Dana J. Hubbard
Cleveland State University
Previous Research
History of Sexual Abuse
Methods
Data Analysis
Results
Discussion
IT IS NO SURPRISE that the number of offenders under correctional supervision has been steadily increasing. In fact, this number has increased more than 188 percent since 1973 (Clear, 1994). Moreover, it is likely that on most days, more than 1.8 million Americans are behind bars (Bureau of Justice Statistics, 1998). Women and girls have not been left out of these increases in correctional populations, with data indicating that the rate of imprisonment for women increased twice as much as for men in the 1980s (Immarigeon and Chesney-Lind, 1992; Danner, 1998) and continues to increase each year at a greater rate than that for men (Bureau of Justice Statistics, 1998). Despite these trends, studies indicate that the public tends to support the idea of using the correctional system to rehabilitate offenders (see Cullen, Skovron, Scott, and Burton, 1990; Applegate, Cullen, and Fisher, 1997). Furthermore, there is a vast amount of literature indicating that rehabilitation efforts can “work.” Given the increases in the numbers of people under correctional supervision and the fact that most offenders return to the streets, the need to find ways to effectively reduce recidivism through rehabilitation of offenders is imperative.
While research has demonstrated that programs can reduce recidivism for offenders, not all programs have been shown to be equally effective. There have been several generations of research called the “what works” literature dedicated to improving correctional treatment. Gendreau (1996) and others (see Andrews and Bonta, 1999; Andrews et al., 1990) have put forth "principles of effective intervention" intended to guide programs for offenders in rehabilitation efforts. These principles include such things as using assessment to classifying offenders on their level of risk to recidivate, targeting offenders’ criminogenic needs in treatment, and matching offenders to the appropriate staff and/or type of treatment. This last principle mentioned above is called the responsivity principle and is the least researched of the principles.
There are two types of responsivity according to Andrews and his colleagues (1990), general and specific. General responsivity refers to the idea that treatment programs will be most successful if they utilize behavioral techniques such as role-playing, role-modeling, problem-solving, and graduated reinforcement techniques (Andrews, Bonta, and Hoge, 1990). This is primarily because these techniques incorporate various teaching tools and strategies which are geared to the greatest amount of learning styles. Examples of treatment programs currently being used in corrections that utilize these techniques are those based in cognitive-behavioral theory and/or social learning theory. Meta-analysis and reviews of the treatment literature have found these treatment strategies to be the most effective with offenders (Izzo and Ross, 1990; Andrews et al., 1990; Antonowicz and Ross, 1994).
The other type of responsivity discussed in the literature is referred to as specific responsivity. This is the notion that personal characteristics of offenders may make them more or less amenable to treatment. Thus, for programs to be effective, these responsivity characteristics must be addressed through assessment and through matching offenders to appropriate staff and the right type of treatment for that offender. To date, there is not much research regarding what personal characteristics are most important for programs to consider. There is speculation, however, that factors such as race, gender, age, sexual abuse, depression, self-esteem, and intelligence are factors that are important to consider when providing correctional rehabilitation.
As discussed previously, the responsivity principle is one of the least researched principles. Since many correctional treatment programs are utilizing cognitive-behavioral treatment, it is important to understand what personal characteristics of offenders are related to success in this type of program. Cognitive-behavioral treatment meets the general responsivity principle and therefore in theory should be able to reach the greatest variety of offender types. However, there is some evidence that intelligence matters in whether offenders can understand this type of treatment (Ross and Fabiano, 1985). In addition, other important characteristics of offenders have not been studied to determine whether they are related to success and/or failure in this type of treatment.
This study sought to determine what responsivity characteristics were important in a program that meets the general responsivity principle, cognitive-behavioral treatment. That is, personal characteristics of offenders were assessed and then the subjects were given cognitive-behavioral treatment. Outcome data were then analyzed to determine if various responsivity characteristics were related to whether the program "worked." It is hypothesized that individual characteristics of offenders such as gender, age and others such as low intelligence, low self-esteem, a history of sexual abuse and depression may interfere with the offender’s likelihood of success in the program (program completion) or not recidivating.
Previous ResearchIn the correctional rehabilitation literature, both individual outcome studies and the numerous meta-analyses have demonstrated that cognitive-behavioral strategies are among the most effective treatment approaches for offenders. One of the most researched cognitive-behavioral programs for offenders is the Reasoning and Rehabilitation program (Ross, Fabiano, and Ewles, 1988; Robinson, Grossman, and Porporino, 1991; Robinson, 1995). These studies have all shown consistent effects of this cognitive-behavioral program over control groups. Another cognitive-treatment program that has shown effectiveness is Moral Reconation Therapy (MRT) (Little, 2000). Moreover, the meta-analyses have also consistently shown the effectiveness of cognitive-behavioral programs in reducing offender recidivism (Andrews, Zinger, Hoge, Bonta, Gendreau, and Cullen, 1990; Antonowicz and Ross, 1994; Garrett, 1985; Izzo and Ross, 1990; Lipsey, 1990).
The notion of specific responsivity refers to the idea that individual personal characteristics may make offenders more or less responsive to treatment (Andrews, Bonta, and Hoge, 1990). Although this is not a new idea, the label for the practice of matching treatment to personal characteristics (responsivity) is relatively new. As early as the 1950s Freud warned that psychotherapists should be aware that their highly verbal style of therapy was inappropriate for certain offender types such as those with poor verbal abilities (Freud, 1953). Recently, there has been some research indicating that personal characteristics such as intelligence and personality may mediate the effects of treatment. However, despite one study that found that IQ may affect success in cognitive- behavioral treatment (Ross and Fabiano, 1985), few studies have addressed the issue of responsivity in cognitive-behavioral treatment (VanVoorhis, 1997).
Gender
There is a debate in the literature about how much gender should be considered when developing treatment programs for offenders. Feminists and advocates of gender-specific treatment argue that men and women are qualitatively different in that they develop differently, have different needs, and have different pathways to offending (Bloom and Covington, 2001). As a result, they argue that the programs used to treat them also need to be different. The “what works” researchers claim that gender should be considered a responsivity concern. That is, programs need to be aware of gender differences when delivering the treatment; however, the principles of effective intervention are the same. Nonetheless, both groups would argue that gender is a responsivity consideration.
Depression
Many researchers have suggested that depression could be an important responsivity characteristic (VanVoorhis, 1997; Kennedy and Serin, 1997; Bonta, 1995). How depression is related to success or failure, however, is still not known. While many researchers have mentioned it as a potential responsivity factor, there is little research on its effects. Due to the fact that depression has been linked to self-esteem and sexual abuse, feminists have argued that programs need to consider this mental health need in programming for women (Kearney-Cooke, 2002).
Self-esteem
There has been much discussion about the importance of self-esteem as a predictor of criminal behavior; however, there is little discussion about self-esteem as a responsivity consideration. Some research has indicated that it may have differential effects on criminal behavior depending on personal characteristics. For example, Hubbard (2006) found that self-esteem was positively related to arrest in African Americans and negatively related to arrest in white offenders, regardless of gender. Evidence does exist that narcissistic people are at risk for criminal behavior (Bushman and Baumeister, 1998). Given this, it is possible that at the very least, self-esteem may interfere with treatment success (Bloom, 1998; Bonta, 1995). Those offenders with low self-esteem may not participate in group therapy and may need more skilled staff members. Again, the role of this characteristic is not yet known. However, the gender-specific literature identifies it as an important need for women that should be addressed in treatment.
A history of sexual abuse is recognized as a risk factor for crime in both males and females, yet it is also likely an important responsivity consideration. Women tend to have a greater likelihood of a sexual abuse background than men. The incidence of sexual abuse in female offenders is even higher. A study by the American Correctional Association (1990) found that approximately 36 percent of all female offenders had been sexually abused. Many researchers, however, claim that this statistic is actually much higher due to the amount of underreporting. While the prevalence of sexual abuse is high in women, the effects of this personal characteristic on success in treatment program are not yet known.
Intelligence
Intelligence has also been considered as a potentially important responsivity characteristic. For example, Ross and Fabiano (1985) found that intelligence was related to success in a cognitive-behavioral treatment program. They found that offenders with IQ’s lower than 85 may not be successful in a cognitive program. Offenders with lower intelligence levels may not have the ability to understand cognitive curriculums. For example, many curriculums emphasize learning the difference between thoughts and feelings and learning to act only on thoughts rather than feelings. This might be too hard a concept to grasp for those offenders with low IQ’s. Thus, they may appear to be uninterested or unsuccessful when they cannot understand the material. Despite this one study, little research has addressed the issue of intelligence as a responsivity consideration.
MethodsThe literature has indicated that certain personal characteristics of offenders may be important in moderating the effects of treatment. This study attempted to determine if these characteristics matter in determining an offender’s success or failure in cognitive-behavioral treatment. The research question addressed in this study is: What specific responsivity characteristics affect whether an offender is successful in cognitive-behavioral treatment (a program that meets the general responsivity principle)? That is, what characteristics are related to whether an offender completes the program as well as what characteristics are related to whether an offender is rearrested and/or reincarcerated? In order to answer the above questions, a cognitive-behavioral program was implemented at four treatment sites. Data were collected on offenders at each site and analyzed to determine if these characteristics were related to success in the program.
The cognitive-behavioral program that was implemented at the four treatment programs is Corrective Thinking. The Corrective Thinking curriculum developed by Rogie Spon (1999) is based on the work of Yochelson and Samenow (1976). Instead of thinking errors, however, the curriculum teaches offenders how to recognize their “barriers in thinking” and replace them with the appropriate “correctives.” The curriculum consists of a series of exercises aimed at teaching the offenders the nine barriers in thinking and the nine correctives. The program is generally considered a cognitive “restructuring” program rather than a skill-building program. However, some exercises involve role playing and the practicing of new behaviors. The program is offered to participants in the form of a group. Currently, there are no outcome evaluations on the effectiveness of this particular cognitive-behavioral program. Cognitive-behavioral programming in general, however, has been found to be effective with offender populations (see Ross, Fabiano, and Ewles, 1988; Robinson, Grossman, and Porporino, 1991; Robinson, 1995).
This study was part of a larger study funded by the Ohio Office of Criminal Justice Services. Offenders at four treatment sites were included to increase the sample size. While all offenders in the above treatment programs were required to participate in the cognitive treatment groups in each study site, participation in this study was voluntary. Staff at each study site explained the study to offenders upon intake into each program and offenders were asked to participate.
The offenders were then asked to sign a release if they agreed to participate in the study. It is not clear how many offenders refused to participate.
Offenders in four study sites were included in this sample. The first site was a 20-bed halfway house for women. The offenders were felons under various supervision levels, and a majority were chemically dependent. The treatment program lasted approximately 90 days, over the course of which offenders received approximately 26 hours of Corrective Thinking. In addition to cognitive-behavioral programming, offenders received substance abuse treatment as well as attending other life skills groups, such as parenting. They were also expected to work full-time.
The next study site was a 60-bed halfway house serving adult males. The length of treatment was approximately 90 days, during which time offenders received approximately 30 hours of Corrective Thinking groups. Other programming included chemical dependency, parenting, relationship building, and life skills. The residents were expected to work full time while in the facility.
The third study site was a 100-bed community-based correctional facility that served primarily adult male felony probationers. The length of treatment was approximately four months. Offenders at this facility received approximately 120 hours of Corrective Thinking. Other programming included chemical dependency, GED classes, vocational assistance, family services, and life skills training.
Finally, the last site used in this study was a 20-bed residential drug treatment program. The program lasted approximately 90 days, during which time offenders received approximately 36 hours of Corrective Thinking. Other programming included drug/alcohol treatment, counseling, and family services. 1
Sample
The sample included men and women from four different treatment facilities who received cognitive-behavioral treatment. The groups in this case were combined to increase sample size and to insure a diverse population was represented. 2 Included in this sample were 344 men and 102 women. The total number of offenders included in this study is 446. The majority of the sample was white (65.6%) and male (77.1%). The majority of the sample were under 40 (82.2%) while close to 50 percent were under 30 years of age. In terms of current offense characteristics for the sample, approximately 40 percent of the people in the sample were serving time for a violent offense (41.0%), while a third of the people in the sample were convicted of a drug offense (32.9%). Most of the offenders in the sample had prior arrests (82.5%). The majority of Differences in number of hours of Corrective Thinking and other program characteristics were controlled for by a quality of programming variable. Differences in the groups will be controlled for through the use of risk/need level and quality of program.the sample had not previously served time in prison (71.9%) or had probation.
Responsivity Assessments (Independent Variables)
In addition to demographics, data on responsivity or personal characteristics were gathered. One often-discussed personal characteristic that is said to be related to success in a program is sexual victimization. Thus, these data were gathered from client files. It is hypothesized that those offenders who have been victimized will perform worse than those who have not. If the files contained no information about sexual abuse, the data were coded as “no.”
Rosenberg’s Self-Esteem Scale
It is suggested that self-esteem is a responsivity issue. Thus people with low self-esteem are hypothesized to perform worse in the program than offenders with high self-esteem. The instrument was developed by Morris Rosenberg (1979) and was originally validated and normed on high school students, but has since been validated on a variety of populations (see Fleming and Courtney, 1984). The Rosenberg Self-Esteem Scale is a 10-item scale with higher scores on the test correlating with higher self-esteem. Thus, the total scores could range from 0-30.
The Culture Fair Intelligence Test
It is hypothesized that those individuals with lower intelligence levels will not perform as well as individuals with moderate to higher intellectual levels. The Culture Fair Intelligence Test measures individual intelligence in a manner designed to reduce the influence of verbal fluency, culture climate, and educational level. The test, which contains four subtests involving different perceptual tasks, was designed by Catell and Catell (1963) and the subtests have since undergone several revisions. The test used in this study was designed for use with all ages. The test has been studied extensively with both reliability and validity data supportive of the test.
Center for Epidemiologic Studies Depression Scale (CES-D)
It is hypothesized that offenders who score high on the depression scale will perform worse in the treatment than offenders who score low on the depression scale. Like self-esteem, it is hypothesized that offenders who are depressed are less likely to participate in the groups and therefore be successful. The scale used in this study is from the Center for Epidemiologic Studies. The CES-D is a 20-item self-report instrument developed at the National Institute of Mental Health. The scale is widely utilized as a screening instrument to distinguish depressed subjects from non-depressed subjects in non-clinical settings (Radloff, 1977). Subjects were asked whether they had experienced a variety of symptoms in the previous week. The CES-D scale has been found to correlate with other measures of depression, including the DSM-III (Fechner-Bates, Coyne, and Schwenk, 1986) and has been found to be an effective screening device for depression, despite differing personal characteristics such as age, gender, and cognitive impairment (Lewinsohn, Seeley, Roberts, and Allen, 1997). If more than five items were missing from the instrument, the score was not used.
Control Variables
The Level of Service Inventory-Revised
The risk principle suggests that those individuals at a higher risk of recidivating will benefit from treatment more than lower-risk individuals (Andrews and Bonta, 1999) and lower-risk individuals may be made “worse” from intensive services. Moreover, higher-risk offenders are more likely to be unsuccessful and recidivate than lower-risk offenders. It is important then, to collect information on risk/need level. Risk/need level was collected using the Level of Service Inventory.
The LSI is a risk/need instrument designed to assess an offender’s likelihood of recidivism. The LSI includes static factors, such as criminal history, as well as dynamic attributes of offenders, such as substance abuse and attitudes about crime. The LSI-R (the version for adults) collects 54 pieces of information about each offender. The items of the LSI are scored in a 0-1 format and added to produce a total score. This makes scoring and adding the scores fairly simple (see Bonta, 1995 for a complete description). The LSI has been found to predict successful completion of probation, institutional misconduct, as well as future criminal offending. Moreover, the LSI has been shown to be valid for a variety of offender types, such as females and juveniles (Kirkpatrick, 1999; Hoge, Andrews, and Leschied, 1996). The LSI was administered to participants upon intake into the program. The data were collected by a project staff member. Each interview took approximately one hour.
Dependent Variables
Data were gathered from client files regarding whether they terminated the program successfully or not. Arrest data were collected in January 2002, and were gathered through official records. Re-incarceration data were collected from the Ohio Department of Rehabilitation and Correction. The average follow-up period was approximately 20 months.
Data Analysis
First, univariate analyses were conducted in order to describe the sample on the various responsivity characteristics, control variables, and dependent variables. Second, logistic regression analysis was conducted to determine what characteristics were related to program completion and recidivism. Independent variables were assessed for multicollinearity and no problems existed.
ResultsTable 1 presents the responsivity characteristics of offenders in the sample. As was stated earlier, 22.9 percent of the sample were women and 34.4 percent of the sample were African American. Data on intelligence level were collected on 438 offenders. Close to 20 percent of the sample had IQ levels at 85 and below (n=84). The largest category of the sample fell between 86 and 99 (n=194, 44.2 percent). The mean intelligence level was 98.32. Most of the sample had no history of sexual abuse as an adult or as a child as recorded in the files (92.4 percent). It should be noted that this information was collected through use of the files and it is probable that this information was not gathered by staff on all offenders who had abuse histories. Moreover, many offenders failed to report this information even if asked by program staff.
About one quarter of the sample appeared to be high in the area of depression as measured
by the Center for Epidemiological Studies Depression Inventory (25.9 percent, n=104). On this scale, the higher the score, the more depressed the individual was. The mean score for the sample of offenders was 8.32 out of 20. In terms of self-esteem, the higher the score, the higher the self-esteem level. The highest score was 30 and the mean was 19.1. Eighty percent of the sample scored 16-30 on the assessment. Twenty percent of the sample scored in the lower half of the scale.
Most of the offenders included in the sample completed the programs successfully (n=353, 79.1 percent). About a quarter of the sample was arrested in the year-and-a-half follow-up time (n=116, 26 percent).
In Table 2, logistic regression coefficients are reported with the standard error in parentheses. The first model predicts whether the offender did not complete the program successfully. In this model, no variable was significantly related to being unsuccessful (not completing the program) at the .05 level or below. One variable however, approached significance. Younger offenders were more likely to fail in the program than older offenders (p<.08). Overall, however, the model was not significant.
Model 2 in Table 2 predicted arrest. While this model had greater predictive ability than the last model, only two variables were related. The results indicate that women were less likely to be arrested than men and offenders with higher risk/need scores as measured by the Level of Service Inventory were more likely to be arrested.
Finally, model 3 in Table 2 predicted re-incarceration. Like the model predicting arrest, gender and risk/need score were significantly related to whether an offender was reincarcerated. Men and people with higher risk/need scores were more likely to be reincarcerated. It should be noted that self-esteem did approach significance. Thus, people with high self-esteem were less likely to be reincarcerated.
DiscussionThis study attempted to sort out the least researched principle of effective intervention: specific responsivity. The responsivity principle assumes that certain personal characteristics of offenders may mediate the effects of treatment. That is, personal characteristics may affect whether an offender can succeed in correctional treatment. The goal of this research was to determine what personal characteristics were related to success in a program that meets the general responsivity principle, cognitive behavioral treatment.
Cognitive behavioral therapy has been gaining attention for a variety of reasons. One reason for its popularity is that it is easier to use than traditional counseling. Correctional staff can easily be trained to conduct the treatment in a relatively short period of time. Another reason for the increase in the use of cognitive-behavioral programs is that positive effects for this type of treatment have been found in the correctional treatment literature (Ross, Fabiano, and Ewles, 1988; Robinson, Grossman, and Porporino, 1991; Robinson, 1995). Cognitive-behavioral programming also targets one of the strongest correlates of crime, antisocial attitudes (Andrews and Bonta, 1999). Finally, cognitive-behavioral programs meet the principle of general responsivity. This principle states that programs should be behavioral in nature and include techniques such as role-playing. For these reasons, cognitive curriculums are dominating correctional treatment. It is important, then, that as more and more treatment programs adopt this form of treatment, the exact effects are sorted out. Although there is evidence that cognitive-behavioral curriculums work, what is still unclear is which kinds of offenders do best in such programs and which do worst. Despite Ross and Fabiano’s (1985) research that found intelligence to be related to success in cognitive-behavioral treatment, there is only speculation as to what personal characteristics may be important in mediating success in treatment.
The current study utilized a longitudinal design with a sample of 344 male and 102 female offenders. A battery of assessments was given to offenders as they entered one of four programs that offered the same cognitive-behavioral treatment. Outcome data were then gathered to determine if any of the personal characteristics of offenders was related to whether an offender did not complete the program, was arrested, and/or incarcerated. What is interesting in this study is that despite previous research that has suggested that certain personal variables may interfere with an offender’s ability to succeed in the program, no such characteristics were found to be related in this study.
While gender was not related to whether an offender completed the program, it was found to be significantly related to recidivism (both re-arrest and incarceration). Men were significantly more likely than women to be arrested and incarcerated. While men tend to have higher recidivism rates in general, it is still interesting that despite controlling for many factors that would likely explain this relationship away, such as level of risk, sexual abuse, depression, and self-esteem, gender emerged as an important responsivity consideration. One explanation for this finding could be that women are more receptive to cognitive-behavioral treatment. In this case, the question becomes: What is it about women that makes them more receptive? While more research is needed in this area, perhaps women are more motivated, able to take responsibility for their actions, and/or relate better to the curriculum. Due to the lack of literature in this area, the reasons for this finding are speculation at this point.
Ross and Fabiano (1985) found that offenders with intelligence levels less than 85 perform worse in cognitive-behavioral treatment than offenders with higher IQ’s. In this study, intelligence level as measured by the Culture Fair was not related to program completion or recidivism. Those offenders with low IQ’s performed as well in cognitive-behavioral treatment as those with higher IQ’s. Thus, in this study, intelligence did not emerge as a responsivity consideration. More research however, is needed in this area to determine whether this finding is related to the type of cognitive-behavioral program and/or whether it is a function of the curriculum and/or staff.
While one study found depression was related to female delinquency (see Obeidallah and Earls, 1999) it is generally thought of as a responsivity consideration (VanVoorhis, 1997; Kennedy and Serin, 1997; Bonta, 1995). Depression as measured by the CES-D scale was not related to successful program completion and/or recidivism. Those who were depressed performed equally as well as those who were not depressed.
Self-esteem has received a lot of attention in the literature on correctional rehabilitation. The Feminist literature has asserted that self-esteem is an important need of female offenders and thus treatment programs should address this need. Other researchers have suggested that self-esteem is not a criminogenic need. It has been suggested in this study that self-esteem is a responsivity issue. However, like depression, self-esteem as measured by Rosenberg’s Self-Esteem Inventory was not related to program completion or recidivism. Those offenders with low self-esteem were just as likely to be unsuccessful in the program, arrested, and incarcerated.
As expected, level of risk was found to be related to many of the outcome indicators. It was highly significant in predicting whether an offender was successful in the program, whether the offender was arrested and whether the offender was incarcerated. This demonstrates the need to match level of programming to the level of risk. Higher-risk offenders need more intensive treatment for a longer duration than lower-risk offenders. Moreover, this finding highlights the importance of collecting information on risk/need level and controlling for it in future evaluation studies.
This study has implications for treatment programs as well as for evaluators studying responsivity in the future. First, more research is needed to determine if, as this study indicates, women are more responsive to cognitive-behavioral treatment than men. This entails collecting data on both men and women and including them in the same sample. As was discussed earlier, many research studies fail to include women and men in the same sample. Finally, other responsivity considerations such as intelligence, sexual abuse, and personality still need to be researched.
One of the most important implications has to do with the link between the general and specific responsivity principles. The general responsivity principle suggests that programs that are behavioral in nature tend to reach the greatest numbers of offenders because of the variety of techniques used. The specific responsivity principle asserts that certain personal characteristics of offenders may interfere with their ability to succeed in the treatment. This study found little support for the specific responsivity principle. In so doing, the study may support the general responsivity principle. The treatment program studied in this research is a cognitive-behavioral program. Cognitive-behavioral programming has been found to be successful at reducing recidivism for many reasons discussed earlier, such as the fact that it targets a known correlate of crime (antisocial attitudes) and uses a variety of behavioral techniques such as role-playing, reinforcement, and modeling. Perhaps specific responsivity characteristics were not found to be related to the outcomes because the cognitive- behavioral program succeeds at reaching a wide range of learning styles of offenders. While more research is needed in the future, this finding suggests that behavioral programs may help negate the effects of offender personal characteristics on treatment success.