“Prison Will Either Make Ya or Break Ya”: Punishment, Deterrence, and the Criminal Lifestyle

Abstract

Although a prison sentence is often considered to be among the worst punishments that the state can provide, previous research indicates that offenders do not necessarily share this view. Some inmates, for example, adjust to prison life with relative ease, do not view their time in prison as severe punishment, and may even prefer prison to alternative sanctions such as boot camp or probation. To help explain such views, we point to the utility of a “criminal lifestyle” perspective. We argue that offenders who are committed to the values of the criminal subculture tend to view prison in a unique way. For various reasons, such offenders are less likely than others to view imprisonment as difficult or severe and they are less likely to be deterred by prison. Drawing on data from a large inmate survey, we find initial empirical support for these arguments. Implications for deterrence and future research are discussed.

Acknowledgments

This article was presented at the 2011 meeting of the American Society of Criminology, Washington, DC. We thank Mark Reed, Sue Collins, Volkan Topalli, and David May for their helpful comments and suggestions during the early stages of our research. We also thank several anonymous reviewers for their comments.

Notes

¹Although early studies of perceptual deterrence raised doubts about the role of sanction severity in the deterrence process, it appears that such doubts were partly a function of study methodology. Based on a comprehensive review of perceptual deterrence studies using different methodologies, Nagin (1998) concludes that the collective evidence favors the existence of sanction effects that are consistent with deterrence. On average, people respond to the perceived severity of punishment as well as the perceived certainty of punishment. Recent research has increasingly focused on the conditions under which these sanction effects operate with greater or lesser force (e.g., Loughran et al. 2011; Piquero et al. 2011).

As a result of listwise deletion, the number of missing cases for each item range from 11 to 15.

²A small number of respondents (generally no more than 1% of the sample) failed to follow instructions and circled both “agree” and “disagree” for some of the questionnaire items. For all such items, we classified these responses as missing values. We obtain the same pattern of results whether or not these values are treated as valid.

³To ensure that the study findings are not dependent on our selected coding procedure, we conducted sensitivity analyses (not shown) by experimenting with alternative coding schemes. For example, we counted simultaneous “agree” and “disagree” responses as legitimate values (see note 1), constructed an additive scale by summing the two prison severity items, and conducted analyses based on ordinary least squares regression. These analyses produced an essentially identical pattern of findings (complete set of results available from the authors on request).

⁴Technically, these dichotomous items violate the assumption of continuous distributions required by ordinary factor analysis. However, Percy (1976) presents data showing that ordinary factor analysis can be used effectively with such dichotomous variables.

⁵For dichotomous scale items, alpha is equivalent to the Kudar-Richardson 20 (KR20) coefficient. Although the reliability coefficient for this scale is rather low, it should be noted that the alpha coefficient is a lower bound for true reliability. In addition, the low alpha may be a function of the small number of items in the scale (Carmines and Zellner 1979). Further, the most likely consequence of low reliability is that it is more difficult to detect significant effects, resulting in a conservative test of the hypotheses. As described later in the article, the Commitment to Crime scale exhibits a consistent and robust pattern of effects in our analyses, suggesting that the reliability of the scale is not a serious problem in this context.

⁶To confirm that perceived prison severity is empirically distinct from the respondent's commitment to the criminal lifestyle, we submitted the original prison severity questionnaire items to factor analysis along with the criminal lifestyle items (again using principal components extraction and a varimax method of rotation). The prison severity items loaded on their own separate factor (results available from the authors on request).

⁷A follow-up questionnaire item asked the respondent if s/he followed instructions and included his or her current term in the final count of prison terms served. For respondents who stated they had failed to include the current term, we added a value of 1 to their total number of prison terms. Ten respondents confirmed that they had included their current term, but nonetheless reported zero prison terms served. In such cases, we classified zero as a missing value. We obtained the same pattern of results whether or not such cases were treated as missing values.

*p < .05.

⁸To obtain the percent change in odds, we subtract 1 from the odds ratio and multiply by 100: (2.05 − 1) * 100 = 105.

*p < .05.

Additional information

Notes on contributors

Beverly R. Crank

BEVERLY R. CRANK is a doctoral candidate in the Department of Criminal Justice and Criminology at the Andrew Young School of Policy Studies, Georgia State University. Beverly's research areas include juvenile delinquency, offender decision-making, and punishment and deterrence.

Timothy Brezina

TIMOTHY BREZINA is Associate Professor of Criminal Justice and Criminology at Georgia State University. His research interests include criminological theory, youth violence, and punishment and deterrence. Recent publications appear in the journals Criminology, Criminal Justice and Behavior, and Theoretical Criminology. He is co-author of Juvenile Delinquency: Causes and Control, 4th edition (with Robert Agnew), Oxford University Press.