Does family intervention work for delinquents?: Results of a meta-analysis.

• Author: Dowden, Craig

Introduction

The contributing role of poor family functioning to the aetiology of juvenile delinquency was documented in some of the earliest, seminal studies reported in the broader criminological literature (Glueck and Glueck 1950; Hawkins, Herrenkohl, Farrington, Brewer, Catalano, and Harachi 1998; Hirschi 1969), a contribution that has been maintained in several recent meta-analyses (Gendreau, Little, and Goggin 1996; Lipsey and Derzon 1998). Given the strong influence of various family factors on the development and maintenance of youth crime, it is not surprising that family interventions are among the most frequently evaluated forms of correctional treatment (Andrews and Bonta 1998).

In light of the popularity of this form of correctional intervention, several meta-analysts have explored the effectiveness of this type of program in their reviews of the correctional treatment literature. Despite the relatively small number of family intervention studies that generally contributed to these reviews, each review provides strong supporting evidence regarding the ability of such treatment approaches to reduce recidivism (Dowden and Andrews 1999a; Garrett 1985; Lipsey and Wilson 1998; Roberts and Camasso 1991).

However, a recent meta-analysis by Latimer (2001) questioned the reliability of this conclusion. More specifically, Latimer argued that perhaps the methodological design of the evaluation study might affect the size of the program effects. The degree of methodological rigour is important to explore because studies that do not employ rigorous designs may be open to criticism on the grounds that their effects are due to some variable other than program participation.

To test this possibility, Latimer (2001) measured the degree of methodological rigour applied in the program evaluation and examined its impact on the reported effects. He selected four variables that he believed provided an indication of enhanced methodological rigour, including random assignment, sample size (n > 100), follow-up periods longer than 12 months, and the use of an external evaluator to conduct the program evaluation. A simple "count" procedure was employed, with methodological rigour scores ranging from 0 to 4, higher scores representing more rigorous research designs. The results revealed that under the most stringent methodological conditions, the positive effects of family interventions disappeared.

Although this finding seems seriously to call into question the therapeutic impact of family intervention programs, Latimer (2001) noted that no effort was made in his study to separate the programs in any way. This was a shortcoming, since a sizeable literature has documented the importance of the principles of risk, need, and general responsivity in delivering effective correctional treatment to offender populations. Therefore, Latimer suggested that future meta-analytic work should explore the influence of methodological rigour on correctional program effectiveness while attending to these principles. This was the purpose of the present paper.

What are the principles of risk, need, and general responsivity?

The principles of risk, need, and responsivity answer questions relating to the who, the what, and the how of correctional intervention, and their clinical relevance has been demonstrated across many different samples, including young, violent, and female offenders (Dowden and Andrews 1999a; 1999b; 2000). Although these principles have been well described elsewhere (Andrews 1995; Andrews and Bonta 1998; Andrews, Bonta, and Hoge 1990), a brief overview of these principles will be provided below.

The risk principle holds that the level of treatment services must be appropriately matched to the risk level of the offender. More specifically, higher-risk offenders should receive more intensive and extensive services, whereas lower-risk clients should receive minimal or no intervention.

Conversely, the need principle focuses specifically on offender needs and classifies them under two separate categories. Criminogenic needs are defined as dynamic risk factors, changes in which are associated with reduced levels of criminal activity (Andrews and Bonta 1998; Andrews, Bonta, and Hoge 1990; Andrews, Zinger, Hoge, Bonta, Gendreau, and Cullen 1990). Non-criminogenic needs are also dynamic, but changes in these particular need areas are not associated with subsequent reductions in criminal activity. The need principle identifies the types of treatment targets that should be selected within correctional interventions. More specifically, in order effectively to reduce recidivism, criminogenic needs must be targeted. Furthermore, although targeting non-criminogenic needs of offenders may also be important for reasons other than reducing recidivism, targeting these areas should not be expected to reduce reoffending.

Two of the criminogenic needs outlined by Andrews and Bonta (1998) focused on different dimensions of family functioning. First, Andrews and Bonta (1998) hypothesized that youths raised in families that used poor monitoring and supervision practices were more likely to engage in delinquent activity. Gendreau et al. (1996) provided empirical evidence for this normative dimension within their meta-analysis of the predictors of adult criminal recidivism. Of the three family structure variables that were measured, they found that child-rearing practices was the strongest correlate of recidivism within this group.

The relationship dimension, on the other hand, is concerned with the ways that family members interact. More specifically, appropriate parent-child interactions, based on open, honest communication and warm expressions of affection, are seen as key elements of appropriate family functioning. Consequently, youths raised in families characterized by inappropriate forms of communication and weak emotional bonds are more prone to developing delinquent behaviour. Meta-analytic evidence presented by Simourd and Andrews (1994), in their examination of the prediction literature for young offenders, provided strong support for the effectiveness of this particular family-functioning variable. More specifically, "having poor parent-child relations" was the third strongest predictor of youthful offending of all of the risk factors surveyed.

Finally, the general responsivity principle is predominantly concerned with the style of program delivery. It states that in order for correctional treatment to be effective, it is of paramount importance for the styles and modes of service to be matched to the learning styles of the clients (Andrews and Bonta 1998). The...