Prisoners’ Coping Skills and Involvement in Serious Prison Misconduct

Abstract

Prison misconduct generates serious problems and costs in prisons across the United States. This study examined whether prisoners’ ways of coping affected their involvement in serious prison misconduct. The study also examined the traditional predictors of serious prison misconduct and their relationship to coping. The research included self-report surveys administered to a stratified random sample of 312 prisoners in medium- and maximum-security facilities as well as data extracted from the prison system’s database. Three of the eight ways of coping studied were directly related to serious misconduct along with several of the traditional predictors.

• prison misconduct
• coping
• support
• venting

Notes

1. This study examined “serious misconduct and violence,” which is defined in the section on measures. Though it is also referred to as “misbehavior,” “d-reports,” prison “offenses,” and prison “infractions,” it all refers to very serious misbehavior as defined and never includes less serious misbehavior. References to “serious misbehavior and violence” refer to misbehavior that is either violent and/or considered to be very serious.

2. During the 1970s and 1980s prison misconduct research was conducted within a larger debate about inmate behavior in general and many researchers adopted either the “importation” or “deprivation” model frameworks. The deprivation model (Sykes, 1958) was not applied to empirical research of prison misconduct until the late 1970s, when researchers began studying overcrowding and using the model to begin focusing on prison-level characteristics. Irwin (1970) blamed prison violence and misconduct on prisoners themselves, alleging that they “imported” their bad behavior into prisons. Thus the debate is whether offenders introduce their tendencies toward violence and misconduct into prison or whether the harsh conditions of confinement result in the use of violence to stay safe in the prison. Since penologists began integrating prison-level factors and prisoner characteristics into their studies of prison misconduct (Gaes & McGuire, 1985; Wright, 1991), there has been less of a focus on the importation/deprivation framework.

3. (Rhode Island Department of Corrections, 2009).

4. Since the data were stratified into two categories and one of the categories (those involved in serious disciplinary misconduct in 2008) was oversampled, the dataset needed to be weighted to make up for the oversampling. All of the multivariate analysis was conducted and reported on the weighted data.

5. Oversampling is appropriate when a study is focusing on a group of people that makes up a very small proportion of a population (Toch et al., 1989). The oversampling deliberately samples a much higher proportion of this rare group than the rest of the population. This is done to obtain reasonable estimates about the characteristics of the group. In this case, prisoners with serious d-reports make up a small proportion of the prisoner population, so oversampling was warranted.

6. It was anticipated at the beginning of the study that there would be a 40% refusal rate. However, the actual refusal rate was 30% (as calculated using the American Association for Public Opinion Research calculator for calculating response rates).

7. In addition to the proposed instruments, others include (1) Ways of Coping (Folkman & Lazarus, 1985); (2) Construction Thinking Inventory (Epstein & Meier, 1989); (3) Coping Strategies Inventory (Tobin, Holroyd, Reynolds, & Wigal, 1989).

8. Carver and colleagues (1989) tested the internal consistency of the 13-subscale original version of the COPE with two samples of college students. The Cronbach’s alpha reliability coefficients ranged widely from .45 to .92.

9. (Carver, n.d.).

10. Five of the coping scales—acceptance (Alpha = .63), turning to religion (Alpha = .91), focus on and venting of emotions (Alpha = .64), denial (Alpha = .78), and use of humor (Alpha = .84)—mimic Carver et al.’s COPE subscales (1989) and each contained four items. The next scale, getting social support (Alpha = .83), contained the eight items in two of Carver’s COPE subscales, emotional support and instrumental support. The items on the behavioral and mental disengagement (Alpha = .74) scale loaded onto two separate factors, five on one and three on another. Two of them (watching television and daydreaming) were added to the five because they strengthened that scale. The other (“I turn to work or other substitute activities to take my mind off things”) weakened the scale and was thus eliminated from it. The final coping scale, active coping (Alpha = .70), was a 19-item factor made up of all but one of the items in five of Carver’s subscales (active coping, planning, suppression of competing activities, restraint coping, and positive reinterpretation and growth).

11. If more than half of the items of any subscale were missing for an individual, then the subscale was made missing for that individual. Only five individuals did not answer most or any of the coping questions and thus their coping scales were made missing. However, for the remaining cases that had any missing values, most of the scales were only missing one individual item with a few missing two if they were four-item scales and a few missing up to three if they were made up of seven to nineteen items. Imputation was conducted for those cases with little missing data by taking the average of those items of a scale that were not missing for each case. Thus for the denial scale, if the item “I refused to believe that it has happened” was missing for prisoner X, the average of his remaining three items would be calculated and that number imputed for the missing item. Therefore if prisoner X scored low on the other three items, his score for this missing item would also be low.

12. Class 1 Highest Predatory is considered the most serious and was multiplied by a factor of 6; Class 1 Highest Nonpredatory, the next most serious, was multiplied by a factor of 5; and Class 2 High was considered not nearly as serious, so it was multiplied by a factor of 3. The number of RIDOC-designated “violent” offenses against staff was multiplied by a factor of 3 because these offenses are considered the most egregious acts. These included “violent” offenses in the Class I category (ranging from assault with a dangerous weapon on staff in Predatory to threatening a staff member in the Nonpredatory category). Finally the number of other violent offenses from the top four out of five categories was added (ranging from rioting in Class 1 Predatory to making threatening gestures in the Class 3 Moderate category).

13. The Poisson model has two additional assumptions—that there can be no overdispersion or underdispersion and that predicted counts of the dependent variable are independent of each other (Osgood, 2000). In this study the variance was greater than the mean of the dependent variable—the annual misconduct score (variance = 254.32; mean = 5.59) violating the first assumption. Misconduct data also violated the second assumption that instances of the dependent variable, in this case events of serious misconduct and violence, are independent. In reality, acts of misconduct often occur in clusters, where acting out elicits an official response from staff or other prisoners, which in many cases triggers further responses by the prisoner.

14. The likelihood ratio chi-square examines the “difference in likelihood values between the researcher’s model and the model with the intercept only (null model). If significant, the researcher concludes that the coefficients in the model are different from 0 and the model is accepted” (Garson, 2009, p. 21) or technically rejects the null hypothesis that there is no difference between the models.

15. For mediation to occur there would need to be several statistically significant relationships: between the personal predictors and misconduct; between the personal predictors and coping; and between coping and misconduct. In addition, in the full model, the relationship between the personal predictors and misconduct would be decreased if there was partial mediation or disappear if there was full mediation (Baron & Kenny, 1986).

16. http://www.gallup.com/poll/12091/tracking-religious-affiliation-state-state.aspx