The Importance of Ecological Context for Correctional Rehabilitation Programs: Understanding the Micro- and Macro-Level Dimensions of Successful Offender Treatment

Abstract

The resurgence of support for offender rehabilitation has led to an increased emphasis on correctional program integrity. Treatment programs are now being evaluated and tailored in accordance with the principles of effective intervention, which are rooted primarily in methods of individual behavior modification through a social learning approach. The problem with this exclusive focus on the individual is that it fails to recognize the importance of certain ecological factors that have been shown to be significant predictors of recidivism. The purpose of the current research, therefore, is to examine the impact of structural characteristics on both treatment program quality and effectiveness of halfway house programs in Ohio. Our results reveal that ecological context influences the magnitude of program treatment effects largely indirectly though its influence on program quality. The implications of these findings for correctional theory and practice are discussed.

Keywords:

• ecological context
• offender rehabilitation
• recidivism

Notes

1. It is important to note that, in agreement with the original arguments set forth by Gottfredson and Taylor, nearly all researchers studying these phenomena acknowledge the interaction between the macro- and micro-levels. Nevertheless, their research questions are generally focused on only one of the two directions identified here (cf. Hipp & Yates, 2009).

2. Later, Andrews, Bonta, and Hoge (1990) would add to these the importance of professional override in exercising discretion upon considerations of risk, need, and responsivity.

3. It is instructive to note that the principles of effective intervention are beginning to creep into the literature on reentry. Travis (2005) provided what he calls the principles of effective reentry in his book on offender reintegration. The five principles are: prepare for reentry, build bridges between prisons and communities, seize the moment of release, strengthen the concentric circles of support, and promote successful reintegration. This presents a similar framework for understanding recidivism within the context of effective programming, but not necessarily one that emphasizes the specific components of individual programs. Similarly, several scholars (e.g. Listwan, Cullen, & Latessa, 2006; Lowenkamp & Latessa, 2005b) have begun to stress the importance of considering evidence-based practice and what works in corrections for reentry programs. While these represent significant advances, they do not specifically consider existing ecological theory.

4. The full version of the CPAI contains 77 items. Due to the limited nature of the data collection, the total number of items scored for each halfway house was reduced to 62. The current analysis uses a reduced form of the CPAI that includes only the 33 items that were found to be significantly and positively correlated with treatment effect (Lowenkamp et al., 2006). Doing so subjects the strongest possible relationship between CPAI and treatment effects to the structural characteristics.

5. Although a sample size of 38 may be small by traditional quantitative standards, it is worthy of note that this number is comparable to prior criminological research in general (e.g. Piquero, Jennings, & Farrington, 2010) and macro-level criminological research in particular (e.g. Messner & Rosenfeld, 1997; Pratt & Godsey, 2003).

6. There has been considerable debate about the appropriate unit of analysis for measuring the impact of macro-level indicators on criminal justice phenomena (Sampson et al., 2002). It is potentially problematic to use county-level data in that unobserved heterogeneity within counties may lead to an ecological fallacy. Nevertheless, the use of county-level structural indicators is consistent with prior criminal justice research in general (Baller, Anselin, Messner, Deane, & Hawkins, 2001; Osgood & Chambers, 2000) as well as with recidivism research specifically (Mears et al., 2008; Reisig et al., 2007). Further, prior research has indicated that the structural correlates of crime are generally robust across city, county, and state levels of aggregation (Land, McCall, & Cohen, 1990; Pratt & Cullen, 2005). Finally, research indicates that when ex-offenders move they typically stay within the same county (LaVigne & Parthasarathy, 2005). Thus, using counties as a unit of analysis allows us to limit the potential bias resulting from relocation (see also Reisig et al., 2007).

7. Scales based on the summation of equally weighted z-scores divided by the number of items in each scale produced similar results.

8. The data made available to us limit our ability to determine how many offenders were convicted outside of the county the HWH was located in; yet we have every reason to believe that this number represents a small subset of offenders. Consistent with prior ecological and recidivism research, and our arguments presented in the current analysis, it is likely that offenders are concentrated in a small number of especially disadvantaged areas (Clear, 2007)—precisely the areas in which these HWH are located in.

9. Calculating effect sizes using successful terminations only produced treatment effects that favor the treatment group in 23 out of 38 programs (61%). Given that the treatment group is likely under stricter surveillance than the control group, it could be expected that they are more likely to “fail” (i.e. be terminated due to a technical violation or new offense) than the control group in the community. Using successful terminations only accounts for this bias and produces treatment effects in the expected direction (favoring the treatment group). Data on the reason for return to prison were not available and therefore this explanation could not be tested directly.

10. A notable omission from the current construct is a measure of unemployment. Principal components analysis revealed that percent unemployed loaded poorly (i.e. factor loading < .50) on the concentrated disadvantage construct. This is consistent with prior research suggesting that unemployment is conceptually distinct and empirically independent from other indicators of concentrated disadvantage (Land et al., 1990). Given the significance of economic conditions for the prediction of criminal behavior in general (Chiricos, 1987), and for reentry in particular (Hannon & DeFina, 2010), we retain independent measures of the change in the percentage of individuals unemployed in our analyses.

11. Massey (1996) recommended defining affluent as families whose income is four times the poverty level (which equates to $54,000 in 1990). Due to data availability, affluent in the current analysis is defined as $50,000 and above for 1990 and $75,000 and above for 2000).

12. The weighting procedure for the current analysis is described more fully in Lowenkamp (2004). Program effectiveness scores were calculated assessing the difference in re-incarceration rates (for any reason, technical violation or new criminal offense) between HWH participants and a matched comparison group over a 2 year period. An effect size was then developed for each halfway house, and weights were calculated by determining the inverse of the effect size variance (see Lipsey & Wilson, 2001, pp. 36–37, 54). Thus, we give more weight to programs that we can be more confident have produced a treatment effect that represents the true value (weights based solely on program size produced similar results).

*Significant difference at p < .05.