Abstract
Extant literature on prison misconduct focuses primarily on male inmates or pooled samples of male and female inmates. Moreover, this research typically combines several types of misconduct into one global measure. To add to the literature, the current study gathered self-report data from over 300 female inmates to explore the impact of variables from three different theoretical perspectives upon four types of misconduct (i.e., physical assaults, nonviolent sexual activity, drug/alcohol, and property). Results show that importation and social control variables predict all types of misconduct, whereas deprivation variables provided some modest support for sexual misconduct and property misconduct.
FUNDING
Research made possible by Grant #232606, Office of Research and Sponsored Programs, Wright State University.
Notes
1. The variables labeled as informal social control variables in this study are only proxy measures of social control rather than specific, direct measures. These measures could also serve as proxies for other theoretical perspectives, such as strain or self-control. Some of the variables classified as social control variables in this study have also been identified as both deprivation and importation variables in past work on prison misconduct. Many of the variables employed in this study can be and have been explained from all three theoretical perspectives throughout the literature. Employment, marital status, education, and the presence of children are identified as social control variables in this study even though they occur prior to prison (a la importation) in order to expand beyond the traditional scope of just importation and deprivation.
2. Missing data were substituted for with the appropriate measure of central tendency. None of the variables included in the analysis had more than 3% missing data.
3. Bivariate correlations were obtained for all of the independent and dependent variables. None of the correlations were more than .50. Moreover, none of the variance inflation factors were over 4. This indicates that multicollinearity was not an issue.