Clinical prediction making: examining influential factors related to clinician predictions of recidivism among juvenile offenders.
Despite the well-documented limitations of clinical judgments to assess recidivism risk among adolescent offenders (Hanson & Bussiere, 1998; Meehl, 1996), clinicians continue to make such predictions. (In fact, further evidence of these limitations was found in the present study, with clinician accuracy rates to predict recidivism at 48.7%.) Absent the existence of standardized assessment tools or actuarial methods that have proved effective in predicting recidivism, these predictions are largely based on the sum of data that clinicians have available to them. These judgments often have significant ramifications for youthful offenders in terms of both treatment and legal decision making. Clinicians may base treatment planning decisions on predicted level of risk (Viljoen et al., 2008) or psychosocial variables (O'Donnell & Lurigio, 2008), and more significantly, courts may use these judgments to make decisions regarding dispositions and placements (Hecker & Steinberg, 2002; Hoge, Andrews, & Leschied, 2002; O'Donnell & Lurigio, 2008).
In addition, the larger meaning associated with a prediction of recidivism (e.g., community safety) may further complicate how one goes about making such a prediction. Whereas clinicians oftentimes naturally engage in making clinical predictions, particularly in attempts to evaluate the effects of their therapeutic work (Aegisdottir et al., 2006; Spinhoven, Giesen-Bloo, Van Dyck, & Arntz, 2008), when additional ramifications and external conditions related to their predictions exist, this otherwise natural process may become somewhat more complicated.
However, despite such speculations regarding just how complex this decision making might be, and despite the potential ramifications associated with clinical predictions related to recidivism, surprisingly little is known about how such predictions are made. This is therefore an area that warrants further study, particularly because understanding what contributes to these decisions may prove critical to the manner in which such predictions are used.
Review of the Literature
To date, a tremendous amount of research has been dedicated to examining the accuracy of clinical predictions (Aegisdottir et al., 2006; Hanson & Morton-Bourgon, 2009), and a good deal of research has focused on examining the efficacy of various standardized risk assessment tools (Caldwell, Ziemke, & Vitacco, 2008; Viljoen et al., 2008). However, a paucity of research has examined precisely how clinical predictions are formed. In fact, only a handful of studies have examined factors that might influence clinical predictions, and these studies have been limited to examining predictions related to client characteristics (Bryan, Dersch, Shumway, & Arredondo, 2004; Murphy, Faulkner, & Behrens, 2004; Quinsey, Harris, Rice, & Cormier, 1998; Spinhoven et al., 2008), therapeutic alliance (Escudero, Friedlander, Varela, & Abascal, 2008), and risk and protective factors (O'Donnell & Lurigio, 2008). We discuss each of these briefly in the following paragraphs.
In examining the relationship between specific client characteristics and prediction of treatment outcome, Bryan et al. (2004) found that therapist predictions were related specifically to client ethnicity. In fact, therapists perceived European American clients to be less distressed than non-European American clients. As a result, therapists were more likely to predict more positive treatment outcomes for European American clients than for non-European American clients.
In a study specifically examining the influence of racial differences and similarities between clinicians and clients, Murphy et al. (2004) found that clinicians who were racially different from their clients tended to evaluate the treatment outcomes of these clients more negatively than they did clients who were racially similar. This tendency to negatively evaluate racially different clients was found related to both Black and White clinicians, thus suggesting that treatment outcome predictions were influenced by client-therapist racial similarities and differences.
Moving beyond examining the influence of client demographic characteristics on clinician predictions, Ramnero and Ost (2004) investigated the impact of treatment-related aspects. They found that clinicians were more likely to predict positive treatment outcomes for clients who were considered to be actively participating and goal directed in treatment than those who were not. In addition, these predictions were found to correlate positively to successful treatment outcomes.
O'Donnell and Lurigio (2008) were specifically interested in learning which factors influenced clinician recommendations at the dispositional decision-making stage. They found that behaviors such as anger control, dangerousness to others, and other antisocial behaviors that were identified through psychological evaluation and other initial assessment activities were influential factors in clinician recommendations. In addition to these psychosocial factors, they also found that clinicians were more likely to recommend secure placement for youth who were younger at the time of their initial offense.
Although O'Donnell and Lurigio (2008) specifically examined the process by which clinicians made recommendations for use in dispositional decision making, there continues to be a dearth of research examining the process by which clinicians make predictions of treatment outcomes. As a result, our exploratory study sought to address this gap in the literature. We were interested in examining potential factors involved in clinical prediction making following treatment to determine what, if any, impact each had on the decision-making process. The clinicians in the study were treating adolescent offenders in a secure residential treatment program. The variables that were examined included (a) age at initial juvenile justice system involvement, (b) age at discharge from residential treatment, (c) program completion status, (d) clinician's perceived strength of the therapeutic relationship, and (e) clinician's perception of client commitment to treatment. Because this was an exploratory study, we hypothesized that each of these variables would contribute to clinical prediction making. With the exception of age at discharge, each of the variables selected had previously been associated with positive treatment outcomes for juvenile offenders (i.e., age at initial juvenile justice system involvement, treatment completion) and for other client groups (i.e., therapeutic alliance). The five variables are discussed briefly in the next section.
Age at Initial Juvenile Justice System Involvement
The age at which an adolescent offender initially offends is an empirically supported risk factor that has been found to be a predictor of recidivism (e.g., Mulder, Brand, Bullens, & van Marie, 2011). In the context of offender risk assessment, age at which a juvenile initially offends is included on most standardized risk assessments today (e.g., the Juvenile Sex Offender Assessment Protocol-II, Estimate of Risk of Adolescent Sexual Offense Recidivism, and North Carolina Assessment of Risk). Because of its research basis and prominence in risk assessment, we hypothesized that age at initial juvenile justice system involvement would likely contribute to clinical prediction making.
Age at Discharge From Residential Treatment
Of the five selected variables, age at discharge from residential treatment had not been previously found to be directly associated with treatment outcome. However, age at discharge was believed to be a potentially influential variable because of its relationship with other age-related risk factors. Namely, it was hypothesized that age at discharge may indicate the length of time an adolescent has been in treatment, particularly because it is somewhat common to retain youth who are perceived as high risk in residential treatment facilities until adulthood. In addition, because age at discharge could be related to case closure status and subsequent diminishing supports, education status, and length of time to adulthood, this was hypothesized as a potential influential factor. Age at discharge also had specific relevance to this study because the residential treatment program was a secure facility, and as a result, youth who were 18 years or older at discharge may not have been eligible for long-term reintegration services to effectively support the transition back into the community, thus possibly influencing therapist treatment predictions.
Program Completion Status
Shorter or incomplete residential treatment stays have been identified as factors associated with recidivism among juvenile offenders (Miner, 2002); as a result, program completion status was also included. In addition to its evidence basis, we hypothesized that clinicians might be naturally predisposed to associate treatment success with treatment completion, particularly in involuntary structured programs that are developed to systematically address needs, such as the treatment program participating in this study.
Clinician's Perceived Strength of Therapeutic Relationship and Clinician's Perception of Client's Commitment to Treatment
Both clinician's perceived strength of the therapeutic relationship and clinician's perception of client's commitment to treatment were selected because each has been historically conceptualized as contributing to the therapeutic alliance (Beck, 1973; Horvath & Luborsky, 1993), which has been found to be a predictor of treatment outcomes (Horvath & Symonds, 1991; Orlinsky, Grawe, & Parks, 1994). To address the lack of a standardized tool to assess therapeutic alliance, we developed two items in an effort to operationalize therapeutic alliance (see the Appendix). These items consisted of the clinician's perceived strength of the therapeutic relationship and perceived client commitment to treatment. These definitions were consistent with Ramnero and Ost's (2004) findings that perceptions of client participation and goal-directed behavior positively influenced positive prediction making.
Juvenile offenders. There was a total of 166 participants in this study. All of the participants were juvenile offenders placed in a residential treatment facility. Participant race was as follows: .006% (n = 1) American Indian, 4.8% (n = 8) biracial, 78.3% (n = 130) Black, and 16.2% (n = 27) White. In terms of ethnicity, 100% of the youth who were Black were African American, and 7.4% (n = 2) of the youth who were White were Arab American and 92.6% (77 = 25) were European American.
Participant age at release from residential treatment ranged from 14 to 21 years, with the age breakdown as follows: 1.8% (n = 3) at age 14, 10.8% (n = 18) at age 15, 21% (n = 35) at age 16, 35.5% (n = 59) at age 17, 22.8% (n = 38) at age 18, 4.2% (n = 7) at age 19, 3% (n = 5) at age 20, and 0.06% (n = 1) at age 21. Five of the participants were missing age data.
Treatment program. The residential treatment program is a locked, 80-bed facility for male juvenile offenders. Because the program is a locked facility, it typically housed more serious offenders. Juveniles placed in the program have committed a range of crimes, including armed robbery, sexual offenses, and drug-related crimes.
Youth are housed in all-inclusive units of 10 with single bedrooms. The educational and treatment programs are delivered within each unit, with one clinician assigned to each unit. Clinical treatment consists of group counseling 5 times per week with individual and family counseling occurring biweekly. A seven-stage treatment model that is based on cognitive behavior therapy is used for the group treatment program (Calley, 2007). The treatment model emphasizes behavioral control, interpersonal skill development, and relapse prevention. The treatment model is sequential, requiring youth to complete one stage before moving on to the next. The program is individualized so that each youth progresses through the program at an individual pace. The average length of the program was 11.5 months.
Clinicians. Ten clinicians participated in this study. Each of the clinicians was assigned to one of the eight units within the facility, with each unit comprising 10 youth. Two clinicians resigned during the study and were replaced with new clinicians.
There were four male clinicians and six female clinicians. Each of the clinicians was a master's-level practitioner with degrees and state licensure in one of the following mental health disciplines: counseling, social work, or marriage and family therapy. With the exception of one clinician who was hired immediately after completing her master's degree, each of the clinicians had at least 1 year of prior clinical experience.
Data for this study were collected at two intervals: youth admission to residential treatment and youth discharge from residential treatment. Two data forms were developed and used specifically to collect the data: the Initial Youth Information Form and the Discharge Tracking Form.
Although both of the data collection tools were used to collect comprehensive information, only specific items from each of the tools were used for this study. Within 30 days of a youth's admission to the program, the assigned clinician completed the Initial Youth Information Form, which included historical information about each youth. Age at initial involvement with the juvenile justice system was one of the variables recorded at this point, and it was defined as the first formal interaction the youth had with the juvenile court. The data were supported by historical court reports that were a part of each youth's case record.
Upon each youth's discharge, the treating clinician completed the Discharge Tracking Form, which documented information about the youth's treatment program and progress. Age at discharge and program completion status were recorded at this point. Program completion status was defined as having completed all seven stages of the treatment program. In addition, each clinician was asked to respond to the two items pertaining to his or her perception of the client's commitment to treatment and his or her perception of the strength of the therapeutic relationship.
We performed a frequency analysis to examine prediction type made by clinicians. There were two possible types of predictions: that the youth would recidivate or that the youth would not recidivate within 1 year of release from residential placement. Recidivism included any type of offense, both sexual and nonsexual. Clinicians predicted that within 1 year after treatment, 60.8% (n = 101) of the youth would recidivate, whereas 39.2% (n = 65) would not recidivate. As stated earlier, the clinician accuracy rate in the 1-year follow-up was 48.7%. The results for each of the variables examined are provided in the following sections.
Client Age at Initial Juvenile Justice System Involvement
Client age at initial juvenile justice system involvement was separated into two categories: ages 13 and younger and ages 14 and older. Of the 166 participants, 36.7% (n = 61) were initially involved with the juvenile justice system at age 13 or younger, whereas 55.4% (n = 92) had initial contact with the juvenile justice system at age 14 or older. Thirteen (7.8%) participants were missing data on this variable. We used a chi-square test to compare the age groups with predictions of recidivism. No significance was found, [chi-square] (1) = 0.42, p > .05.
Client Age at Program Discharge
Client age at program discharge was separated into two categories: 15-17 years and 18 years and older. Of the 166 participants, 69.3% (n = 115) of the clients were 15-17 years of age at discharge from residential treatment, whereas 30.7% (n = 51) were 18 years or older at discharge. We used a chi-square test to compare the age of discharge and prediction of recidivism. No significance was found for this variable, [chi-square] (1) = 2.93, p > .05.
Program Completion Status
Program completion status was separated into two categories: complete or incomplete. Of the 166 participants, 70.5% (n = 117) of clients completed the treatment program, whereas 29.5% (n = 49) did not. We used a chi-square test to compare the counselor's prediction of recidivism and program completion status. Significance was found on this variable, [chi-square] (1) = 10.26, p < .05, with prediction of no recidivism related to program completion status.
A logistic regression was used to further analyze this relationship. Completing the program significantly affected the clinicians' prediction of recidivism (likelihood ratio: [chi-square] = 10.89, p = .001), thus increasing the likelihood of the clinician predicting no recidivism, as indicated by the positive parameter estimate (0.62; SE = 0.20). Clinicians were 3.46 times more likely (95% confidence interval [CI; 1.58, 7.58]) to predict no recidivism for youth who completed the program.
Clinician's Perceived Strength of Therapeutic Relationship
Clinician's perceived strength of the therapeutic relationship was broken down into four possible categories: tenuous, mild, strong, and very strong. Of the 166 participants, 9.6% (n = 16) of the therapeutic relationships were perceived as tenuous, 22.3% (n = 37) were perceived as mild, 48.2% (n = 80) were perceived as strong, and 19.9% (n = 33) were perceived as very strong. We used a chi-square test to compare the clinician perception of relationship and prediction of recidivism. Clinician prediction of recidivism was significantly related to the perceived strength of the therapeutic relationship, [chi-square](3) = 31.23, p < .05.
We used a logistic regression to further analyze this relationship. Therapeutic relationship significantly affected the clinicians' prediction of recidivism (likelihood ratio: [chi-square] = 31.66, p < .000). Increasing levels of therapeutic relationship increased the chance of having the clinician predict no recidivism, as indicated by the positive parameter estimate (1.22; SE = 0.25). The chance of a clinician predicting no recidivism increased by 3.38 times (95% CI [2.07, 5.54]) for each level of increase in the therapeutic relationship.
Clinician Perception of Client Commitment to Treatment
Clinician perception of client commitment to treatment was separated into four possible categories: not committed, mildly committed, committed, and very committed (see the Appendix for definitions). Of the 166 participants, 6.6% (n = 11) were perceived by the assigned clinician as not committed to the treatment program, 24.1% (n = 40) were perceived as mildly committed, 48.2% (n = 80) were perceived as committed, and 21% (n = 35) were perceived as very committed to the program. We used a chi-square test to compare the commitment level and prediction of recidivism. The perceived commitment level of the client was significantly related to clinician prediction of recidivism, [chi-square](1) = 24.24, p < .05.
A logistic regression was used to further analyze this relationship. Commitment to treatment significantly affected the clinicians' prediction of recidivism (likelihood ratio: [chi-square] = 48.80, p < .0001). Increasing levels of commitment to treatment (i.e., commit) increased the chance of the clinician predicting no recidivism, as indicated by the positive parameter estimate (1.73; SE = 0.31), and the chance of a clinician predicting no recidivism increased by 5.63 times (95% CI [3.07, 10.33]) for each level of increase in perceived client commitment.
A final logistic regression model included clinician prediction of recidivism and all three predictors (program completion status, perceived strength of the therapeutic relationship, and perceived level of client commitment to treatment; see Table 1). The model was significant (likelihood ratio: [chi-square] = 56.45, p < .0001); however, with all three variables in the model, program completion status was no longer significant (Wald [chi-square] = 0.22, p = .6407). Clinicians' perceived strength of the therapeutic relationship (Wald [chi-square] = 5.96, p = .0147) and clinician perception of client commitment to treatment (Wald [chi-square] = 18.87, p < .0001) both significantly affected clinicians' predictions of recidivism. Commitment to treatment had the largest positive effect (parameter estimate = 1.44, SE = 0.33), whereas the effect of therapeutic relationship was less (parameter estimate = 0.71, SE = 0.29) but significant. The effect size was also evident in the odds ratios. Each higher level (i.e., greater level of commitment) of clinician perception of client commitment to treatment increased the chances of the clinician predicting no recidivism by 4.23 times, and each higher level of clinician's perceived strength of the therapeutic relationship increased the chances of the clinician predicting no recidivism by 2.04 times.
To determine which model had the best fit (single variable or the three-variable model), we used Akaike's Information Criteria (AIC; see Table 2). The three-variable model had the best fit (indicated by the lowest AIC value), followed by commitment to treatment, therapeutic relationship, and program completion status.
In sum, multiple lines of evidence--parameter estimates from the individual models and the three-variable model, as well as AIC--suggest that perceived commitment to treatment has the strongest influence on whether the clinician will predict recidivism. Therapeutic relationship also has a strong influence on whether the clinician will predict an individual will recidivate. Program completion, when considered alone, did significantly influence the clinical predictions, although this predictive power was lost when the other variables were included in the model.
There were some surprising findings in this study, including the following: Clinicians were much more likely to predict that youth would reoffend within 1 year following treatment; age-related variables did not influence predictions of recidivism; and clinicians' perception of the therapeutic relationship, including client commitment to treatment and strength of the therapeutic relationship, significantly influenced clinician prediction of recidivism.
Clinician Prediction of Recidivism
With regard to prediction type, clinicians predicted that youth would reoffend within 1 year postrelease in a majority of the sample (60.8%). There could be several reasons for this tendency to predict recidivism among this population. First, all of the youth included in the study were considered high risk, evidenced by their placement in a locked setting. Second, clinicians working directly with this population were well aware of the youths' offense history and comparable risk status to other offenders within the broader juvenile justice system. Both these factors could have influenced clinicians to be guarded in their prognoses, leading them to believe that regardless of potential treatment effectiveness, treatment would be unsuccessful for most of the youth in this population.
Age-Related Variables and Prediction of Recidivism
Neither age at initial juvenile justice system involvement nor age at discharge from treatment influenced clinician prediction of client recidivism. Because age at initial juvenile justice system involvement is a well-documented risk factor for juvenile offenders--incorporated into the majority of standardized risk assessments in use today--this finding was especially surprising. However, although age at discharge did not have a similar evidence basis as a risk factor, we hypothesized that clinicians would be influenced by this variable because of its potential indicator of status of length of time in treatment and lack of available resources following case closure.
The fact that neither of these age-related variables significantly influenced clinician decision making could be because clinicians had limited to no ability to influence these factors. Age at which youth were initially involved in the juvenile justice system was a historical factor and therefore completely outside the realm of clinician control. Clinicians may also have perceived little control over determining the age at which clients are discharged. This is especially true in today's managed care climate of juvenile justice in which residential placements are often viewed as a last resort rather than as a key part of the treatment continuum, and residential stays are typically limited in efforts to save money.
Program Completion Status and Prediction of Recidivism
The finding that treatment completion status was an influential factor in recidivism predictions was consistent with previous research, specifically when examined as a single variable. However, this seems to be the first study in which a three-variable model including the therapeutic relationship, commitment, and program completion status was compared with single variables. Therefore, the finding that program completion status did not have significance when included with the other two variables is a new finding.
Therapeutic Relationship and Prediction of Recidivism
The two variables related to therapeutic alliance--clinician perception of therapeutic relationship strength and client commitment to treatment--were found to significantly influence clinician prediction making. Because a sizable amount of research supports the relationship between treatment outcomes and therapeutic alliance, we hypothesized that these variables would influence clinician prediction making. However, because no previous research had been conducted specifically on therapeutic alliance and juvenile offender outcomes, we were uncertain if the relationship between therapeutic alliance and outcome prediction making would apply to this population.
Particularly interesting was the finding that at least some indication of a therapeutic alliance was necessary for clinicians to predict that youth would be successful following treatment. In fact, of the 65 clients whom clinicians predicted would not reoffend, none were perceived as tenuous therapeutic relationships, and none were perceived as not committed to the treatment process. This suggests that clinicians believe that without the development of even a hint of a therapeutic alliance (i.e., mild commitment to treatment, mildly strong therapeutic relationship), successful treatment outcomes cannot be attained.
This finding could be interpreted as clinician ego insofar as clinicians are more apt to attribute client outcomes--both positive and negative--primarily to their clinical work, and more specifically, their ability to form a therapeutic alliance. Thus, they might conceive a certain degree of power over client outcomes. Although this could be a reflection of clinician ego, it could also be viewed as an act of self-preservation because it may serve to further justify the clinician's role and the clinician's value in the treatment process.
Finally, this finding may be related to the serious implications that are associated with clinician predictions related to work with juvenile offenders (i.e., court decision making). Because clinician predictions are typically requested by the court for use in legal decision making, clinicians may be most comfortable basing such decisions on what matters most to them in their work: the therapeutic relationship. The counseling relationship is after all central to the treatment process; therefore, clinicians may, in turn, have more trust in this than other factors when making decisions about client prognoses.
Limitations of the Study
Two methodological problems may have influenced the results of this study. First, because there was no comparison group, some caution must be used when interpreting the results. Second, the results cannot be broadly generalized because the study consisted of a single sample. Including participants from multiple residential treatment programs would have been preferred; however, this is understandably quite challenging when working with public systems.
Although not a primary objective of this study, the inaccuracy of clinician recidivism predictions was consistent with previous findings related to other client groups. As a result, this finding was not necessarily surprising. However, whereas the inaccuracy of predictions was consistent with previous findings, it was surprising that clinicians were much more apt to lean on the side of conservatism in predicting that youth would reoffend when in fact the majority of youth did not. Particularly because of the unique population investigated, these findings prompt further questions as to why such weight continues to be given to clinical outcome predictions for juvenile offenders. Because of the serious ramifications that such predictions often have on involuntary minors, one of our most vulnerable populations, this issue requires much further investigation.
Overall, the findings suggest that clinician predictions are primarily influenced by factors related to clinician perceptions of the therapeutic relationship, with the absence of a perceived therapeutic alliance leading to a negative prediction of treatment outcome. It is interesting that although these factors were found to be influential, age at initial juvenile justice involvement--the evidence-based risk factor--was ignored.
Because the trend related to the use of clinical input in court decision making will likely continue, it is imperative that we fully understand how such predictions are made. Jurists and other legal professionals would be well served knowing that clinicians are more likely to predict a negative treatment outcome over a positive one perhaps as a result of the ramifications of such decisions. In addition, the types of factors that influence prediction making should be known so that those requesting such predictions are fully equipped with this knowledge before making such requests. Perhaps by better understanding how such decisions are conceived, one may demand more effective methods by which to make such predictions, or one may discontinue this practice of clinical prediction making, especially with juvenile offenders, and instead allow actual outcome data to serve as a guide to outcome predictions.
More important, however, are the implications that these findings have for counselors and other clinicians. In an era of accountability in clinical practice, we as counselors must rethink our ability to effectively engage in clinical prediction making with specific client groups. The effects of our predictions may indeed cause harm to our clients because treatment and legal decisions may be made for them on the basis of our own faulty information. We cannot afford to ignore existing evidence and instead rely on our own perceptions of the importance of the therapeutic relationship. Doing so may indeed serve to undermine our ability to be fully trusted by those who have been involuntarily entrusted to us and to be fully accountable to our profession. It is our ethical responsibility as counselors to be cognizant of our limitations and not engage in practices that are outside the scope of our competence. And because clinical prediction making related to outcomes of juvenile offenders is clearly one area in which counselors and other clinicians have yet to achieve competence, this should not be included in our clinical practices.
Much more research is needed to further examine the issue of clinical prediction making, especially as related to juvenile offender outcomes. With the results of this exploratory study as a first step, it is our hope that work in this important area will continue and that such work can be used to guide policy and practice issues in juvenile justice.
Therapist Perceptions and Predictions
Directions. For each item, please check the one response that best describes your opinion related to the youth.
1. In your opinion, how committed to treatment was the youth?
 Very committed:completed all work, and then some; only typical problems and resistance demonstrated
 Committed: completed work; however, did nothing additional to indicate extreme commitment; only typical problems and resistance demonstrated
 Mildly committed: required continuous prompting to complete work; commitment typically appeared superficial
 Not committed: completely resistant to treatment throughout the majority of the treatment process; majority of work incomplete
2. In your opinion, what best describes the therapeutic relationship that you had with this client?
 Very strong: characterized by mutual constant honesty and openness regarding the therapeutic relationship and belief in the treatment process
 Strong: characterized by a solid working relationship; however, consistent mutual honesty and openness was not always present; relationship could have become even stronger
 Mildly strong: characterized by intermittent feelings of relationship strength, more often feeling that relationship could become much more strong
 Tenuous: characterized by a lack of trust from either the client or the therapist
3. Based upon the information that you have at the youth's discharge, what do you believe is the likelihood that he will commit any type of crime following release from the program?
 More than likely the youth will reoffend within 1 year following treatment
 More than likely the youth will not reoffend within 1 year following treatment
Aegisdottir, S., White, M., Spengler, P., Maugherman, A., Anderson, L., Cook, R., ... Rush, J. (2006). The meta-analysis of clinical judgment project: Fifty-six years of accumulated research on clinical versus statistical prediction. The Counseling Psychologist, 34, 341-382.
Beck, A. T. (1973). Cognitive therapy and the emotional disorders. New York, NY: International Universities Press.
Bryan, L., Dersch, C., Shumway, S., & Arredondo, R. (2004). Therapy outcomes: Client perception and similarity with therapist view. American Journal of Family Therapy, 32, 11-26.
Caldwell, M. F., Ziemke, M. H., & Vitacco, M. J. (2008). An examination of the Sex Offender Registration and Notification Act as applied to juveniles: Evaluating the ability to predict sexual recidivism. Psychology, Public Policy, and Law, 14, 89-114.
Calley, N. G. (2007). Integrating theory and research: The development of a research-based treatment program for juvenile male sex offenders. Journal of Counseling & Development, 85, 131-142.
Escudero, V., Friedlander, M., Varela, N., & Abascal, A. (2008). Observing the therapeutic alliance in family therapy: Associations with participants' perceptions and therapeutic outcomes. Journal of Family Therapy, 30, 194-214.
Hanson, R. K., & Bussiere, M. T. (1998). Predicting relapse: A meta-analysis of sexual offender recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348-362.
Hanson, R. K., & Morton-Bourgon, K. (2009). The accuracy of recidivism risk assessments for sexual offenders: A meta-analysis of 118 prediction studies. Psychological Assessment, 21,1-21.
Hecker, T., & Steinberg, L. (2002). Psychological evaluation at juvenile court disposition. Professional Psychology: Research and Practice, 33, 300-306.
Hoge, R. D., Andrews, D. A., & Leschied, A. (2002). Youth Level of Service/Case Management Inventory: YLS/CMI manual. Toronto, Ontario, Canada: Multi-Health Systems.
Horvath, A. O., & Luborsky, L. (1993). The role of the therapeutic alliance in psychotherapy. Journal of Consulting and Clinical Psychology, 61,561-573.
Horvath, A. O., & Symonds, B. D. (1991). Relation between working alliance and outcome in psychotherapy: A meta-analysis. Journal of Counseling Psychology, 38, 139-149.
Meehl, P. E. (1996). Clinical versus statistical prediction: A theoretical analysis and a review of the evidence. Northville, NJ: Jason Aronson. (Original work published 1954)
Miner, M. H. (2002). Factors associated with recidivism in juveniles: An analysis of serious juvenile sex offenders. Journal of Research in Crime and Delinquency, 39, 421-436.
Mulder, E., Brand, E., Bullens, R., & van Marie, H. (2011). Risk factors for overall recidivism and severity of recidivism in serious juvenile offenders. International Journal of Offender Therapy and Comparative Criminology, 55, 118-135.
Murphy, M., Faulkner, R., & Behrens, C. (2004). The effect of therapist-client racial similarity on client satisfaction and therapist evaluation of treatment. Contemporary Family Therapy, 26, 279-292.
O'Donnell, P. C., & Lurigio, A. J. (2008). Psychosocial predictors of clinicians' recommendations and judges' placement orders in a juvenile court. Criminal Justice and Behavior, 35, 1429-1447.
Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994). Process and outcome in psychotherapy. In A. E. Bergin & S. L. Garfield (Eds.), Handbook of psychotherapy and behavior change (4th ed., pp. 270-376). New York, NY: Wiley.
Quinsey, V., Harris, G., Rice, M., & Cormier, C. (1998). Violent offenders: Appraising and managing risk. Washington, DC: American Psychological Association.
Ramnero, J., & Ost, L. G. (2004). Prediction of outcome in the behavioural treatment of panic disorder with agoraphobia. Cognitive Behaviour Therapy, 33, 176-180.
Spinhoven, P., Giesen-Bloo, J., Van Dyck, R., & Arntz, A. (2008). Can assessors and therapists predict the outcome of long-term psychotherapy in borderline personality disorder? Journal of Clinical Psychology, 64, 667-686.
Viljoen, J. L., Scalora, M., Cuadra, L., Bader, S., Chavez, V., Ullman, D., & Lawrence, L. (2008). Assessing risk for violence in adolescents who have sexually offended. Criminal Justice and Behavior, 35, 5-23.
Nancy G. Calley, Department of Counseling and Addiction Studies, University of Detroit Mercy, and Spectrum Human Services, Inc., Westland, Michigan; Emily M. Richardson, Department of Counseling and Addiction Studies, University of Detroit Mercy. Emily M. Richardson is now at Harbor Oaks Hospital, New Baltimore, Michigan. This research was supported by the Office of Juvenile Justice and Delinquency Services (Grant 2008-DD-BX-1003). Correspondence concerning, this article should be addressed to Nancy G. Calley, Department of Counseling and Addiction Studies, University of Detroit Mercy, 4001 West McNichols Road, 234 Reno Hall, Detroit, MI 48221 (e-mail: firstname.lastname@example.org).
TABLE 1 Three-Variable Model and Prediction of Recidivism Wald 95% Wald CL Parameter and Point [chi Effect df Estimate SE square] Lower Upper p Analysis of Maximum Likelihood Estimates Intercept 1 -6.87 1.21 32.32 <.0001 Complete 1 1 0.11 0.24 0.22 .6407 Relationship 1 0.71 0.29 5.96 .0147 Commit 1 1.44 0.33 18.87 <.0001 Odds Ratio Estimates Complete 1 vs. 1.26 0.48 3.27 2 Relationship 2.04 1.51 3.62 Commit 4.23 2.21 8.10 Note. For testing the global null hypotheseis: [beta] = 0. For the likelihood ratio: [chi square] (3) = 56.45, p < .0001. CL = confidence limit; Complete = program completion status; Relationship = clinician's perceived strength of the therapeutic relationship; Commit = clinician's perception of client commitment to treatment. TABLE 2 Akaike's Information Criteria (AIC) Table [DELTA] Model AIC AIC Completion status, therapeutic 173.8 0.0 relationship, and commitment Commitment 177.5 3.7 Therapeutic relationship 194.6 20.8 Completion status 215.4 41.6
|Printer friendly Cite/link Email Feedback|
|Author:||Calley, Nancy G.; Richardson, Emily M.|
|Publication:||Journal of Addictions & Offender Counseling|
|Date:||Oct 1, 2011|
|Previous Article:||Using therapeutic letters to navigate resistance and ambivalence: experiential implications for group counseling.|
|Next Article:||Evaluation of web-based and counselor-delivered feedback interventions for mandated college students.|