Online Program

The Impact of a Person Level Incentive on Establishment Level Response and Prevalence Rates within Correctional Facilities
Marcus Berzofsky, RTI International 
*Rachel Caspar, RTI International 
Chris Krebs, RTI International 

Keywords: Incentives, National Inmate Survey, response rates, multilevel model, prisons, jails, correctional facilities

Researchers often use incentives to increase response rates while, hopefully, not biasing the estimates for a survey’s outcomes of interest. The National Inmate Survey (NIS) is a congressionally mandated nationally representative, probability-based survey sponsored by the U.S. Bureau of Justice Statistics about the sexual victimization of inmates within correctional facilities (prisons and jails). During the 2007–08 NIS inmates were offered a non-monetary incentive (a bag of cookies); however, not all facilities allowed inmates to receive incentives. This paper will analyze the impact of an incentive on response rates and prevalence rates, controlling for facility-level and inmate-level characteristics. Separately for prisons and jails, we will conduct bivariate analyses to determine which characteristics significantly increase the response rate or make a significant difference in the prevalence rate. Next, for each facility type, we will fit two multi-level models to determine the impact that the incentive had on an inmate’s 1) propensity to participate in the survey, and 2) propensity to report being a victim while controlling for facility-level and inmate-level characteristics.