Gender Diversity and the Prospects for Progressive Prison Reform

Abstract

For decades, scholars and policymakers have thought that changing the demographics of the prison officer workforce—in particular, including more female officers—might facilitate progressive prison reform. This article explores the assumption that female officers are more empathetic and pro-rehabilitation and less punitive than their male counterparts by analyzing the relationship between gender and prison officer attitudes toward inmates and the purpose of imprisonment. Drawing on data from more than 900 prison officers in Minnesota, our findings suggest that although female and male officers may import different attitudes toward punishment and rehabilitation, organizational and cultural factors affect the officers' perspectives more than dispositional characteristics. Because prison officers are chiefly responsible for implementing policy on the ground, prison reform efforts require isolating factors that shape these workers' orientations.

KEYWORDS:

• gender
• Minnesota
• officers
• prison
• reform

Notes

*Significant difference from female officers (or the female prison) mean at p < .05.

Note: Ordinal logistic regression (ordinal logistic models [OLM]) is used in Models 1 and 2; linear regression (ordinary least squares [OLS]) is used in Model 3. Female prison is included as a dichotomous variable (1 = female prison, 0 = male prison).

*p < .05, **p < .01, ***p < .001.

Note: Ordinal logistic regression (ordinal logistic models [OLM]) is used in Models 1 and 2; linear regression (ordinary least squares [OLS]) is used in Model 3. All adult state prisons are included, with MCF-Shakopee (the female facility) as the reference. MCF = Minnesota Correctional Facility.

*p < .05, **p < .01, ***p < .001.

Note: MCOS = Minnesota Correctional Officer Survey; MDOC = Minnesota Department of Corrections; MCF = Minnesota Correctional Facility.

^a MCOS does not equal n because of respondents' ability to mark multiple categories.

Historically speaking, prison officers have resented treatment staff because they feel that state officials, prison authorities, the press, and the public celebrate social workers and other workers charged with rehabilitating inmates while disparaging officers (Irwin 1980; Jacobs 1977; Page 2011; Rothman 1980).

The survey questionnaire was adopted from a survey previously administered in California. The California Correctional Officer Survey (CCOS) was administered in 2006 to more than 5,000 prison officers working in the California adult state prison system (Lerman 2008). The CCOS was an initial large-scale effort to obtain information on prison officer attitudes on job satisfaction, work-related stress, personal safety and security; attitudes toward inmates; and profession orientation. In an effort to provide comparisons by state, the current survey replicated the CCOS with slight modifications tailored to the Minnesota Department of Corrections.

To facilitate this study, we collaborated with the labor union that represents correctional officers in Minnesota state prisons, the American Federal of State and Municipal Employees. We decided to work with the union rather than the Department of Corrections because correctional officers typically distrust prison managers and state officials; we anticipated, therefore, that officers would not participate in the study if they received the survey from their bosses. We reasoned that the officers would trust their union more than the Department of Corrections. The union maintains a roster of all correctional officers employed in Minnesota. Using this roster, the union addressed envelopes containing the 2007 MCOS; a cover letter; and prepaid, addressed return envelope. The cover letter described the purpose of the survey, explained that we were not affiliated with the Minnesota Department of Corrections or the American Federal of State and Municipal Employees, and guaranteed the officers' anonymity. We collected boxes of surveys from the union (without looking at individual surveys to ensure officers' anonymity) and mailed them from the University of Minnesota. Respondents mailed completed questionnaires to the Department of Sociology at the University of Minnesota.

One of the most important assumptions underlying the use of ordinal logistic models is the parallel regression assumption. This assumption requires that the explanatory variable (i.e., perceptions of punishment) have a parallel effect across the different cutpoints, or response categories, of the dependent variable. Rejection of the assumption of parallelism implies that at least some of the variables have a differential effect across outcome levels (Long and Freese 2006). In order to examine the parallel regression assumption we computed an approximate likelihood ratio (LR) test for our dependent variables. The LR test indicated that the parallel regression assumption was violated for our first dependent variable, punishment, in the first model, χ²(17, n = 819) = 72.16; but not the second χ²(33, n = 832) = 41.16. The first model was run using linear regression, and there were no significant differences in the results. The LR test indicated that the parallel regression assumption was not violated for our second dependent variable, rehabilitation, in either the first model, χ²(24, n = 824) = 31.87; or the second, χ²(44, n = 837) = 49.85.

We also ran analyses to look for interactions with gender in all models; however, our results did not provide any significant findings.