Did Mass Incarceration Leave Americans Feeling Less Afraid? A Multilevel Analysis of Cumulative Imprisonment and Individual Perceptions of Fear

Abstract

Much of the political rhetoric that facilitated mass incarceration was predicated on the promise of reducing fear among the public. Yet, it remains unclear whether the large increases in imprisonment experienced in many areas made residents feel less afraid. We examine this issue by integrating geographic data on imprisonment with individual-level data on fear from the General Social Survey (GSS). We find that people from states and counties with greater “cumulative imprisonment” rates were no less afraid than their counterparts from areas that imprisoned many fewer people. These findings hold for the public overall and for non-Latino whites and members of the working and middle classes, who frequently were target audiences for political rhetoric linking mass incarceration era policies to fear reduction. Our study supports growing calls to decouple crime and criminal justice policy from politics and electoral cycles, and to develop evidence-based punishment approaches organized around transparent normative principles.

Keywords:

• Mass incarceration
• fear
• criminal justice policy
• GSS

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1 This measure diverges from indicators of persons housed in prisons within a given state or county. The latter are useful for advancing understanding of the factors related to prison siting (e.g. Eason, 2017; Hooks et al., 2010), but provide an incomplete portrait of the share of the state/county population sentenced and admitted to prison irrespective of where that prison sentence is served.

2 The choice of a ten-year window to measure cumulative imprisonment is arbitrary, but in our view logical for defining a state’s or county’s engagement with mass imprisonment. Using a five-year window to define “cumulative imprisonment” yielded the same conclusions.

3 The log percentage change in crime rates is calculated as 100 log_e (Crime Rate_year/Crime Rate_year-10). This measure of relative change is symmetrical and additive, making it preferred over alternatives that lack these desirable qualities (see Törnqvist et al., 1985). Using a five-year change measure yielded substantively identical results.

4 A total of 22,146 GSS respondents from 47 states were asked about fear between 1985 and 2010. The analysis is limited to 18,010 of these respondents due to missing data on individual-level control variables (primarily family income and the presence of other race persons in respondents’ neighborhoods). There were 14,418 GSS respondents asked about fear between 1993 and 2010 (for whom county geographic codes were recorded), 9,676 of whom were sampled from states and counties that consistently reported data on imprisonment in the NCRP. The analysis of county effects is limited to 7,053 of these respondents who had complete data on the other variables included, with most (73%) of the sample loss due to missing data on family income and the presence of other race persons in respondents’ neighborhoods and the remainder due to missing data on UCR county crime rates. Comparing Table 1 with Appendix B in the online supplement reveals that the analysis samples closely mirror the maximum potential GSS samples that could be used for the study. This suggests that any bias due to missing data is minimal.

5 As shown in Appendix C of the online supplement, the state (Panel A) and county (Panel B) variables exhibit low to moderate inter-item correlations with one another, which indicates that they capture unique features of the geographic areas in which GSS respondents resided and suggests that the potential for bias due to multicollinearity is minimal.

6 We replicated the models described after adding state fixed effects to account for potentially important unmeasured state conditions and we also expanded our model specification to include indicators of changes in social, demographic, economic, and political conditions that corresponded to the periods captured by our measure of cumulative imprisonment. These additional analyses yielded results (not shown) that parallel those reported in the manuscript.

Additional information

Funding

This work was supported by the Population Research Institute at Pennsylvania State University.

Notes on contributors

Andrea Corradi

Andrea Corradi is a Ph.D. Candidate in the Department of Sociology and Criminology at Pennsylvania State University. Her research areas include incarceration, extremism, security, gender, and the use of social control mechanisms.

Eric P. Baumer

Eric P. Baumer is a professor of sociology and criminology at Pennsylvania State University and a faculty affiliate of the Population Research Institute. His research explores demographic, temporal, and spatial patterns of violence, the mobilization of law, and the application of criminal justice sanctions.