Research on Religion-Accommodative Counseling: Review and Meta-Analysis Michael E. McCullough National Institute for Healthcare Research The present meta-analysis examined data from 5 studies (N = 111) that compared the efficacy of standard approaches to counseling for depression with religion-accommodative approaches. There was no evidence that the religion-accommodative approaches were more or less efficacious than the standard approaches. Findings suggest that the choice to use religious approaches with religious clients is probably more a matter of client preference than a matter of differential efficacy. However, additional research is needed to examine whether religionaccommodative approaches yield differential treatment satisfaction or differential improvements in spiritual well-being or facilitate relapse prevention. Given the importance of religion to many potential consumers of psychological services, counseling psychologists should devote greater attention to religion-accommodative counseling in future studies. The United States is a highly religious country; 92% of its population are affiliated with a religion (Kosmin & Lachman, 1993). According to a 1995 survey, 96% of Americans believe in God or a universal spirit, 42% indicate that they attend a religious worship service weekly or almost weekly, 67% indicate that they are members of a church or synagogue, and 60% indicate that religion is "important" or "very important" in their lives (Gallup, 1995). In addition, many scholars acknowledge that certain forms of religious involvement are associated with better functioning on a variety of measures of mental health. Reviews of this research (e.g., Bergin, 1991; Bergin, Masters, & Richards, 1987; Larson et al., 1992; Pargament, 1997; Schumaker, 1992; Worthington, Kurusu, McCullough, & Sandage, 1996) suggested that several forms of religious involvement (including intrinsic religious motivation, attendance at religious worship, receiving coping support from one's religious faith or religious congregation, and positive religious attributions for life events) are positively associated with a variety of measures of mental health. For example, various measures of religious involvement appear to be related to lower degrees of depressive symptoms in adults (Bienenfeld, Koenig, Larson, & Sherrill, 1997; Ellison, 1995; Kendler, Gardner, & Prescott, 1997) and children (Miller, Warner, Wickramaratne & Weissman, 1997) and less suicide (e.g., Comstock & Partridge, 1972; Kark et al., 1996; Wandrei, 1985). Koenig, George, and Peterson (1998) reported that depressed people scoring high on measures of intrinsic religiousness were significantly more likely to experience a remission of depression during nearly a 1-year follow-up than were depressed people with lower intrinsic religiousness, even after controlling for 30 potential demographic, psychosocial, and medical confounds. Other studies have Correspondence concerning this article should be addressed to Michael E. McCullough, National Institute for Healthcare Research, 6110 Executive Boulevard, Suite 908, Rockville, Maryland 20852. Electronic mail may be sent to Mike@nihr.org. shown that religious involvement, as gauged through singleitem measures of frequency of religious worship and private prayer as well as more complex measures of religious coping, is related to positive psychological outcomes after major life events (e.g., Pargament et al., 1990; Pargament et al., 1994; Pargament, Smith, & Brant, 1995). This is the case even though several patterns of religious belief and religious coping (e.g., the belief that one's misfortunes are a punishment from God) are associated with greater psychological distress (Pargament, 1997). Religion in Counseling and Psychotherapy Some scholars (e.g., Bergin, 1991; Payne, Bergin, & Loftus, 1992; Richards & Bergin, 1997; Shafranske, 1996; Worthington et al., 1996) posited that considering clients' religiousness while designing treatment plans might have an important effect on the efficacy of treatment. Surveys of psychiatrists (Neeleman & King, 1993), psychologists (Bergin & Jensen, 1988; Shafranske & Malony, 1990), and mental health counselors (Kelly, 1995) also indicate that many mental health professionals believe that religious and spiritual values can and should be thoughtfully addressed in the course of mental health treatment. Moreover, a variety of analogue and clinical studies (e.g., Houts & Graham, 1986; T. A. Kelly & Strupp, 1992; Lewis & Lewis, 1985; McCullough & Worthington, 1995; McCullough, Worthington, Maxey, & Rachal, 1997; Morrow, Worthington, & McCullough, 1993) indicate that clients' religious beliefs can influence both (a) the conclusions of clinicians' structured psychological assessments and (b) the process of psychotherapy (cf. Luborsky et al., 1980). Evidence From Comparative Efficacy Studies Given the existing research on religion and mental health, an important question for counseling psychologists is whether supporting clients' religious beliefs and values in a structured treatment package yield clinical benefits that are equal to or greater than standard methods of psychological prac 92 RELIGIOUS COUNSELING META-ANALYSIS 93 tice. Several empirical studies have addressed this issue. Although the findings of studies that have examined such questions have been reviewed in narrative fashion elsewhere (e.g., W. B. Johnson, 1993; Matthews et al., 1998; Worthington et al., 1996), no researchers have used metaanalytic methods to estimate quantitatively the differential efficacy of such treatments. Metaanalytic reviews that compare religious approaches to counseling with standard approaches to counseling are one of three meta-analytic strategies that can be used to examine whether a given therapeutic approach has therapeutic efficacy (Wampold, 1997). In the present article, I review the existing research on such religious approaches to counseling using quantitative methods of research synthesis (e.g., Cooper & Hedges, 1994; Hunter & Schmidt, 1990) to estimate the differential efficacy of religious approaches in comparison to standard forms of counseling for depressed religious clients. Method Literature Search The PsycLIT, PsycINFO, Medline, ERIC, and Dissertation Abstracts electronic databases were searched through August 1998 for published and unpublished studies that examined the differential efficacy of a religion-accommodative approach to counseling in comparison to a standard approach to counseling. The reference sections of relevant articles were searched for other studies that would be relevant to this review. This search process continued until no new studies were revealed. In addition, several experts in the field of religion and mental health were contacted to identify unpublished studies. Studies had to meet four criteria to be included in the metaanalytic sample: They had to (a) compare a religion-accommodative approach to counseling to a standard approach to counseling; (b) randomly assign patients to treatments; (c) involve patients who were suffering from a specific set of psychological symptoms (e.g., anxiety or depression); and (d) offer equal amounts of treatment to clients in the religion-accommodative and standard treatments. Five published studies and one unpublished dissertation (W. B. Johnson, 1991), which was later reported in W. B. Johnson, DeVries, Ridley, Pettorini, and Peterson (1994), met these inclusion criteria. Several studies that investigated religious approaches to psychological treatment (e.g., Azhar & Varma, 1995a, 1995b; Azhar, Varma, & Dharap, 1994; Carlson, Bacaseta, & Simanton, 1988; Richards, Owen, & Stein, 1993; Rye & Pargament, 1997; Toh & Tan, 1997) were obtained, but these studies failed to meet all four inclusion criteria. Thus, they were omitted from the metaanalytic sample. A single rater determined which studies met inclusion criteria. This rater's decisions were made without reference to the results or discussion sections of the articles. The resulting meta-analytic sample included five studies representing data from 111 counseling clients. Descriptions of study populations, measures used, and effect size estimates (with 95% confidence intervals) are given in Table 1. The Studies Researchers interested in accommodative forms of religious counseling have taken standard cognitive-behavioral protocols or specific techniques, such as cognitive restructuring (Beck, Rush, Shaw, & Emery, 1979), cognitive coping skills (Meichenbaum, 1985), and appeals to rational thinking (e.g., Ellis & Grieger, 1977), and have developed religion-friendly rationales for and versions of such protocols or techniques (W. B. Johnson & Ridley, 1992b). These adapted protocols or techniques are thought to be theoretically equivalent to standard cognitivebehavioral techniques (Propst, 1996), but more amenable to the religious world view and religious language that religious clients use to understand their lives and their problems. The five studies are described in greater detail next. Propst (1980). Propst (1980) examined the differential efficacy of a manualized, religion-accommodative approach to cognitive restructuring and imagery modification. Volunteers who scored in the mild or moderate range of depression on the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) and in at least the moderate range on the King and Hunt (1972) religion scales were randomly assigned to one of two treatments. The standard treatment was an integration of Beck's (1976) cognitive therapy for depression and Meichenbaum's (1973) cognitive-behavior modification. During eight 1-hr sessions conducted over 4 weeks, clients were trained to observe their cognitions and imagery during depressed moods. After clients were convinced of the links between their moods, thoughts, and images, they practiced cognitive restructuring skills for modifying their thoughts and images using imagery and positive self-statements (e.g., "I can see myself in the future coping with that particular situation"). Ten of eleven clients assigned to this condition completed it. In the religion-accommodative treatment, clients completed the same therapeutic protocol as that used in the standard treatment. The only difference is that participants were trained to replace their negative cognitions and imagery with religious images (e.g., "I can visualize Christ going with me into that difficult situation in the future as I try to cope"). Seven of 9 clients assigned to this condition completed the treatment. Pecheur and Edwards (1984). Pecheur and Edwards (1984) assessed the differential efficacy of Beck et al.'s (1979) cognitive therapy for depression and a religion-accommodative version of the same therapy. Clients were students from a Christian college who met research diagnostic criteria for major depressive disorder They also scored in the depressed range on the BDI, the Hamilton Rating Scale for Depression (HRSD; Hamilton 1960), and a single-item visual analogue scale. In the standard treatment, clients completed eight 50-min sessions of cognitive behavior modification. All 7 clients who were assigned to this treatment completed it. In the religion-accommodative treatment, clients completed the standard cognitive therapy tasks specified in Beck et al. (1979); however, challenges to negative cognitions were placed in a religious context. For example, rather than replacing negative views of self with statements such as "Our self-acceptance and self-worth are not lost or lessened when we fail," the religionaccommodative approach trained clients to use self-statements such as, "God loves, accepts, and values us just as we are." This treatment was also administered according to a manual, which appears in Pecheur (1980). Propst, Ostrom, Watkins, Dean, and Mashburn (1992). Propst et al. (1992) compared the efficacy of Beck et al.'s (1979) cognitive therapy for depression with a manualized, religion-accommodative version of the same therapy (see Propst, 1988). Clients were recruited from the community and scored at least 14 on the 28-item version of the HRSD. They also scored at least in the moderate range on standard measures of religious commitment (e.g., Allport & Ross, 1967; King & Hunt, 1972). Clients in the standard treatment completed 18 sessions of individual cognitive therapy for depression. All 19 clients enrolled in this condition completed it. In the religionaccommodative treatment, clients completed 18 sessions of cognitive therapy that challenged negative cognitions and images by replacing them with positive thoughts and imagery of a religious nature, as in Propst (1980). All 19 clients enrolled in this condition completed it. W. B. Johnson & Ridley (1992a). Johnson and Ridley (1992) compared the efficacy of rational-emotive therapy (RET), using Walen, DiGiuseppe, and Wessler's (1980) treatment manual, with a manualized, religion-accommodative version of the same therapy. Clients were theology students and local church members who scored in at least the mildly depressed range on the BDI. They also scored in the "intrinsic" range on a standard measure of religious motivation (Allport & Ross, 1967), suggesting that their religious faith was highly internalized. In the standard RET condition, clients completed six 50-min sessions in 3 weeks, including homework sessions and in-session rehearsal of rational-emotive techniques. All 5 clients assigned to this condition completed it. In the religion-accommodative treatment, three explicitly Christian treatment components were added. First, clients were directed to dispute irrational beliefs using explicitly Christian beliefs, as in Propst (1980). Second, clients were encouraged to use Christian prayer, thoughts, and imagery in their homework assignments. Third, counselors used brief prayers at the end of each session. All 5 clients assigned to this condition completed it. W. B. Johnson et al. (1994). W. B. Johnson et al. (1994) compared the efficacy of standard RET and a religionaccommodative form of RET, as in W. B. Johnson and Ridley (1992a). Selection criteria were almost identical to those reported in W. B. Johnson and Ridley (1992a). The standard RET condition was an eight-session protocol delivered over 8 weeks, and was based on two popular RET treatment manuals (Ellis & Dryden, 1987; Walen et al., 1980). All 16 clients assigned to this condition completed it. The religion-accommodative treatment was based on two treatment manuals discussing Christian versions of RET (Backus, 1985; Thurman, 1989). Although the basic structure of RET was kept intact, clients were encouraged to dispute irrational beliefs based on scriptural beliefs and biblical examples. Homework assignments also used biblical examples and beliefs. All 16 clients assigned to this condition completed it. Effect Size Estimates Effect sizes and homogeneity statistics were calculated from means and standard deviations using the DSTAT statistical software, Version 1.10 (B. T. Johnson, 1989), using the formulas prescribed by Hedges and Olkin (1985). Effect sizes were based on the difference between the mean of clients in the standard counseling condition and the mean of clients in the religionaccommodative conditions. This difference was divided by the pooled standard deviation of clients in both conditions. All effect size estimates, expressed as d+ values, are corrected for the bias that is present in uncorrected g values, as recommended by Hedges and Olkin (1985). Effect sizes can be interpreted as the increased amount of symptom reduction afforded to participants in the religion-accommodative condition, expressed in standard deviation units. In calculating aggregate effect size estimates, individual effect sizes were weighted by the inverse of their sampling error variance, so that studies with larger samples were given greater weight in the calculation of d+ (Hedges & Olkin, 1985). The Q statistic was also used to estimate the degree of variability among the effect sizes. The Q statistic is basically a goodness-of-fit statistic with a roughly x2 distribution that enables a test of the hypothesis that all observed effect sizes were drawn from the same population. Significant Q values imply a heterogeneous set of effect sizes (Hunter & Schmidt, 1990). Handling Multiple Dependent Measures All five studies used the BDI as a dependent measure of depression. Although two of the studies also used the HRSD or a single-item visual analogue measure of depression, or both (Pecheur & Edwards, 1984; Propst et al., 1992), effect size estimates were based exclusively on the BDI for three reasons. First, the BDI has been shown to produce conservative effect size estimates in comparison to rating scales that are completed by clinicians, such as the HRSD (Lambert, Hatch, Kingston, & Edwards, 1986). Second, single-item visual analogue measures of depression (e.g., Aitken, 1969) appear to contain remarkably little true score variance (Faravelli, Albanesi, & Poli, 1986). Third, the aggregation of data across multiple dependent measures requires knowing their intercorrelations, which were not available for all five studies. Thus, the individual and mean effect size estimates reported here can be considered to be somewhat conservative. Handling Data From Multiple Follow-Up Periods All five studies collected follow-up data within 1 week of the termination of the trial. Although three of the studies (W. B. Johnson et al., 1994; Pecheur & Edwards, 1984; Propst et al., 1992) also reported follow-up data collected between 1 and 3 months after the termination of the trial, and one study (Propst et al., 1992) reported an effect size for a 24-month follow-up, we based our effect size estimates only on the data from the 1-week follow-up. Other Problems With Coding Effect Sizes Some studies reported data on additional experimental conditions, including self-monitoring and therapist contact conditions (Propst, 1980), waiting list control conditions (Pecheur & Edwards, 1984; Propst et al., 1992), and pastoral counseling conditions (Propst et al., 1992). Because none of these conditions were relevant to the central goal of this study, these data were neither coded nor included in the present meta-analytic study. RELIGIOUS COUNSELING META-ANALYSIS 95 Two other problems arose in coding effect sizes. First, although Propst (1980) reported posttreatment means on the BDI for both conditions, standard deviations were not reported. On the basis of the assumption that the other four studies in the present metaanalysis would yield similar pooled standard deviations for the BDI, a mean standard deviation for posttest scores on the BDI from these studies (5.81) was used as an imputed standard deviation for Propst (1980). This imputed standard deviation produced a nonsignificant test statistic for the comparison of the religious and standard counseling conditions, as Propst (1980) reported, giving us confidence that our imputed standard deviation was not wholly inaccurate. Second, Propst et al.'s (1992) results reported treatment effects separately for religious and nonreligious therapists, which was an independent factor in their experimental design. To collapse treatment effects across levels of the therapist religiousness factor, means and standard deviations obtained for religious and nonreligious therapists within each of the two religious counseling conditions were pooled before calculating an effect size for the treatments. Corrections of Findings for Unreliability in Dependent Measures Scholars in meta-analysis advise that effect size estimates be corrected for biases (Hunter & Schmidt, 1990, 1994). One of the easiest biases to correct is attenuation resulting from unreliability in the dependent variable. This bias can be corrected by dividing observed effect sizes and standard errors by the square root of the internal consistency of the dependent variable. Because metaanalytic estimates of the BDI's internal consistency were readily available (Beck, Steer, & Garbin, 1988, estimated its internal consistency at a = .86), the observed mean effect size and its confidence interval (CI) were divided by the square root of .86, or .927. Corrections for attenuation resulting from unreliability of the dependent variable produce increased effect size estimates but also a proportionate increase in confidence intervals; thus, a nonsignificant effect size will not become significant as a result of this correction (Hunter & Schmidt, 1994). Estimating Clinical Significance We were also interested in whether religion-accommodative and standard approaches to counseling yielded clinically significant differences in efficacy (Jacobson & Revenstorf, 1988; Jacobson & Truax, 1991). Thus, we calculated meta-analytic summaries of clinical significance for two studies that reported clinical significance data (using BDI > 9 as a cutoff for "mild clinical depression"; Kendall, Hollon, Beck, Hammen, & Ingram, 1987). Results Observed Mean Effect Size and Attenuation-Corrected Effect Size The mean effect size for the difference between religious and standard counseling during the 1week follow-up period (number of effect sizes = 5, N = 111) was d+ = +0.18 (95% CI: -.20/+0.56), indicating that clients in religionaccommodative counseling had slightly lower BDI scores at 1-week follow-up than did clients in standard counseling conditions. This effect size was not reliably different from zero (p = .34). The five effect sizes that contributed to this mean effect size were homogeneous, Q(4) = 5.38, p > .10. The mean effect size after correcting the effects for attenuation resulting from unreliability was d+ = +0.20 (95% CI: -0.19/+0.61). Differences in Clinical Significance Two studies (W. B. Johnson & C. R. Ridley, 1992a; Propst, 1980) reported the percentage of participants in the religious and standard psychotherapy conditions who manifested evidence of at least mild clinical depression (BDI scores >9) during the 1-week follow-up period. Aggregation of these data indicated that, among the 20 religionaccommodative counseling clients in the two studies, 4 (20%) were still at least mildly depressed at the end of treatment. Among the 26 standard counseling clients in the two studies, 9 (34.6%) were at least mildly depressed when treatment ended. This difference clinical significance was not statistically significant, x2(l, N = 46) = 1.19,p > .10. Discussion The goal of the present study was to review the existing empirical evidence regarding the comparative efficacy of religion-accommodative approaches to counseling depressed religious clients. These data suggest that, in the immediate period after completion of counseling, religious approaches to counseling do not have any significant superiority to standard approaches to counseling. Given that the differences in efficacy of most bonafide treatments are surprisingly small (e.g., Lambert & Bergin, 1994; Wampold, 1997), the existing literature on psychotherapy outcomes would have portended the present meta-analytic results. These findings corroborate some narrative reviews that claim equal efficacy for religion-accommodative and standard approaches to counseling (e.g., Worthington et al., 1996), and help to resolve the inconsistencies that others have observed among these studies (e.g., W. B. Johnson, 1993; Matthews et al., 1998). Although it is true that the religious approaches to counseling were no more effective than the standard approaches to counseling, it is equally true that they were no less effective than the standard approaches to counseling. Thus, the decision to use religion-accommodative approaches might be most wisely based not on the results of comparative clinical trials, which tend to find no differences among well-manualized treatments, but rather on the basis of patient choice (see Wampold, 1997). Not every religious client would prefer or respond favorably to a religionaccommodative approach to counseling. Indeed, the available evidence suggests that all but the most highly religious clients would prefer an approach to counseling that deals with religious issues only peripherally rather than focally (Wyatt & Johnson, 1990; see Worthington et al., 1996, for review). On the other hand, many religious clientsespecially very conservative Christian clientswould indeed be attracted to a counseling approach (or counselor) precisely because the counseling approach (or the counselot) main 96 McCULLOUGH tained that the clients' system of religious values were at the core of effective psychological change (Worthington, et al., 1996). The research reviewed herein indicates that no empirical basis exists for withholding such religionaccommodative treatment from depressed religious clients who desire such a treatment approach. The Last Word? There is inherent danger in publishing meta-analytic results. Because of their ability to provide precise-looking point estimates and short CIs (especially when the observed effect size estimates are relatively heterogeneous), metaanalytic summaries can be perceived to be the last word in evaluating research questions. It would be unfortunate if the present results were interpreted as the last word in evaluating the efficacy of religious approaches to counseling, however, because interesting and important questions remain. For example, although religion-accommodative approaches to counseling do not appear to be differentially efficacious in reducing symptoms (at least depressive symptoms), they might produce differential treatment satisfaction among some religious clients. Also, comparative studies of religion-accommodative therapy are needed with longer follow-up periods. It is possible that religion-accommodative approaches might prove to be superior to standard treatments in longer term follow-up periods, particularly in helping clients from relapsing, for example, back into depressive episodes. The differential effects of religionaccommodative and standard approaches to treatment also need to be investigated for a wider variety of disorders, including anxiety, anger, alcohol and drug problems, and marital and family problems. As well, although religionaccommodative and standard approaches to counseling do not appear to influence clients' religiousness or religious values differentially (Worthington et al., 1996), it is possible that religion-accommodative counseling yields differential improvements in religious clients' spiritual well-being. Finally, on a technical note, it should be noted that the studies in this body of literature currently have been seriously underpowered (i.e., in all cases fewer than 20 clients per treatment). This literature would benefit enormously from as few as three or four very high-quality, large-sample (i.e., 30 or more clients per condition) studies that investigated these questions in greater detail. W. B. Johnson (1993) provided other helpful methodological recommendations to which research on religionaccommodative counseling should adhere. Limitations The stability of meta-analytic findings comes from the number of studies included in the metaanalysis as well as the number of participants in the constituent studies. Thus, the findings from meta-analyses with small numbers of studies, such as the present study, are more easily overturned than meta-analyses that include larger numbers of studies. Although meta-analytic methods can be used to synthesize the results of as few as two studies (for examples of small-it meta-analyses, see Allison & Faith, 1996; Benschop et al., 1998; Kirsch, Montgomery, & Sapirstein, 1995; Uchino, Cacioppo, & Kiecolt-Glaser, 1996), our findings would obviously be considered more trustworthy if more studies had been available. A second limitation of the present findings relates to the nature of the meta-analytic sample. The five studies reviewed herein all investigated religion-accommodative counseling with depressed Christian clients. We can only speculate whether the present pattern of results would generalize to different religious populations or to people with different sets of presenting problems. Obviously, research is needed to fill in such gaps. Conclusion A variety of empirical data now suggest that certain forms of religious involvement can help prevent the onset of psychological difficulties and enhance effective coping with stressors. In addition, the majority of mental health professionals and the general public believe that patients' religious beliefs should be adequately assessed and taken into consideration in mental health treatment. Moreover, data indicate that patients' religious commitments can play a substantial role in counseling processes (Worthington et al., 1996). Data from the present study also indicate that religious approaches to counseling can be as effective as standard approaches to counseling depressed persons. Thus, for some clients, particularly very religious Christian clients, religionaccommodative approaches to counseling could be, quite literally, the treatment of choice. It is hoped that the present study will encourage counseling psychologists to examine whether religion-accommodative approaches yield similar or even superior benefits on other important metrics of therapeutic change and with other common difficulties in living. Your thread must be at least 300 words Discussion Board Forum 2: Religious Counseling Meta-Analysis (Module/Week 7) Preliminary Reading: McCullough, M.E. (1999). Research on religion-accommodative counseling: Review and meta-analysis. Journal of Counseling Psychology, 46(1), p. 92-98. Full text available through the Jerry Falwell Library. Log in to access the article for free. Thread Prompt: As you have been studying this module/week, a meta-analysis is a study that statistically combines the results of several studies on a similar topic. By calculating an overallmean effect size, meta-analyses help provide information on the effects of treatments over and above the individual studies themselves. McCullough's meta-analysis examines several studies that compare Christian counseling techniques to standard, non-religious counseling techniques. Imagine you are working in a community clinic with a variety of counselors, and you are interested in alerting the clinic supervisors to this meta-analysis and possibilities for future research that could be conducted in your clinic. Write a professional email to your supervisor and include the following required information: 1. Briefly describe how many studies are represented in this meta-analysis and what the main findings of the meta-analysis were. 2. In your opinion, why is this meta-analysis important in the context of Christian counseling, even though the results are not significant? Support your answer with information from the article. 3. Propose a study: In the meta-analysis discussion section, the subsection entitled \"The Last Word\" contains 4-5 suggestions for further studies in Christian counseling. Choose 1 that interests you and briefly propose a quantitative study to investigate this topic. The description of your proposed study must include: a) Your research question; b) A short description of the target participants (gender, age, diagnosis or problem area, etc.); c) The independent and dependent variables, and how many levels/groups the independent variable would have (review Module/Week 1 if necessary); 4. The type of outcome measure to serve as your dependent variable, as well as its level of measurement (nominal, ordinal, scale?see Module/Week 1). The measure could be a questionnaire, self-report, therapist rating of illness severity, number of diagnostic criteria met, etc. This can be something you come up with yourself. You do not have to name an actual existing instrument unless you know of one. 5. Include references in current APA style for any sources, including the meta-analysis