Meta-Analytic Evaluation of Skills Training Research for Individuals With Severe Mental Illness

Melody Nichols Dilk
Psychological Resources

Gary R. Bond
Department of Psychology Indiana University–Purdue University at Indianapolis

Citation: Journal of Consulting and Clinical Psychology, December 1996 Vol. 64, No. 6, 1337-1346

A meta-analysis of 68 studies examined the effectiveness of skills training for individuals with severe mental illness and the influence of such factors as methodological rigor, choice of outcome measures, and service settings. The methodological quality of these studies was generally very good. Skills training was found to be moderately to strongly effective in increasing skill acquisition and reducing psychiatric symptoms. However, effect size varied by type of outcome measure, with situationally specific measures yielding larger effects than measures of skill usage and role functioning. Studies rarely examined whether acquired skills were used outside the training setting. Surprisingly, most studies were confined to inpatient settings, and most focused exclusively on social skills. Implications are discussed for future research.

An earlier version of this article was presented in August 1995 at the 103rd Annual Convention of the American Psychological Association, New York, New York.
This work was supported in part by an American Psychological Association Science Directorate Award and by National Institute of Mental Health Grant MH00842.
We thank Betsy Becker, Mark Lipsey, Larry Hedges, and Jack Hunter for their consultation regarding some technical aspects of the meta-analysis.
Correspondence may be addressed to Melody Nichols Dilk, Psychological Resources, 9801 Fall Creek Road, Number 103, Indianapolis, Indiana, 46256.

For over 2 decades, skills training has been regarded as a major psychosocial intervention strategy for persons with severe mental illness (SMI), as suggested in numerous narrative reviews ( Bellack, Turner, Hersen, & Luber, 1984 ; Brady, 1984a, 1984b ; Curran, 1985 ; Dion & Anthony, 1987 ; Donahoe & Driesenga, 1988 ; Halford & Hayes, 1991 ; Hayes, 1989 ; Hersen & Bellack, 1976 ; Hersen, Eisler, & Miller, 1973 ; Herz, 1986 ; Kovacs, 1980 ; Margules & Anthony, 1976 ; Slade & Bentall, 1989 ; Stravynski & Shahar, 1983 ; Test & Stein, 1978 ; Wallace et al., 1980). Although most reviews have focused specifically on social skills or assertiveness training, a few also have examined efforts to teach other types of skills, such as those needed for activities of daily living, employment, and independent living ( Dion & Anthony, 1987 ; Halford & Hayes, 1991 ; Hayes, 1989 ; Slade & Bentall, 1989 ; Test & Stein, 1978 ). Without exception, reviews have concluded that skills training using behavioral methods is effective in teaching a variety of social and living skills to persons with a wide variety of diagnoses, including schizophrenia and major depression. The reviews are quite variable, however, both in the way information is presented and in the strength of the conclusions. For example, Dion and Anthony (1987) concluded optimistically that skills training is effective for virtually every area of role functioning for persons with severe psychiatric disabilities. In contrast, Wallace et al. (1980) , reached the more conservative conclusion that, although social skills training has demonstrated effectiveness in changing topographical features of behavior and self-reports of anxiety, such changes have not been demonstrated to result in substantial differences in patients' quality of life. The review literature displays almost as much diversity and contradiction as does the collection of the primary studies themselves. Furthermore, methodological problems noted in some reviews ( Curran, 1985 ; Donahoe & Driesenga, 1988 ; Kovacs, 1980 ; Wallace et al., 1980 ) raise doubts about the strength and validity of general conclusions regarding efficacy.

Until recently, there were no published quantitative reviews of this literature. Two meta-analyses have been reported since 1990 ( Benton & Schroeder, 1990 ; Corrigan, 1991 ); however, both were limited by including only a small number of studies and by focusing solely on social skills. Currently, then, despite the popularity of skills training, a comprehensive, quantitative review is lacking.

The current meta-analysis examines the following issues: (a) What has been done; that is, what is the extent and quality of the published and unpublished research? (b) What has been found; that is, how effective is skills training, for what kinds of skills, over what period of time, and for which outcome domains? (c) What factors influence effectiveness; that is, what methodological and substantive factors affect the impact of skills training?


The sample consisted of studies examining the effectiveness of behavioral approaches to teaching skills to adults with SMI. The search encompassed studies published between 1970 and 1992 and doctoral dissertations and master's theses completed during that period. Studies with at least 5 participants, using either between-group or within-group (i.e., one-group, pretest—posttest) designs were included. Between-group designs included both two-group and multiple-group comparisons. We excluded studies with vaguely described interventions or with inadequate outcome information.

To locate published studies, we conducted computer searches of Medline and PsycINFO and manual searches of bibliographies of skills training reviews. Meeting the inclusion criteria were 23 of the 27 studies identified by Benton and Schroeder (1990) and 20 of 23 identified by Corrigan (1991) . These searches yielded 168 published studies, of which 42 were included in the meta-analysis. Of those excluded, 18% were program descriptions, lacking an appropriate study design; 10% had fewer than 5 participants; 17% evaluated approaches that were not specifically skills training; 16% did not pertain specifically to SMI; and 10% had insufficient diagnostic or outcome data. The search for published studies was complemented by a computer literature search of Dissertation Abstracts International. It yielded 41 studies potentially meeting the inclusion criteria. We obtained the full texts of relevant studies and subsequently included 25 dissertations and 1 thesis in the analyses.

A coding system adapted from Benton and Schroeder (1990) was used to code methodological, training program, client, and outcome method characteristics. Methodological variables included document source (publication vs. dissertation or master's thesis), year of publication, design type (between- vs. within-groups), randomization of participants, use of standardized diagnostic interviews, concurrent drug treatment, use of a well-defined treatment protocol, use of naive raters, and use of trained or experienced raters. A composite index of methodological quality was calculated on the basis of all but the first three of these methodological variables. Training program variables included intensity and duration, orientation (behavioral vs. cognitive—behavioral), inclusion of core behavioral techniques, content focus, and location of training. Client variables included age, diagnosis, gender, and hospitalization history.

Outcome method variables were coded on four dimensions: content, level of acquisition, generalizability, and modality: The content dimension included (a) skill acquisition, (b) symptom reduction, (c) personal adjustment (e.g., Global Assessment of Functioning scale), (d) hospitalization, and (e) vocational readiness. The level of standardization dimension included (a) researcher-developed measures, (b) unstandardized measures borrowed from other research, and (c) standardized measures. The generalizability dimension included (a) measures that used methods that were similar to the training and obtained in the same setting as the training, (b) measures that used methods that were similar to the training but ratings that were obtained in a different setting, and (c) inferential measures that were measures using methods that differed from the training and were obtained in a different setting. The modality dimension included (a) self-report, (b) rating of behavior (other than role play) in the training group, (c) in vivo behavioral rating, (d) rating of confederate interaction, (e) rating of role play, (f) homework assignment completion, (g) physiological measures, (h) response to a situation presented on audio or video tape, (i) clinical assessment, and (j) family member rating of the participant.

Operational definitions for each variable were detailed in a coding manual (available from Melody Nichols Dilk). All of the coding was conducted by M.N. Dilk, who later repeated all coding as a double-check. Gary R. Bond served as an independent coder on key variables requiring some degree of subjective judgment. He coded all of the published studies and a sample (62%) of the dissertations. Ratings were compared and discrepant codings were discussed and reconciled. In addition, a doctoral-level student independently coded all of the published studies on the outcome method variables. Discrepancies in coding these variables were reconciled by reinspection of the original study or by information from other studies.
Statistical Analysis

The goals of meta-analysis include assessing overall significance for a group of studies, determining whether the studies are homogeneous with respect to outcome, and if they are not, determining whether moderator variables explain the variation among the studies. The unit of analysis in meta-analysis is the effect size. It reflects the level of change in the treatment group relative to the comparison group, or the posttest relative to the pretest, as measured in standard deviation units. A positive effect indicates the amount of improvement the treatment group had beyond the comparison group or pretest assessment. Lipsey (1990) has developed empirical norms for describing the magnitude of effect sizes: a value less than .33 is "small," a value between .33 and .55 is "medium," and any larger value is "large."

The specific techniques used in the current meta-analysis are described by Cooper and Hedges (1994) . For between-group studies, the effect size d is defined as the difference between treatment and control group means divided by the pooled within-group standard deviation, using Hedges' (1984, p. 32) correction for small sample size. Study effect sizes are then combined to obtain an overall effect size for the sample of studies. The overall effect, d., is calculated by using a weighted mean that takes into account the differing variances in the individual studies ( Shadish & Haddock, 1994, pp. 265—267 ). To test whether d. is significantly different from 0, z is calculated, using the mean weighted effects and the conditional variance. The overall effect size is significantly different from zero if z exceeds 1.96; that is, the 95% critical value for the standard normal distribution ( Shadish & Haddock, 1994 ). To determine whether mean effects of studies share a common population effect size, homogeneity of variance is calculated as follows:

Homogeneity of aggregated effects ( d. ) is rejected if Q exceeds the upper-tail critical value of chi-square at k - 1 degrees of freedom, where k is the number of studies. When homogeneity is rejected, the overall effect is disaggregated on the basis of potential moderator variables. For categorical variables, an analog to analysis of variance developed for meta-analysis partitions the total variance, Q (or QT), into QB, the explained variance (i.e., variance between groups), and QW, residual variance (i.e., unexplained variance remaining within groups of the grouping variable). The significance of QBis tested in the same fashion as the QTstatistic, using p - 1 degrees of freedom, where p is the number of groups. For moderator variables that are metric, Pearson correlations are used to examine the association with study-level effect sizes.

For within-group studies, the effect size is defined as the difference between the mean posttest score and the mean pretest score, divided by the standard deviation of the pretest scores. The correction for small samples ( Becker, 1988 ) requires knowledge of the correlation between pretest and posttest scores, which studies only rarely report. Following the usual convention, we inputed the value of zero for the correlation when this information was missing. Further meta-analytic procedures for within-group designs, analogous to the between-group procedures, have been described by Becker (1988) .

Effect sizes were calculated at the time of coding. The coding manual detailed decision rules for calculating effect sizes when means and standard deviations were not reported. Nonsignificant results lacking supporting statistical information were coded as an effect size of zero ( Glass, McGaw, & Smith, 1981 ; Hunter & Schmidt, 1990 ).

An overall effect size was calculated for studies with multiple outcome measures, averaging across measures so that each study yielded one effect for posttest and one effect for follow-up outcomes. Separate analyses were conducted to examine the characteristics of outcome measures. Within each study, effect sizes were averaged within categories of the outcome method variables, so that each study yielded no more than one effect size for each category.

Between-groups and within-groups studies were analyzed separately, because of the potential for inflated within-group effects ( Lipsey & Wilson, 1993 ). In between-group studies with multiple comparison groups, all comparison groups were combined into a composite group. Examination of individual studies suggested that this decision rule did not materially affect the results of the meta-analysis.


Study Characteristics

The sample consisted of 59 between-group and 9 within-group studies. Of the between-group studies, 39 (66%) used random assignment, 13 (22%) had multiple comparison groups, 39 (66%) used a standardized treatment protocol, but only 26 (44%) explicitly reported using raters unaware of the study hypothesis or treatment assignment, and only 20 (34%) reported using experienced raters. Study sample sizes ranged from 12 to 133 participants for between-group studies (total N = 2,753), and from 7 to 286 for within-group studies (total N = 725). The mean sample size for the skills training treatment group was 23.8 in between-group studies and 69.2 in within-group studies. Control groups had a mean of 20.7 participants and alternate comparison groups had a mean of 14.8 participants. Fewer than half of the between-group studies, but seven of the nine within-group studies reported attrition rates. Attrition for between-group studies ranged from 0% to 58%, with a mean of 18%. The range for within-group studies was 5% to 73%, with a mean of 26%. Only 14 (24%) of the between-group studies and 4 (44%) of the within-group studies included follow-up measures. In most instances, the follow-up period was 3 months or less.
Training Program Characteristics

Training program characteristics are described in Table 1 . For most studies, the training period was brief. The training programs evaluated by the between-group studies were substantially less intensive than those evaluated by the within-group studies, as judged by both mean number of hours (17.1 vs. 61.0) and mean number of weeks (6.2 vs. 13.0) of training. Most training programs incorporated standard core behavior techniques (e.g., instruction, modeling, behavioral rehearsal, and feedback). Fewer than half, however, included a specific mechanism to promote generalization.

Although the literature search criteria explicitly included studies examining training in all types of skills, social skills training clearly predominated, with 81% of the between-group studies focusing on training either general interpersonal or assertiveness skills. Other skill areas included prevocational skills (five studies), activities of daily living (three studies), and micro-interpersonal, affective management, and cognitive skills (one study each). Within-group studies focused on general interpersonal skills (five studies) or activities of daily living (four studies). About two thirds of the studies were conducted in inpatient settings. Few of the inpatient studies included postdischarge information, and of those that did, most provided anecdotal information only.
Participant Sample Characteristics

Participant sample characteristics are presented in Table 1 . Client characteristics were not reported in a standardized fashion, and those characteristics that could be tabulated did not vary greatly across studies. Schizophrenia was the most widely represented diagnostic group, included in over four fifths of the studies. By contrast, few studies explicitly included participants with major depression or bipolar disorder (with 84% and 86%, respectively, of the studies reporting diagnostic information including no participants with these diagnoses). Of the 49 between-group studies reporting the gender distribution, almost half of the studies used all-male samples (mostly in Veteran Administration hospitals).
Overall Effect Sizes

For between-group studies, the overall effect size was medium at posttest ( d. = .40, n = 58, z = 9.18, p < .001) and large at follow-up ( d. = .56, n = 14, z = 5.34, p < .01). For within-group studies, the overall effect size was also medium at posttest ( d. = .48, n = 9, z = 7.59, p < .01), but small at follow up ( d. = .30, n = 4, z = 2.37, p < .05). Thus, the overall posttest effect size for the within-group studies was similar to that for the between-group studies, whereas the overall follow-up effect size was smaller for the within-group studies than the corresponding statistic for the between-group studies.

The test for homogeneity was examined for each of the four overall effect sizes at posttest and follow-up. Only the between-group posttest effect size was significant QT = 150.4, p < .01. Therefore, only subanalyses testing for potential moderator variables in between-group studies at posttest are reported later. (The corresponding subanalyses on the within-group studies and on follow-up effect sizes for the between-group studies yielded few significant findings, partly because of the small number of studies in these analyses.)
Moderating Variables in Between-Group Studies

Overall posttest effects were disaggregated relative to a number of methodological, training program, client, and outcome method variables, as shown in Table 3 , Table 4 , and Table 5 . The purpose of the subanalyses was twofold: to document the features of the body of studies reviewed and to determine which factors might explain differences in the effects of skills training.
Methodological characteristics.

No significant relationship was found between effect size and any coded methodological variable. Both published and unpublished studies yielded medium effects ( d. = .42, and d. = .38, respectively) that were not significantly different from one another QB = .20, ns. This finding is contrary to the popular assumption that published studies yield larger effect sizes than do unpublished studies. Similarly, year of publication was unrelated to effect size. Sample size was not significantly correlated with the unweighted effect size. Finally, a composite measure of methodological quality was not correlated with effect size ( r = .01, n = 58).
Training program characteristics.

The content focus of the training was examined, as shown in Table 3 . Mean effect size ranged from .30 (general interpersonal skill training) to .73 (prevocational skill training) among those areas represented by two or more studies. Studies of training in assertiveness skills had a significantly larger overall effect than studies of training in general interpersonal skills QB = 60.9, p < .01. Neither the use of a specific mechanism to promote generalization outside the treatment setting QB = .29, nsnor the incorporation of a cognitive—behavioral approach had any substantive impact on treatment outcomes QB = .92, ns. Similarly, both inpatient and community-based settings yielded medium effects, with no significant difference in the effect size, QB = .1.01, ns. Therefore, location of the training had no discernible impact on outcome. Six measures examining the intensity or comprehensiveness of the training approach are shown in Table 4 . Of these, only weeks of training was significantly correlated with effect size. Overall, variability in training characteristics contributed little to explaining differences between studies in the impact of the training.
Client characteristics.

No significant relationships were found between effect size and the study-level percentage of women or of clients with a specific diagnosis (e.g., schizophrenia); nor were significant relationships found between effect size and sample means for age, number of prior hospitalizations, or months of prior hospitalizations.
Outcome domains.

Not counting measures tailor made by researchers for their own studies, over 100 different instruments were used to measure outcomes. Testing significance between categories of outcome measures presents thorny statistical issues, because of the lack of consistency across studies in the categories of measures used and because the correlations between measures were usually not reported ( Gleser & Olkin, 1994 ; Raudenbush, Becker, & Kalaian, 1988 ). To circumvent these difficulties, we used a nonparametric statistical test. Within each outcome method domain, we used a series of sign tests to determine which effect size was larger across those studies that included two different outcome method categories. Only significant differences are reported later.

With regard to instrument standardization, researcher-developed instruments yielded the largest effects ( d. = .59), followed by nonstandardized instruments developed by other researchers ( d. = .38), and, finally, by standardized instruments ( d. = .25). Study-specific effect sizes were larger than effect sizes based on other-researcher measures in 18 (69%) of 26 studies, z = 1.96, p < .05, and larger than effect sizes based on standardized measures in 15 (79%) of 19 studies, z = 2.52, p < .05.

The generalizability of the outcome measure was also examined. As shown in Table 5 , a large effect was found for measures that were both similar to the training stimuli and administered in the same setting ( d. = .82), compared with a medium effect for measures similar to training but administered in a setting different from the treatment setting ( d. = .39), and a small effect for inferential measures ( d. = .25). Among the 25 studies measuring both same-setting measures and inferential measures, the same-setting measure effect size was larger in 22 (88%) studies, z = 3.80, p < .001.

Also shown in Table 5 , self-report was the most common measurement modality, representing roughly one third of all effect sizes. Role play and in vivo behavioral modalities were the next most common, each accounting for roughly one fifth of the effect sizes. Role play measures yielded a large effect ( d. = .94), in contrast to in vivo behavioral ratings ( d. = .20) and self-report measures ( d. = .28). Role-play effect sizes were larger than self-report effect sizes in 15 (88%) of 17 studies, z = 3.15, p < .01, and larger than effect sizes based on in vivo measures in 9 (90%) of 10 studies, z = 2.52, p < .05.


This meta-analysis was prompted by the contrast between optimistic predictions made over the past 2 decades about effective applications of skills training and the well-documented limitations of short-term rehabilitation approaches with this population ( Lehman, Thompson, Dixon, & Scott, 1995 ). The reality of everyday practice in community treatment settings does not square with the more expansive assertions sometimes made on behalf of skills training. If skills training does have the promise that proponents claim, is it merely a failure to implement a proven technology? Are discrepant conclusions a result of different reviews examining somewhat different subsets of primary studies? Are the methodological shortcomings of narrative reviews at fault? Or have some proponents overstated the case?

The intent of this meta-analysis was to minimize review biases by seeking an exhaustive data base, using a quantitative coding scheme and statistical analysis, and accounting for confounding variables. In terms of sheer quantity of studies, our meta-analysis is more exhaustive than any previous review. Most of what we found reinforced what previous reviews have already concluded. We did, however, uncover a couple patterns that were not so frequently noted.

In agreement with most other reviews, our overall conclusion is that behavioral skills training for persons with SMI is effective for teaching inpatients interpersonal and assertiveness skills, as indicated by measures of skill acquisition and symptom reduction. This set of findings is solidly grounded in a substantial number of experimental and quasi-experimental studies, many of which have rigorous designs, low attrition, careful attention to implementation, and thoughtful measurement of outcomes. We should not lose sight of either the quantity or methodological quality of studies in the skills training area, which stands in contrast to the body of research in most, if not all, other areas of psychiatric rehabilitation ( Lehman et al., 1995 ).

Encouragingly, effect sizes were little affected by methodological variables. Thus, we found no indication of methodological artifacts explaining program outcomes. Unlike most meta-analyses ( Lipsey & Wilson, 1993 ), we did not find larger effect sizes for within-group designs. The lack of difference is puzzling in that the within-group studies evaluated training approaches that were on average more intensive than those in the between-group studies, which would also lead to the prediction of larger effects for within-group studies. We cannot explain the lack of differences.

Having recognized the accomplishments of the skills training literature, we hasten to add that it mostly has been evaluated with a restricted set of circumstances. Most importantly, the current data base does not yet permit extrapolation to everyday functioning, which is, after all, the bottom line of skills training interventions. Many studies have rediscovered the same basic set of findings without extending the findings beyond a narrow scope. The limitations concern treatment settings, study populations, intensity of training, skill domains, outcome measurement, and time frame for examining outcomes. These limitations also help explain the lack of substantive moderator variables correlated with outcome within this body of studies.
Treatment Settings

One clear gap is the paucity of studies examining skills training in settings other than psychiatric hospitals. Moreover, few studies have systematically examined if such inpatient training improves postdischarge adjustment. Fortunately, there appears to be a trend in studies published in the past 5 years toward evaluating skills training in community settings. Also encouraging is the similarity of effect sizes for studies conducted in community settings, compared to inpatient settings.
Study Populations

The scope of the skills training literature is also surprisingly parochial in its overrepresentation of mostly male or all-male samples. Not only should researchers ensure adequate gender representation but also there is a need to study more diverse populations, including individuals from ethnic minority groups and other diagnostic groups besides schizophrenia. Skills training studies should also address frequently occurring comorbid conditions, such as substance abuse.
Intensity of Training

Few studies provided a fair test of the ecological validity of skills training. In most studies, the training period was short. Also, although many experts advocate intensively training skills in the settings in which clients actually use the skills ( Test & Stein, 1978 ), we did not find a single study to rigorously test this approach.
Skill Domains

Social skills training is clearly the predominant approach in the skills training literature. The few studies focusing on activities of living yielded effect sizes as least as large as the social domain. Nonetheless, strong claims about the effectiveness of skills training in other skills areas are premature.

Liberman (1988) has theorized that improved social functioning may lead to improved role functioning in areas such as work and independent living. The meta-analysis did not reveal that such generalization of learning occurs, however. Despite this widely held assumption, little research directly shows that social skills training results in improvement in role functioning. For example, the vocational rehabilitation literature has produced little evidence that social skills training facilitates competitive employment ( Bond, 1992 ). One recent study has suggested that incorporating classroom skills training as a prerequisite to entry into an employment program may, in fact, delay benefits from a more direct placement approach ( Drake, McHugo, Becker, Anthony, & Clark, 1996 ).
Outcome Measurement

One consistent influence on effect size was the type of outcome measure used. Effect sizes were substantially smaller when the assessment was made outside of the training setting or differed in form from the training exercises. This finding is reminiscent of other meta-analyses, which have shown that a measure's reactivity ( Smith, Glass, & Miller, 1980 ) or specificity ( Shadish et al., 1993 ) influences effect size. These findings suggest caution in interpreting the generalizability of large effects observed in the training setting itself.

Another pattern was the frequent use of outcome measures that were close approximations of exercises used in the training exercises. Although measuring outcomes that are theoretically linked to the intervention is sound research practice, many of the studies depended solely or primarily on such measures. Ideally, studies should balance their outcome packages with a mix of a range of outcome instruments, including standardized measures that provide reliable and valid information as to whether specific interventions are therapeutically and pragmatically beneficial (or not) in broader behavioral domains and functional contexts.
Time Frame for Examining Outcomes

Research is needed to demonstrate that, once obtained, skills are maintained over a period of time, particularly given the long-term nature of SMI. Relatively few studies addressed the question of durability of outcomes. Those that did typically used only one follow-up probe, typically only a few months after treatment termination. Follow-up probes did not always include the full outcome battery measured at posttest, further clouding interpretation of findings. For those studies that did attempt to address durability, we found, as did Benton and Schroeder (1990) , that follow-up effects were as large as those immediately posttreatment.
Future Directions

Since our review was completed, Penn and Mueser (1996) , Scott and Dixon (1995) , and Wallace (1995) have identified six more recent studies ( Bradshaw, 1993 ; Dobson, McDougall, Busheikin, & Aldous, 1995 ; Hayes, Halford, & Varghese, 1995 ; Marder et al., in press ; Tarrier, et al., 1993 ; Wallace, Liberman, MacKain, Blackwell, & Eckman, 1992 ). Including these recent studies would not substantively alter the estimation of overall effect sizes.

Although their results are mixed, these recent studies do suggest movement toward addressing several deficits that reviews repeatedly have identified. A prominent example is the dissemination of user-friendly skills training modules to enhance medication and symptom self-management in residential programs and other community settings ( Eckman & Liberman, 1990 ; Eckman, Liberman, Phipps, & Blair, 1990 ; Wallace et al., 1992 ).

Experimental studies are needed to determine whether increasing the intensity and duration of skills training enhances outcomes. Such studies would be especially relevant to managed care (Newman & Tejeda, 1996 ). Our findings hint at a dose—effect relationship, similar to findings in the psychotherapy literature ( Howard, Kopta, Krause, & Orlinsky, 1986 ). The lack of a stronger relationship in our meta-analysis may have been a result of a restriction of range problem, with the preponderance of studies examining time-limited training programs.

As research on skills training moves toward evaluating applications in the mainstream of community treatment practice, a logical next step is to compare them experimentally to other psychiatric rehabilitation approaches. At present, such studies are almost completely absent from the literature.



*Baenninger, L. P. & Tang, W. (1990). Teaching chronic psychiatric inpatients to use differential attention to change each other's behaviors.( Hospital and Community Psychiatry, 41, 425—429.)
*Balzer, F. J. (1974). The development of psychiatric inpatient peer-helpers through a systematic human relations training program (Doctoral dissertation, University of Georgia, 1974).( Dissertation Abstracts International, 35 (10), 5097B. (University Microfilms No. 7508086))
Becker, B. J. (1988). Synthesizing standardized mean-change measures.( British Journal of Mathematical and Statistical Psychology. 41, 257—278.)
*Bellack, A. S., Turner, S. M., Hersen, M. & Luber, R. F. (1984). An examination of the efficacy of social skills training for chronic schizophrenic patients.( Hospital and Community Psychiatry, 35, 1023—1028.)
Benton, M. K. & Schroeder, H. E. (1990). Social skills training with schizophrenics: A meta-analytic evaluation.( Journal of Consulting and Clinical Psychology, 58, 741—747.)
*Berryman, M. L. (1990). A training program to improve emotion recognition in schizophrenics (Doctoral Dissertation, Oklahoma State University, 1990).( Dissertation Abstracts International, 52 (02), 1048B. (University Microfilms No. 9119855))
Bond, G. R. (1992). Vocational rehabilitation.(In R. P. Liberman (Ed.), Handbook of psychiatric rehabilitation (pp. 244—275). New York: Macmillan.)
*Booraem, C. D. & Flowers, J. V. (1972). Reduction of anxiety and personal space as a function of assertion training with severely disturbed neuropsychiatric inpatients.( Psychological Reports, 30, 923—929.) Bradshaw, W. H. (1993). Coping-skills training versus a problem-solving approach with schizophrenic patients.( Hospital and Community Psychiatry, 44, 1102—1104.)
Brady, J. P. (1984a). Social skills training for psychiatric patients.(I: Concepts, methods, and clinical results. American Journal of Psychiatry, 141, 333—340.)
Brady, J. P. (1984b). Social skills training for psychiatric patients.(II: Clinical outcome studies. American Journal of Psychiatry, 141, 491—498.)
*Brown, M. A. (1982). Maintenance and generalization issues in skills training with chronic schizoprenics.(In J. P. Curran & P. M. Monti (Eds.), Social skills training (pp. 90—116). New York: Guilford Press.)
*Brown, M. A. & Munford, A. M. (1983). Life skills training for chronic schizophrenics.( Journal of Nervous and Mental Disease, 171, 466—470.)
*Brown, S. D. (1980). Coping skills training: An evaluation of a psychoeducational program in a community mental health setting.( Journal of Counseling Psychology, 27, 340—345.)
*Brown, S. D. (1983). Coping skills training: Attitude toward mental illness, depression, and quality of life 1 year later.( Journal of Counseling Psychology, 30, 117—120.)
*Carmichael, S. R. (1976). A comparison of three group treatment methods for low assertive patients (Doctoral dissertation, University of Texas at Austin, 1976).( Dissertation Abstracts International, 37, 2497B. (University Microfilms No. 7626606))
*Carter, M. (1990). Assertive behavior training with chronic mentally ill persons using a videotape format (Doctoral dissertation, University of North Dakota, 1990).( Dissertation Abstracts International, 52 (02), 1120B. (University Microfilms No. 9118354))
*Clark, K. W. (1974). Evaluation of a group social skills training program with psychiatric inpatients: Training vietnam era veterans in assertion, heterosexual, and job interview skills (Doctoral dissertation, University of Wisconsin, 1974).( Dissertation Abstracts International, 35 (09), 4642B. (University Microfilms No. 7428795))
*Clark, P. J. (1988). Social skills training to reduce aggressive behavior in chronic psychiatric inpatients (Doctoral dissertation, Brigham Young University, 1988).( Dissertation Abstracts International, 49 (07), 2847B. (University Microfilms No. 8820016))
*Coche, E. & Flick, A. (1975). Problem-solving training groups for hospitalized psychiatric patients.( Journal of Psychology, 91, 19—29.)
*Cohen, S. R. (1978). A comparison of the effects of interpersonal problem solving training groups and social interaction training groups on hospitalized psychiatric patients (Doctoral dissertation, Bryn Mawr College, 1978).( Dissertation Abstracts International, 42 (07), 2981B. (University Microfilms No. 8125520))
Cooper, H. & Hedges, L. V. (1994). Handbook of research synthesis. (New York: Sage)
Corrigan, P. W. (1991). Social skills training in adult psychiatric populations: A meta-analysis.( Journal of Behavior Therapy and Experimental Psychiatry, 22, 203—210.)
*Cuccaro, M. L. (1988). Cognitive retraining: An approach to the remediation of conceptual dysfunction in chronic schizophrenia (Doctoral dissertation, University of South Carolina, 1988).( Dissertation Abstracts International, 50 (02), 1744B. (University Microfilms No. 8910238))
Curran, J. P. (1985). Social skills therapy: A model and a treatment.(In R. M. Turner & L. M. Ascher (Eds.), Evaluating behavior therapy outcome (pp. 122—145). New York: Springer.)
*Daugherty, M. E. (1982). Social skills training with psychiatric patients and their spouses (Doctoral dissertation, University of Notre Dame, 1982).( Dissertation Abstracts International, 43 (06), 1973B. (University Microfilms No. 8225807))
*Davis, N. J. & Matthews, A. T. (1980). An integrated approach to social skills training for schizophrenic clients in a day treatment setting (Doctoral dissertation, Harvard University, 1980).( Dissertation Abstracts International, 41 (08), 3157B. (University Microfilms No. 8100343))
*DeLancey, A. L. (1979). The effects of covert reinforcement and social skills training with psychiatric patients (Doctoral dissertation, Boston College, 1979).( Dissertation Abstracts International, 40 (03), 1358. (University Microfilms No. 7920481))
*Denicola, J. A. (1980). The effects of method of presentation on acquisition of job interview skills in schizophrenic patients (Doctoral dissertation, University of South Florida, 1980).( Dissertation Abstracts International, 41 (06), 2315B. (University Microfilms No. 8028931))
Dion, G. L. & Anthony, W. A. (1987). Research in psychiatric rehabilitation: A review of experimental and quasi-experimental studies.( Rehabilitation Counseling Bulletin, 30, 177—182.)
Dobson, D. J., McDougall, G., Busheikin, J. & Aldous, J. (1995). Effects of social skills training and social milieu treatment on symptoms of schizophrenia.( Psychiatric Services, 46, 376—380.)
Donahoe, C. P. & Driesenga, S. A. (1988). A review of social skills training with chronic mental patients.(In M. Hersen, R. M. Eisler. P. M. Miller (Eds.), Progress in behavior modification (pp. 131—164). Newbury Park, CA: Sage.)
*Doty, D. W. (1972). Role-playing and incentives in the modification of the social interaction of chronic psychiatric patients (Doctoral dissertation, University of Illinois at Urbana—Champaign, 1972).( Dissertation Abstracts International, 34, 871B. (University Microfilms No. 7317186))
*Doty, D. W. (1975). Role playing and incentives in the modification of the social interaction of chronic psychiatric patients.( Journal of Consulting and Clinical Psychology, 43, 676—682.)
*Douglas, M. S. & Mueser, K. T. (1990). Teaching conflict resolution skills to the chronically mentally ill.( Behavior Modification, 14, 519—547.)
Drake, R. E., McHugo, G. J., Becker, D. R., Anthony, W. A. & Clark, R. E. (1996). The New Hampshire study of supported employment for people with severe mental illness.(I: Vocational outcomes. Journal of Consulting and Clinical Psychology, 64, 391—399.)
*Eckman, T. A. & Liberman, R. P. (1990). A large-scale field test of a medication management skills training program for people with schizophrenia.( Psychosocial Rehabilitation Journal, 13 (3), 31—35.)
*Eckman, T. A., Liberman, R. P., Phipps, C. C. & Blair, K. E. (1990). Teaching medication management skills to schizophrenic patients.( Journal of Clinical Psychopharmacology, 10, 33—38.)
*Eckman, T. A., Wirshing, W. C., Marder, S. R., Liberman, R. P., Johnston-Cronk, K., Zimmermann, K. & Mintz, J. (1992). Technique for training schizophrenic patients in illness self-management: A controlled trial.( American Journal of Psychiatry, 149, 1549—1555.)
*Eisler, R. M., Blanchard, E. B., Fitts, H. & Williams, J. G. (1978). Social skills training with and without modeling for schizophrenic and non-psychotic hospitalized psychiatric patients.( Behavior Modification, 2, 147—172.)
*Eisler, R. M., Hersen, M. & Miller, P. M. (1973). Effects of modeling on components of assertive behavior.( Journal of Behavior Therapy and Experimental Psychiatry, 4, 1—6.)
*Finch, B. E. & Wallace, C. J. (1977). Successful interpersonal skills training with schizophrenic inpatients.( Journal of Consulting and Clinical Psychology, 45, 885—890.)
*Frisch, M. B., Elliott, C. H., Atsaides, J. P., Salva, D. M. & Denney, D. R. (1982). Social skills and stress management training to enhance patients' interpersonal competencies.( Psychotherapy: Theory, Research, and Practice, 19, 349—358.)
*Giantonio, C. K. (1976). Assertion training: A social skill development technique with chronic schizophrenic inpatients (Doctoral dissertation, University of Notre Dame, 1976).( Dissertation Abstracts International, 37 (03), 1433B. (University Microfilms No. 7620094).)
Glass, G. V., McGaw, B. & Smith, M. L. (1981). Meta-analysis in social research. (Newbury Park, CA: Sage)
Gleser, L. J. & Olkin, I. (1994). Stochastically dependent effect sizes.(In H. Cooper & L. V. Hedges (Eds.), The handbook of research synthesis (pp. 339—355). New York: Sage.)
*Goldsmith, J. B. & McFall, R. M. (1975). Development and evalution of an interpersonal skill-training program for psychiatric inpatients.( Journal of Abnormal Psychology, 84, 51—58.)
*Gutride, M. E., Goldstein, A. P. & Hunter, G. F. (1973). The use of modeling and role playing to increase social interaction among asocial psychiatric patients.( Journal of Consulting and Clinical Psychology, 40, 408—415.)
*Gutride, M. E., Goldstein, A. P., Hunter, G. F., Carrol, S., Clark, L., Furia, R. & Lower, W. (1974). Structured learning therapy with transfer training for chronic inpatients.( Journal of Clinical Psychology, 30, 277—279.)
Halford, W. K. & Hayes, R. (1991). Psychological rehabilitation of chronic schizophrenic patients: Recent findings of social skills training and family psychoeducation.( Clinical Psychology Review, 11, 23—44.)
*Hansen, D. J., Lawrence, J. S. & Christoff, K. A. (1985). Effects of interpersonal problem-solving training with chronic aftercare patients on problem-solving component skills and effectiveness of solutions.( Journal of Consulting and Clinical Psychology, 53, 167—174.)
*Harpin, R. E., Liberman, R. P., Marks, I., Stern, R. & Bohannon, W. E. (1982). Cognitive—behavior therapy for chronically depressed patients.( The Journal of Nervous and Mental Disease, 170, 295—301.)
Hayes, R. (1989). Occupational therapy in the treatment of schizophrenia.( Occupational Therapy in Mental Health, 9, 51—68.)
Hayes, R. L., Halford, W. K. & Varghese, F. T. (1995). Social skills training with chronic schizophrenic patients. Effects on negative symptoms and community functioning.( Behavior Therapy, 26, 433—449.)
Hedges, L. V. (1984). Advances in statistical methods for meta-analysis.( New Directions for Program Evaluation, 24, 25—42.)
Hersen, M. & Bellack, A. S. (1976). Social skills training for chronic psychiatric patients: Rationale, research findings, and future directions.( Comprehensive Psychiatry, 17, 559—580.)
Hersen, M., Eisler, R. M. & Miller, P. M. (1973). Development of assertive responses: Clinical measurement and research considerations.( Behavior Research and Therapy, 11, 505—521.)
*Hersen, M., Eisler, R. M. & Miller, P. M. (1974). An experimental analysis of generalization in assertive training.( Behavior Research and Therapy, 12, 295—310.)
*Hersen, M., Eisler, R. M., Miller, P. M., Johnson, M. B. & Pinkston, S. G. (1973). Effects of practice, instructions, and modeling on components of assertive behavior.( Behavior Research and Therapy, 11, 443—451.)
*Hersen, M.,Kazdin, A. E., Bellack, A. S. & Turner, S. M. (1979). Effects of live modeling, covert modeling and rehearsal on assertiveness in psychiatric patients.( Behavior Research and Therapy, 17, 369—377.)
Herz, M. I. (1986). Toward an integrated approach to the treatment of schizophrenia.( Psychotherapy and Psychosomatics, 46, 45—57.)
*Hewitt, K., Wishart, C. & Lambert, R. (1981). Social skills training with chronic psychiatric patients.( British Journal of Occupational Therapy, 44, 284—285.)
*Holmes, M. R., Hansen, D. J. & St. Lawrence, J. S. (1984). Conversational skills training with aftercare patients in the community: Social validation and generalization.( Behavior Therapy, 15, 84—100.)
*Hopkins, L. M. (1985). Transfer effect of instruction, live modeling, reinforcement, and overt or covert rehearsal on assertive non-verbal behavior of hospitalized schizophrenics (Doctoral dissertation, Ball State University, 1984).

Copyright © 2000, Southern Development Group, S.A.  All Rights Reserved.
A Private Non-Profit Agency for the good of all, published in the UK & Honduras