluding those individuals who have been dropped from the sample, respondents have completed, on average, 17.4 of the 19 interviews. In 2000, 64% of the sample had completed a survey in each of the data collection years (Center for Human Resource Research, Citation2001). 2. Due to changes in study protocol, two large sub‐groups of participants became ineligible for interviews during the course of data collection. As of 1984, 638 male members of the military sample were no longer interviewed. In 1990, 731 males from the non‐black, non‐Hispanic economically disadvantaged group were excluded from the sample. 3. In total, 230 males died during the course of data collection, 7 men were incarcerated during 11 or more time periods, and 206 men participated in less than six interviews. These men were omitted from the analysis group. Although the nested structure of the HLM model facilitates the valid estimation of models when both the spacing and the number of observations vary by individual, the preceding cases were removed from the analyses due to poor data quality. 4. This measure includes men incarcerated as young adults and juveniles. In total, 166 men reported being incarcerated prior to 1980. Of those, 104 (63%) were between the ages of 19 and 23 in 1980. Respondents were not asked to provide a date for their most recent incarceration; therefore, it is impossible to ascertain the true age at first incarceration. This measure simply serves as a control for incarceration that occurred prior to the study period. 5. The measure of cognitive ability used in this study is not optimal. The AFQT was originally constructed to be an assessment of trainability for the armed forces and is currently used as the primary criterion for enlistment eligibility in the United States armed forces. In addition, the use of standardized scores to measure cognitive ability has also been debated (see Neisser et al., Citation1996). Due to limitations in the original data‐collection instrument, a better measure of cognitive ability was not available. 6. The appropriateness of the growth trend was considered in a number of ways (see Raudenbush therefore, each of the life‐course event variables was considered fixed in subsequent models. The log of age was allowed to vary in each of the models. This modeling procedure is consistent with traditional HLM methodology (Raudenbush, Citation2001; Raudenbush & Bryk, Citation2002). 9. The reliability of the coefficients is calculated as a ratio of the true parameter variance to the total observed parameter variance (see Raudenbush & Bryk, Citation2002, p. 166). Reliability of (πpi ) = Var(πpi )/Var(πpi ). Additional informationNotes on contributorsBeth M. Huebner Beth M. Huebner is an assistant professor in the Department of Criminology and Criminal Justice at the University of Missouri‐St. Louis. She received her PhD from Michigan State University. Her research interests include the collateral consequences of imprisonment, life course criminology, and quantitative research methods.