Association for Public Policy Analysis & Management

The notion that prison visits may sustain or increase social bonds, which ultimately translate to improved inmate outcomes, has led to renewed interest in studying prison visitation. In most examinations of the effect of visits on inmate behavior both during incarceration and post-release the “treatment” of prison visits is considered as being either present or absent or being a count of the number of visits an inmate receives; however, given that most prison sentences span a number of years it would stand to reason that the effect of prison visits might be related to variations in  “dosage” over time. The goal of this paper is to expand beyond binary or count measures of visits to explore a key feature of prison visit heterogeneity: the pattern of visits over time.

Two recent papers used semiparametric group-based trajectory modeling with binary logit model specification to explore visit-by-month trajectories (Cochran, 2012; Cochran, 2014). Using slightly different samples from the Florida Department of Corrections, spanning one year (2012) and between 8 and 17 months (2014), both papers identified a four-group model as the best fit: no visits, near entry visits, near release visits, and sustained visits. We’re interested in exploring the degree to which these patterns appear in another jurisdiction, across a longer time span, and with even more varied lengths of stay.

To that end, this paper builds on that work by examining longitudinal patterns of visitation in a large sample of inmates receiving visits between 1995 and 2013 in New York state prisons. The data, provided by New York State Department of Corrections and Community Supervision (DOCCS), include information on inmate individual characteristics, criminal history, length of stay, and visitation history. Results will demonstrate the extent to which previously observed patterns are replicated and whether new descriptive patterns are observed. In addition to presenting trajectories of visitation for inmates over time, we will address implications for methodology and policy.