Recidivism and Relationships: Examining the Role of Relationships, Transitions, and Relationship Quality in Reincarceration

Prison Visitation and the Likelihood of Post-Release Employment

“…In order to construct post-stratification weights (see below), the data need to be complete; thus 270 individuals are lost when constructing the strata for the weighting variable. Additionally, 669 individuals are lost due to attrition or loss to follow-up by wave 2 (for a discussion of attrition within the SVORI data and methods to address it, see Wallace et al, 2020; Wallace & Wang, 2020). Finally, 47 individuals are lost due to listwise deletion resulting in our final analytic sample of 711 respondents.…”

“…Since the dependent variable is dichotomous, we utilize logistic regression with post-stratification weights. Post-stratification weighting is an established method of addressing non-response issues in survey research (Kalton & Flores-Cervantes, 2003), which has been previously used to rectify these issues in the SVORI data (see Wallace et al, 2020; Wallace & Wang, 2020). Attrition is concerning in our study given that people in the SVORI data are potentially easier to follow over time, suggesting that observed respondents in the analysis are more stable, which in turn may impact their likelihood of employment.…”

“…2 Post-stratification weighting allows us to weight the respondents in the same way as the initial sample (wave 1 baseline) prior to attrition from the data at wave 2. In this sense, post-stratification weight is necessary to produce unbiased model estimates (Kalton & Flores-Cervantes, 2003; Wallace et al, 2020; Wallace & Wang, 2020), as well as having our analytic sample represent the original sampling structure of SVORI.…”

An Assessment of Prisoner Reentry, Legal Financial Obligations and Family Financial Support: A Focus on Fathers

“…Additionally, we employed post-stratification weighting to address the level of attrition in the SVORI data. Post-stratification is a common method used to remedy issues of non-response in sampling and research [58], and has been previously applied to the SVORI data (see [34,57,59]). In the SVORI data, attrition is problematic since individuals who remain in the study over multiple waves may be qualitatively different from those who are more difficult to follow and attrite from the study.…”

“…Only 60% of the sample was retained between the first and second wave of the project, a significant attrition rate [64]. In line with prior studies employing the SVORI data [57,59], we created post-stratification weights using respondent characteristics. In the current study, prior drug use, race, and age [65,66] were identified as being related to either perceived or objective familial financial support.…”