The effect of prior incarceration on trust and social support: results from two factorial vignette studies

Abstract

Social capital can help formerly incarcerated individuals navigate the challenges of life after prison. Yet, these individuals are unlikely to receive the trust from others that is necessary to build the relationships from which social capital and social support flow. To date, little research has examined individuals’ willingness to extend trust and provide social support to those who are reentering society. To help fill this void, two independent factorial vignette studies were conducted using university-based samples. Study 1 tested the effect of prior incarceration on participants’ willingness to trust. Study 2 examined whether participants provided more social support to recently incarcerated kin as opposed to recently incarcerated friends. Findings from both studies highlight some of the disadvantages formerly incarcerated individuals face when reentering society.

Keywords:

• reintegration
• trust
• recidivism
• desistance
• vignettes

Acknowledgement

The authors would like to thank Michaela Flippin and Ryan Mays for their assistance in the data collection process. This article developed out of Raven Simonds’ thesis project.

Notes

1 The hypothetical scenarios for this study are available from the lead author upon request.

2 Participants were administered four incarceration conditions. That is, the student in the scenario had either been incarcerated one year ago, three years ago, seven years ago, or not at all (baseline condition). Based on preliminary analyses demonstrating that differences across the three incarceration conditions were very modest in terms of trust, the decision was made to collapse these categories in the analysis to make the findings more parsimonious.

3 McEvily and Tortoriello (2011) found that trust was operationalized more than 120 ways. After completing their review of the literature, the authors recommend future scholars pull from validated constructs when appropriate, as well as tailor the trust measures to the unique purposes of the study as a way to acknowledge that “the precise form of trust will vary depending on the empirical setting” (p. 41).

4 When considering the relatively low levels of trust participants expressed concerning their personal property, it is important to note that this university had the highest rates of property crime and larceny thefts among college campuses with at least 10,000 students during the year prior to the study (Insurify Insights, 2018). While it is not possible to directly gauge the impact that the local crime had on survey respondents’ judgements about the vulnerability of their property, it is likely that the effect was nontrivial.

5 To ensure that the observed differences were robust, an ANCOVA model was estimated that include three binary-coded covariates: prior police contact, Hispanic, and racial minority (non-Hispanic whites served as the reference category). The results from the ANCOVA showed that the inclusion of these variables did not influence mean differences between experimental conditions.

6 The hypothetical scenario used in this study is available from the lead author upon request.

7 Cross tabulations and chi-square tests were conducted to ensure that the experimental conditions were approximately evenly distributed across demographic characteristics (i.e., gender, race/ethnicity, age, prior victimization). None of these tests were statistically significant, providing evidence that the experimental manipulations and participant characteristics were appropriately balanced.

8 The dependent variable was disaggregated into its three constituent elements and the effect of the interaction term was assessed across models for instrumental, informational, and emotional support. In terms of sign and significance, the effect of the interaction term persisted.