Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/232466088

Theeffectivenessofpsychodramatictechniques:Ameta-analysisARTICLEinGROUPDYNAMICSTHEORYRESEARCHANDPRACTICEFEBRUARY2003ImpactFactor:0.88DOI:10.1037/1089-2699.7.1.13

CITATIONS29

READS1,166

2AUTHORS,INCLUDING:

TimothyD.RitchieSaintXavierUniversity34PUBLICATIONS305CITATIONS

SEEPROFILE

Allin-textreferencesunderlinedinbluearelinkedtopublicationsonResearchGate,lettingyouaccessandreadthemimmediately.

Availablefrom:TimothyD.RitchieRetrievedon:05February2016

The Effectiveness of Psychodramatic Techniques: A Meta-Analysis

David A. KipperRoosevelt University

Timothy D. RitchieChicago, Illinois

Questions concerning the scientific basis of the clinical application of psychodramatictechniques have been raised primarily because of the infrequency of research publica-tions that validate the clinical observations. A meta-analysis conducted on the basisof 25 experimentally designed studies showed an overall effect size that points to alarge size improvement effect similar to or better than that commonly reported forgroup psychotherapy in general. The techniques of role reversal and doubling emergedas the most effective interventions. Of the 4 techniques investigated, 3 were signifi-cantly different from each other. There was no difference between the techniqueseffectiveness when used with clinical versus student populations or between their usein single versus multiple sessions. The meaning of the findings is discussed.

As a therapeutic modality invented more thanseven decades ago, psychodrama made a signif-icant impact on the development of group psy-chotherapy. Its influence was particularly greatin the early years of the 20th century. Then,together with the psychoanalyticdynamic ori-entation, it dominated the field of group psycho-therapy (Fuhriman & Burlingame, 1994). In theensuing years, its influence in the field of grouppsychotherapy in the United States diminished.Yet along with the observed decline in NorthAmerica, its popularity soared elsewhere, suchas in Europe, South America, and the Far East.The attraction of psychodrama has been attrib-uted in part to its powerful impact as seen inclinical practice. As to the observed decline inthe popularity of psychodrama in North Amer-ica, some have speculated that it is due to un-certainty concerning the research base of psy-chodramatic interventions and their scientificvalidity.

Critics have repeatedly raised the issue of thepaucity of empirical research that supportsclaims for the effectiveness of psychodrama.For example, DAmato and Dean (1988) wrotethe following:

Since its introduction, there has been a clear dichotomybetween believers and nonbelievers in PDs [psycho-dramas] methods. . . . The conflict revolves around thebasic assumptions of PD and the fact that the methodsare not empirically based. Psychodrama as a therapeu-tic technique continues to be practiced internationally,and authors continue to publish case studies and theo-retical discussions. But few empirical evaluations ofthe method have appeared since its introduction. (p.305)

Indeed, practically all of the books on psycho-drama published in the past 30 years have con-tained ample descriptions of anecdotal experi-ences, clinical experiences, and case illustra-tions. Summaries of empirical research,however, have been conspicuously missing(e.g., Blatner, 1996, 2000; Clayton, 1993;Farmer, 1995; Fox, 1987; Goldman & Morri-son, 1984; Holmes, 1992; Holmes & Karp,1991; Holmes, Karp, & Watson, 1994; Keller-mann, 1992; Kellermann & Hudgins, 2000;Kipper, 1986; Leviton, 1977; Roine, 1997;Starr, 1977; Sternberg & Garcia, 1989; Wil-liams, 1989; Yablonsky, 1976).

It is true that in general, empirical data on theeffectiveness of psychodrama have been gener-ated infrequently and have often been based onmethodologically weak experimental designs.The explanation given for this trend has beenthat psychodramatists are clinical practitionersand hence are disinclined to be engaged in sci-entific research. J. L. Moreno, the founder ofpsychodrama, himself was not an avid advocateof the traditional scientific research. It has alsobeen suggested that the absence of courses onpsychodrama from academic curricula has been

David A. Kipper, School of Psychology, Roosevelt Uni-versity; Timothy D. Ritchie, research consultant, Chicago,Illinois.

Correspondence concerning this article should be ad-dressed to David A. Kipper, 919 North Michigan Avenue,Suite 3106, Chicago, Illinois 60611. E-mail: kipperda@aol.com

Group Dynamics: Theory, Research, and Practice Copyright 2003 by the Educational Publishing Foundation2003, Vol. 7, No. 1, 1325 1089-2699/03/$12.00 DOI: 10.1037/1089-2699.7.1.13

13

a factor contributing to the slow rate of empir-ical studies. This said, however, it would beinaccurate to conclude that there is absolutelyno research concerning the effectiveness of psy-chodrama and its primary techniques (most no-tably, role reversal and doubling). Scant as theyare, the available studies provide an initial pic-ture about the potential usefulness of this inter-vention modality.

In the past 23 years, four reviews of psycho-drama research have appeared in English(DAmato & Dean, 1988; Kellermann, 1982;Kipper, 1978; Rawlinson, 2000). Using variousqualitative criteria, these articles reviewed anumber of selected studies that fitted their in-clusion requirements. Two of these four reviews(Kellermann, 1987; Kipper, 1978) focused onexperimentally controlled studies. The othertwo adopted a broader scope and included con-trolled studies; outcome studies using single-group designs (i.e., prepost treatment improve-ment as well as case illustrations).

In general, all four reviews came to a similarconclusion, namely, that although the initialempirical research on the effectiveness of psy-chodrama revealed some encouraging results,the data were insufficient and often lackedmethodological rigor. There were, however, dif-ferent nuances to this conclusion, and the re-viewers comments reflected some disparity ofviews. For instance, DAmato and Dean (1988)voiced the most critical opinion. Their basicview was that studies have been built on un-tested assumptions, and many of the findingsare questionable (DAmato & Dean, 1988, p.311). Kellermann (1982), on the other hand,offered the most positive view. He thought thatin principle, the evidence was supportive ofpsychodrama. He stated,

Although the above studies are so limited in scope thatany generalization of their findings must be very ten-uous, they do indicate that psychodrama is a validalternative to other therapeutic approaches, primarilyin promoting behavior change with adjustment, antiso-cial and related disorders. (Kellermann, 1982, p. 467)

Rawlinsons (2000) overall assessment wasthat there is some research evidence to support the useof psychodrama. This includes its use as a tool forhelping people to develop self-esteem, to change ele-ments of their behavior and to develop empathy andsocial relationships. However, in the present climate,this may not amount to the level of evidence whichwould be required to present a strong case as a cost

effective treatment for the range of mental health prob-lems currently dealt with in other ways. (p. 93)

Kipper (1978) concluded that although psycho-drama research pointed to the existence of asound basis for empirical validation of psycho-drama, more and better research was needed.

Arguably, the above four reviews have twolimitations. One concerns the method chosenfor assessing the validity of psychodrama. Allfour resorted to a qualitative method of analysisthat Glass (1976) called a secondary level ofanalysis. This type of analysis refers to reviewsof data collected by someone else who mayhave had aims and theoretical positions thatdiffer from those of the person conducting theanalysis. A less subjective evaluation would bethe third level of analysis, namely, meta-analy-sis (Glass, 1976). This is a procedure based onstatistical integration of the outcomes of severalindependent studies. It quantifies the effective-ness of treatments by creating a formula thatproduces a common measuring standard knownas effect size (Mullen, Driskell, & Salas, 1998;Mullen & Rosenthal, 1985). The effect sizeprovides an index of the average change thatcan be expected from the average recipient of aparticular treatment. To the best of our knowl-edge, psychodrama research has never been ex-amined by means of meta-analysis. In thepresent study, therefore, we pursue the questionof the scientific validity of psychodrama usingthis type of analysis.

The other limitation is that previous reviewshave not differentiated between studies of theeffectiveness of the entire psychodrama proce-dure (session) and studies concerned with theeffectiveness of individual psychodramatic tech-niques. One may view the psychodramatic pro-cedure as one unit that involves a complexpsychotherapeutic intervention based on a par-ticular sequence of a three-stage procedure, por-traying several scenes, and the application of anumber of techniques used multiple times withseveral group members as helpers or auxiliaries(Kellermann, 1987). Alternatively, one mayview psychodrama as a compendium of tech-niques. The latter definition is more experimen-tal-research friendly and is based on the notionthat the application of such special techniquesaccounts for much of the curative aspect ofpsychodrama (Kipper, 1988b). In the presentmeta-analysis, we differentiate between studies

14 KIPPER AND RITCHIE

that followed these two views and investigatethem both.

Method

DefinitionStudies that were selected to be part of the

present meta-analysis conformed to the defini-tion of psychodrama as a method that uses dra-matizations of personal experiences throughrole-playing and enactment under a variety ofsimulated conditions, which include at least onescene and one psychodramatic technique (Kip-per, 1978).

On the basis of this definition, we identifiedfour groups of studies that used psychodramatechniques. Three of these pertained to investi-gations of single psychodramatic techniquesnamely, role reversal, role-playing, and dou-bling (as described below). One group referredto studies that used multiple techniques.

Literature Retrieval

The studies selected for the present meta-analysis were located by computer (e.g.,through PsycLIT and Social Sciences Index) aswell as by examining published and unpub-lished psychodrama bibliographies (e.g., Sacks,Bilaniuk, & Gendron, 1995). The search fo-cused on studies published during the period19651999. Over 50 articles were included inthe original retrieval, but half were discardedeither because they pertained to sociometry re-search, a topic that was beyond the scope of thepresent study, or because they did not complywith the inclusion criteria. Finally, 25 studieswere admitted to the meta-analysis.

Selection Procedure

To be included, studies had to comply withthe following requirements. They needed to bewritten in English and published in a profes-sional, refereed journal. They had to be de-signed as controlled studiesthat is, composedof at least one experimental and one controlgroupand they had to contain sufficient sta-tistical information to allow the calculation ofeffect sizes.

Two independent raters selected the 25 stud-ies out of the initially retrieved pool (a high

interrater agreement of 92%). The disagreementregarding the discarded studies was attributed toconfusion as to whether articles featuring acombination of sociometry and action interven-tions should be included. In the end, the ratersdecided to exclude studies that essentially in-vestigated hypotheses related to sociometry.Similarly, there was a high degree of agreementbetween the two raters (88%) regarding thesorting of the selected studies into the fourtechniques (as discussed below). The disagree-ment was attributed to confusion with regard tothe difference in the definitions of role reversaland role-playing (enactment). Once the defini-tions were clarified, full agreement between theraters was reached.

Techniques

Table 1 describes the selected 25 studies assorted into the following four techniques.

Role reversal. This technique refers to en-actment that involves changing the initiallygiven role with that of another person who maybe present from the session or absent but rep-resented by an auxiliary (a helper). Typically,the protagonist begins the enactment by por-traying himself or herself with an auxiliary whoportrays the role of someone else. At some pointthe two reverse roles. The reversals need to lastquite a few minutes (310) and may be repeatedduring the enactment. This group included 10studies (D. W. Johnson, 1967, 1970, 1971;D. W. Johnson & Dustin, 1970; Kipper & Har-Even, 1984; Kipper, Har-Even, Rotenberg, &Dagan, 1982; Kipper & Ushpiz, 1987; Muney& Deutsch, 1968; Remer & Betts, 1998; Staven,1985).

Role-playing (enactment). This techniquerefers to the enactment of roles, whether spon-taneously or scripted, without the accompani-ment of role reversing, doubling, or any otherpsychodramatic technique. The actors remain intheir own identity during the entire enactmentwithout changing their identities. The duration ofrole-playing techniques may range from 5 to 15min. This group included four studies (Barrett,1986; Grandvold & Ollerenshaw, 1977; Kipper,1988a; Schutte, Malouf, & ODare, 1990).

Doubling. The doubling technique refers torole enactment in which one person (the protag-onist) portrays himself or herself and an auxil-iary assumes the protagonists persona, acting

15PSYCHODRAMA META-ANALYSIS

Table 1Characteristics of Studies Included in the Meta-Analysis

Source

Subjects SessionsMeasure of

dependent variableType Gen.Exp.group

Cont.group Total

Perweek

Role reversal

Johnson, 1967 S M 20(4) 1 1 Conflicts Solutions, a game/case self-report

Johnson, 1970 S M/F 10(3) 10 1 1 Attitude Questionaire, a game/case

Johnson, 1971 S M/F 17 17 1 1 Attitude Change InventoryJohnson & Dustin, 1970 S M/F 16(8) 1 1 Attitude Questionnaire, a

game/caseKipper & Har-Even, 1984 S M/F 12 13 1 1 Aggression Box,

questionnairesKipper et al., 1982 S F 18 18 1 1 Involvement Questionnaire

(original)Kipper & Ushpiz, 1987 S M/F 9 9, 9 1 1 Two questionnaires (originals)Muney & Deutsch, 1968 S M/F 50 50 1 1 Conflicts Solutions, a game/

case self-reportRemer & Betts, 1998 S M/F 98a 97a 4 1 Two questionnaires on

decision making and realismStaven, 1985 HP M/F 10 10, 10 1 1 Body-image test

Multiple

Carbonell & Parteleno-Barehmi,1999

SP F 6, 6 14 20 1 Youth Self-Report Form,qualitative analysis

Kipper & Giladi, 1978 S M/F 14, 10 12 14 1 Suinn Test AnxietyMartin & Stepath, 1993 HP M/F 53a 8 1 Group Behavior Assessment

Scale, narrative qualitativeanalysis

Schramski et al., 1984 PR M 66a 16 8 1 Correctional InstitutionalEnvironment Scale, HopkinsSymptom Checklist GroupEnvironment Scale

Slawson, 1965 HP M/F 27 27 b 2 Minnesota MultiphasicPersonality Inventory

Stallone, 1993 PR M 22a 22 5 1 Institutional disciplinary report

Role-playing

Barrett, 1986 SP c 19 16 12 1 Social-Emotional AssessmentInventory

Granvold & Ollerenshaw, 1977 S M/F 10 8 1 1 Social Avoidance and DistressScale, Fear of NegativeEvaluation, State/TraitAnxiety Inventory

Kipper, 1988a S M/F 15 15 1 1 Behavior Evaluation FormSchutte et al., 1990 S M/F 45 57 1 1 Social Consequence of

Smoking Scale, SmokingIntention/Behavior Scale

Doubling

J. A. Goldstein, 1971 HP M/F 6, 5 6 40 2 Taped text qualitative analysisS. G. Goldstein, 1967 HP 6 6 15 2 Qualitative analysis of taped

narratives and utterancesHudgins & Kiesler, 1987 S F 8 8 2 1 BarrettLennard Relationship

Inventory, Impact MessageInventory

16 KIPPER AND RITCHIE

alongside as a double. Doubling is a role por-trayed by a group member, an auxiliary, ratherthan by the protagonist. The duration of doubletechniques typically ranges from 3 to 15 min.There were five studies in this group (J. A.Goldstein, 1971; S. G. Goldstein, 1967; Hud-gins & Kiesler, 1987; Kipper & Ben-Ely, 1979;Taylor, 1983).

Multiple techniques. This category refers tostudies that compared the effect of at least onepsychodrama session that involved the use of anumber of psychodramatic techniques, includ-ing but not necessarily limited to the techniquesmentioned above. Customarily, a classical psy-chodrama session (Moreno, 1964) lasts for6090 min and is divided into three parts,which include the application of several psy-chodrama techniques. There were six studies inthis group (Carbonell & Parteleno-Barehmi,1999; Kipper & Giladi, 1978; Martin & Stepath,1993; Schramski, Feldman, Harvey, & Holi-man, 1984; Slawson, 1965; Stallone, 1993).

Role reversal and doubling are the two basictechniques of psychodrama. It is not surprising,therefore, that they constitute 60% of the 25studies that qualified for the meta-analysis. Thereasons Table 1 shows twice as many studies ofrole reversal than any other category are likelythat (a) role reversal is easy to study experimen-tally and (b) two investigators (i.e., Johnson andKipper) were particularly interested in this tech-nique and produced multiple studies.

Additional characteristics of the studies se-lected for the meta-analysis are shown in Ta-ble 1. The participants in the studies varied andwere drawn from both clinical and nonclinicalpopulations. Some studies used either males or

females, and some were composed of mixed-gender groups. All studies used experimentaldesigns with controls. About half of the studies(44%) featured interventions conducted overmultiple sessions. Of these, most involved oneweekly session. Table 1 also describes the mea-sures of the dependent variables used in theselected studies (see the far right column).These descriptions are included to provide ageneral sense of the kind of studies chosen, butthey were not part of the quantitative analysis.

Calculating the Effect SizeThe effect size was the basic unit of the

present calculations and concerned the relativeeffectiveness of four primary psychodramaticinterventions (techniques). Calculating the ef-fect size for each reported result of the indepen-dent variable was based on D-Stat Version 1.1.1for DOS (B. T. Johnson, 1995), which producedboth Hedgess g and Cohens adjusted d coef-ficients. To obtain descriptive and inferentialstatistics about effect size (Cohens adjusted d),we entered data from the D-Stat into SPSS 10.1for Windows 2001. Hedgess g coefficients tendto yield slightly higher effect sizes than themore conservative Cohens d (Cohen, 1992). Inthe present meta-analysis, we report only the dresults.

The standard formula used for the presentmeta-analysis for unequal group size was d Me Mc / within-group S, where Me and Mc arethe averages of the experimental and the controlgroups, respectively, and S is the pooled stan-dard deviations (Mullen & Rosenthal, 1985).For studies with equal size experimental and

Table 1 (continued)

Source

Subjects SessionsMeasure of

dependent variableType Gen.Exp.group

Cont.group Total

Perweek

Doubling (continued)

Kipper & Ben-Ely, 1979 S M/F 16 16, 16 6 1 Accurate Empathy Scale

Taylor, 1983 S M/F 19 19 1 1

Responding to videotapemessages, SemanticDifferential of EmotionalResponses

Note. Gen. gender; exp. experimental; cont. control; S students; M male; F female; HP hospitalizedpatients; PR prisoners/correctional inmates; SP special population.a Divided into small groups. b Multiple unspecified. c Gender was not specified.

17PSYCHODRAMA META-ANALYSIS

control groups, the equation used was d 2 /N (Hunter, Schmidt, & Jackson, 1982). Resultsreported in the form productmoment r werealso considered. Although r is a valid effect sizestatistic, it must be transformed into d in orderto maintain uniformity of effect size across thestudies. The formula used to this end was d 4 r2 / 1 r2 (Mullen & Rosenthal, 1985, p.135). Another common statistic used in psycho-drama research is t, which is also a valid effectsize statistic; however, it must be converted intor and then into d. The effect size variance wascomputed to determine the degree of variationsacross studies. It ought to be pointed out thatmean differences are stated in terms of standarddeviations, so that an effect size of 0.50 indi-cates a difference of one half of a standarddeviation. If all 25 studies had reported approx-imately the same effect size, then the variancewould be estimated to equal zero. Therefore,any effect size that exceeds the 00.50 range isconsidered a very positive outcome.

In the present study we sought to obtain an-swers to several questions. First, considering allthe studies together, do the results show a goodimprovement effect for psychodrama tech-niques? In other words, will the magnitude ofthe average effect size for all psychodramaticinterventions (NES 281) exceed the 0.50 mod-erate level (Cohen, 1992) cutoff point, therebyshowing moderate to large effect size? Second,are there significant differences among the fourtechniques? We were also interested in whetherthere was a statistically significant differencebetween (a) studies using student versus clinicalpopulations, (b) studies using single- versusmixed-gender groups, and (c) studies usingtechniques administered in one session versusmultiple sessions.

Results

The 25 studies included in the meta-analysisgenerated 281 unique effects about their de-pendent measures, with a combined sample ofthe participants of n 1,325 (M 53.00,SD 40.95). The sample size averages reportedhere came from total reported samples.

Overall EffectivenessThe main question concerned the overall ef-

fectiveness of psychodramatic techniques. Ta-

ble 2 lists the 25 studies reviewed, divided intothe four technique groups along with the num-ber of ds, their mean effect sizes, and theirstandard deviations.

Table 2 shows the average adjusted Cohensd for each of the 25 studies. The total averageadjusted d coefficient for all of the studies underinvestigation, 0.95 (SD 0.69), was signifi-cantly different from zero, t(280) 23.20, p .01, at the 99% confidence interval (lower 0.847, upper 1.061). This suggests the pres-ence of an overall moderate to large improve-ment effect size (i.e., greater than the commonly

Table 2Outcomes by Source, Technique, Mean Effect Size,and Standard Deviation

Source nd

Effect size

M SD

Role reversal

Johnson, 1967 10 0.82 0.31Johnson, 1970 17 1.72 0.94Johnson, 1971 9 0.83 0.32Johnson & Dustin, 1970 11 0.73 0.14Kipper & Har-Even, 1984 6 0.59 0.25Kipper et al., 1982 16 0.90 0.39Kipper & Ushpiz, 1987 7 1.30 0.92Muney & Deutsch, 1968 13 0.68 0.99Remer & Betts, 1998 8 0.32 0.30Staven, 1985 3 0.49 0.23

Multiple

Staven, 1985 8 0.66 0.33Kipper & Giladi, 1978 6 0.62 0.46Martin & Stepath, 1993 2 0.66 0.16Schramski et al., 1984 23 0.02 0.00Slawson, 1965 4 0.10 0.17Stallone, 1993 3 0.67 0.10

Role-playing

Barrett, 1986 2 0.02 0.28Granvold & Ollerenshaw, 1977 9 0.28 0.36Kipper, 1988a 1 1.75 0.00Schutte, Malouf, & ODare, 1990 5 0.71 0.33

Doubling

J. A. Goldstein, 1971 63 1.35 0.48S. G. Goldstein, 1967 33 1.26 0.45Hudgins & Kiesler, 1987 9 1.49 0.79Kipper & Ben-Ely, 1979 10 0.80 0.41Taylor, 1983 3 1.32 0.45

Total 281Adjusted d 0.95 0.69

18 KIPPER AND RITCHIE

regarded moderate effect size level of 0.50) forall of the psychodramatic techniques underinvestigation.

Effectiveness of the Specific TechniquesThe second question concerned the effective-

ness of each of the four techniques. Table 3shows the adjusted effect sizes for the fourtechniques, which represents the average of allof the effect sizes, to retaining the n of 281. Theresults show that the average adjusted effectsizes for each of the four techniques rangedfrom 1.29 for doubling to 0.16 for role-playing.The inflated mean (i.e., over ESd of 1.00) fordoubling suggests the influence of outliers (i.e.,occasional very high standard deviations thattend to skew the outcomes). In such an event,the median is often a better descriptive statisticthan the mean. Calculating the median (see Ta-ble 3) did not solve the problem for doubling,however, because the median effect size turnedout to be more inflated (ES 1.35). Again,when we used the 0.50 level as a guide forassessing the importance of the mean effectsizes, role reversal and doubling showed meanssuggesting large improvement effect size. Theresults for the multiple techniques were some-what below the cutoff point for moderate im-provement, with a mean ESd of 0.42 (MdnESd 0.35). At best it suggests an improve-ment effect that falls within the smallmoderaterange. The results for the technique of role-playing showed hardly any improvement effect(0.16).

Effectiveness of Techniques Relative toEach Other

The second question concerned the differ-ences among the four specific techniques. A

one-way analysis of variance of the estimatedeffect sizes revealed that the four categoriesdiffered significantly from each other, yieldinga significant main effect, F(3, 277) 35.47,p .01. The relative effectiveness of each ofthese categories against the others was deter-mined by the Bonferroni post hoc comparisonanalysis (see Table 4).

The results shown in Table 4 reveal signifi-cant mean differences between all of the tech-niques except for the ESds of multiple tech-niques and role-playing (Mdiff 0.25, p .70).Role reversal revealed significantly higher ESdsthan both multiple techniques and role-playingand a significantly lower ESd than doubling. Inaddition, doubling differed significantly frommultiple techniques and role-playing. The neg-ative signs for some of the results shown in themean difference column of Table 4 indicate thatthe average ESd for Technique B was higherthan that of Technique A. The meaning of theseoutcomes may suggest that each technique con-tributes to a psychological effect or effects thatare different from those of other techniques,thus possessing a unique impact. The only ex-ception to this conclusion is that the effects ofrole-playing and multiple techniques seemedundifferentiated.

Gender

The next question concerned the possible ef-fects of gender composition of the participantsin the studies (see Table 1). The average effectsize for studies using single-gender groupswas 0.72 (n 69, SD 0.53). This was com-pared with the mean effect size for studies thatused males and females together (M 0.99,n 177, SD 0.75). The result of a t testrevealed a statistically significant difference be-tween studies using only one gender at a timeand those using mixed-gender groups; t(244) 2.71, p .01. Studies reporting effects fromdata on groups of mixed gender tended to reporthigher effect sizes.

Clinical Versus Nonclinical Participants

Of the 25 studies, 9 (36%) involved hospital-ized patients, prisoners, and special populationswith developmental and physical disability (seeTable 1). The rest of the studies (64%) wereconducted with college and university students.

Table 3Descriptive Statistics for Techniques Using AllAverage Reported Effect Sizes

Techniques nd

Effect size

M Mdn SD

Role reversal 100 0.93 0.79 0.74Multiple techniques 46 0.42 0.33 0.30Role-playing 17 0.17 0.09 0.68Doubling 118 1.29 1.36 0.51

Total 281 0.95 0.85 0.69

19PSYCHODRAMA META-ANALYSIS

To determine whether psychodramatic tech-niques were more effective with either of thesetwo groups, we compared the results by meansof a t test. There was no statistically significantdifference, t(279) 1.32, p .18, betweenthe studies using students (n 134, M 0.89,SD 0.77) and those using clinical populations(n 147, M 1.01, SD 0.60).

Number of SessionsTable 1 shows that some of the studies in-

volved one session intervention (54%), andsome ranged between 2 and 40 sessions. Todetermine whether there was a difference be-tween one-session and multiple-session inter-vention, we conducted a t-test analysis. Theresults revealed that there was no statisticallysignificant difference in the effectiveness ofpsychodramatic techniques applied in a singlesession (n 110, M 0.88, SD 0.78)compared with several sessions (n 167,M 1.02, SD 0.61) t(275) 1.58, p .12.

Independence of the PredictorsThe present meta-analysis included three di-

chotomous independent variables: population(students vs. other populations), gender (singlevs. mixed), and number of sessions (one vs.more than one). The results of 2 2 chi-square(two-tailed Pearsons) analyses showed thatthere were no statistically significant associa-tions between population and gender, 2(1, N246) 0.54, p .46, or between gender andnumber of sessions, 2(1, N 242) 2.35, p.13. However, there was a statistically signifi-cant association between population and num-ber of sessions, 2(1, N 277) 174.68, p

.00. Participants who were members of groupsconducted in hospitals, prisons, and other spe-cial populations were more often exposed tomultiple sessions, whereas students, mostly inuniversity-based research, tended to be limitedto a single session.

The results of three 2 4 chi-square analysesconducted to investigate the association be-tween the four treatment techniques (role-play,role reversal, doubling, and multiple tech-niques) and each of the above three predictorsrevealed the following: There were statisticallysignificant relations between the type of tech-niques by (a) population group, 2(3, N 281) 190.51, p .01; (b) number of sessions,2(3, N 277) 226.68, p .01; and (c)gender, 2(3, N 246) 66.84, p .01. Someof the interesting outcomes of these results are,for instance, that studies with role reversal wereconducted significantly more with students thanwith other populations and involved a single-session exposure. The doubling technique wasconducted significantly more with nonstudentpopulations and in multiple sessions. Multipletechniques were used significantly more withsingle- than with mixed-gender groups

Finally, a question might be raised concern-ing the likelihood that our selection of studiesmight have been biased. Using the calculationsproposed by Rosenthal (1995, pp. 189190;1991, p. 104), we found that it would takesome 1,057 statistically insignificant findings tochallenge the present outcomes. Becker (1994)pointed out that a large fail-safe number lendscredence to the finding of significance, and inmany cases this number will be large. Given thepaucity of psychodrama research, we can safelyassert that the studies used in the present meta-analysis do not represent a selection bias.

Table 4Bonferronis Post Hoc Comparison of Adjusted d Means

Technique A Technique BMean difference

(A minus B) SEMultiple techniques Doubling** 0.87 0.10

Role-playing** 0.26 0.16Role reversal* 0.51 0.10

Doubling Role-playing** 1.13 0.15Role reversal* 0.35 0.08

Role-play Role reversal** 0.77 0.15

* p .0001. ** p .00005.

20 KIPPER AND RITCHIE

Discussion

The implications that emerged from thepresent study concern changing the perception ofthe empirical validity of psychodrama and its pri-mary techniques, integrating such techniques in awider group psychotherapy practice, and outlin-ing directions for future psychodrama research.Some of these issues are clearly supported bythe findings; others, though somewhat specula-tive, are nonetheless worthwhile considering.

The first implication addresses the prevailingview toward psychodrama and its techniques.As described earlier, observers of psychodramahave been torn between the favorable clinicalimpression of the method, on the one hand, andthe absence of empirical validation for its mul-tifarious interventions, on the other. The presentmeta-analysis shows that the overall treatmenteffect size found for the 25 studies was 0.95,above the cutoff level of 0.80 that customarilyindicates a large effect (Cohen, 1992). Thiseffect size is slightly higher than the comparableresults commonly reported in the group psycho-therapy literature for the effectiveness of grouptherapy in general (0.500.70; Fuhriman &Burlingame, 1994). Arguably, the elevated ef-fect size might be partly due to the halo effectoften associated with psychodramatic interven-tions. Nonetheless, the overall effect size iscertainly congruent with the tendency reportedfor group psychotherapy in general, a fact thatlends credence to the conclusion that the avail-able research demonstrated the empirical valid-ity of the basic psychodramatic techniques (i.e.,role reversal and doubling).

This conclusion notwithstanding, the paucityof sound research remains a troubling issue.The 25 studies selected for the present meta-analysis spanned over a period of three de-cadesby any measure, a poor showing. True,these studies may not have constituted the entirebody of published empirical psychodrama liter-ature. First, they included only studies pub-lished in English. Second, there were otherexperimental studies published in English notincluded in the meta-analysis because ofincomplete data or weak designs. Our selectionaimed at representing the best studies available.Obviously, the paucity of studies is an unwel-come trend. The second implication of thepresent study is, therefore, that if one reason forpast diminished research activities has been

a view that psychodrama is not amenable totraditional, quantitative research, the results ofthe present meta-analysis ought to dispel thatnotion.

Our focus on single psychodramatic tech-niques has some relevance for clinical practice.We have repeatedly noted that this meta-analy-sis concentrated on psychodramatic techniquesrather than on the psychodramatic process (theentire session) that characterizes the classicalapproach (e.g., Blatner, 1996; Kellerman, 1992;Moreno, 1953, 1964). Addressing the individualpsychodramatic techniques as a way of demon-strating the merit of the entire therapeutic pro-cedure is a novel approach. It means that eachcomponent (technique) of the process plays animportant role by itself. The implication forclinical practice is that separate psychodramatictechniques can be adopted and incorporated intovarious forms of group psychotherapy, eventhose based on theoretical approaches that differfrom that of J. L. Moreno. Role reversal, dou-bling, and role-playing enactment, singularly ortogether, could add to the psychotherapeuticendeavor of many forms of group treatment.

The final implication of the present analysisconcerns a most promising direction for futureresearch. The classical psychodramatic litera-ture (e.g., Blatner, 2000; Moreno, 1964; Starr,1977; Yablonsky, 1976) did not offer a system-atic rationale for predicting a specific, orunique, therapeutic effect of each psychodra-matic technique. In fact, half of the studies inthe present meta-analysis were conducted be-fore the first rationale for a systematic catego-rization of psychodramatic techniques was pro-posed (Kipper, 1986). In the absence of a cleartheoretical model for the therapeutic effect ofthe various psychodramatic techniques, therewas no coherent direction in psychodrama re-search. Kipper (1981, 1986, 2001) proposed anew model based on the proposition that theparticular design of psychodramatic enactments(techniques) represents one of four unique sim-ulation conditions. Each such condition tends toactivate different psychological (psychothera-peutic) processes and therefore produces differ-ent therapeutic outcomes. The finding that eachof the role reversal, doubling, and multiple tech-niques significantly differs from the others isconsistent with the above proposition, whichexpects the three to be distinctly (psychologi-cally) different interventions.

21PSYCHODRAMA META-ANALYSIS

So far, there have been too few studies basedon this new modelthat is, studies designed totest Kippers specific predictions about the dif-ferential psychological processes activated byeach simulation condition. Yet initial outcomestend to support the model. For example, it hasbeen shown (Kipper, 2001) that role reversalactivates disinhibition processes by creatingpsychological distancing. Cognitively, role re-versal was also shown to have facilitated atti-tude change. On the emotional level it generatedwarmth and trust. Doubling, however, wasshown to have produced empathy (Hudgins &Kiesler, 1987). It is hoped that future researchwill result in a good number of studies based onthe differential psychological effect hypothesis,thus enabling us to incorporate the specificpsychotherapeutic outcome as a predictor insubsequent meta-analyses.

It is not clear, though, why the present resultsfor the technique of role-playing showed a neg-ligible improvement effect size. It is importantto note that in the psychodramatic context, theterm role-playing does not connote enactmentin general. Rather, it refers to a specific form ofenactment, a portrayal of oneself in ones ownidentity. Given the widespread use of role-play-ing enactments in research, therapy, and train-ing, better outcomes for this particular tech-nique were expected. This might have been dueto the fact that it had fewer studies in our listthan the other three groups. Moreover, the role-playing studies generated a miniscule numberof ESs, at most one fifth of the number gener-ated by the studies on role reversal and dou-bling. One can only speculate that with morestudies on the clinical effectiveness of role-playing technique and subsequently a numberof effect sizes similar to those obtained for rolereversal and doubling, the results for this tech-nique would be much better.

The group we called multiple techniques rep-resented the traditional approach to the study ofthe effectiveness of psychodramathat is, stud-ies that considered the entire psychodramaticprocedure as the basic unit of the investigation.The results revealed a small to moderate im-provement effect size. Logically, one wouldassume that if each of the role-reversal anddoubling techniques separately generates alarge effect size, they would produce a bettersynergetic effect when combined with othertechniques. The combined format would boost

the improvement effect size of the multipletechniques package. This did not happen. Thereason for this may lie in the data. A glance atTable 2 reveals that the mean effect sizes fortwo of the studies in the multiple techniquesgroup were close to zero. One study, in fact theonly study in the entire list of 25, actuallyreported no significant differences between thegroup treated with psychodrama and a controlgroup (Slawson, 1965). Given the weakness ofthe available data, one must hold any conclu-sion on the effectiveness of multiple techniquesin abeyance pending further investigations. Thecloseness of the presently found effect size tothe moderate cutoff point would encourage sucha research undertaking.

Other specific outcomes of the present studydeserve comment. One would be hard pressedto explain the lack of a difference in the effec-tiveness between the same therapy rendered inone session and in multiple sessions. With ourfocus on psychodramatic techniques, these re-sults are actually considered positive. It mightbe recalled that the rationale underlying thetherapeutic effect of psychodramatic techniquemakes no differentiation concerning who theactor is. Rather, it hypothesized that beingplaced under a particular simulation condition(a technique) ipso facto activates the specificpsychological process. All that it requires is thatthe actors be able to understand and follow theinstructions on how to role-reverse or become adouble and be able to become involved in theenactment. The psychological effect(s) of agiven technique would be the same regardlessof whether the actor is a patient or a student.The same underlying rationale also expects thathaving identified the specific therapeutic effectof a given technique, a person will experience itevery time the technique is used, whether onceduring a single session or 20 times during 20sessions. The results, therefore, are interpretedas an encouragement for the continued researchinto the therapeutic effectiveness of single psy-chodramatic techniques.

It appears that studies using both male andfemale participants showed greater improve-ment effect sizes than those using only males orfemales. The meaning of this finding is puzzlingand certainly poses a challenge for futureresearch.

Finally, the strength of the findings dependson the number and the scientific rigor of the

22 KIPPER AND RITCHIE

studies that constituted the basis for the meta-analysis. We are aware that the present selectionis restricted. It did not include studies publishedin foreign languages, although to the best of ourknowledge, most of the well-designed quantita-tive studies in the field have been published inEnglish-language journals. Also, we did nothave access to unpublished studies, if suchmanuscripts in drawers exist. Because wewere conducting the first meta-analytic study ofpsychodrama research, we decided to adopt aninclusion criterion that considered only experi-mentally designed studies. Although this deci-sion may have lent more credence to the find-ings, at the same time it has curtailed the scopeof the data. One must bear in mind this limita-tion in assessing the meaning of the presentoutcomes. All in all, the findings appear to sheda positive light on the issue of the validity ofpsychodramatic interventions and to encourageresearch regarding the specific psychotherapeu-tic effects of its basic techniques.

ReferencesReferences marked with an asterisk indicate stud-

ies included in the meta-analysis.*Barrett, M. E. (1986). Self-image and social adjust-

ment change in deaf adolescents participating in asocial living class. Journal of Group Psychother-apy, Psychodrama and Sociometry, 39, 311.

Blatner, A. (1996). Acting-in (3rd ed.). New York:Springer.

Blatner, A. (2000). Foundation of psychodrama (4thed.). New York: Springer.

Becker, B. J. (1994). Combining significance levels.In H. Cooper & L. V. Hedges (Eds.), The hand-book of research synthesis (pp. 215230). NewYork: Russell Sage Foundation.

*Carbonell, D. M., & Parteleno-Barehmi, C. (1999).Psychodrama for girls coping with trauma. Inter-national Journal of Group Psychotherapy, 49,285306.

Clayton, M. (1993). Living pictures of the self. Caul-field, Victoria, Australia: ICR Press.

Cohen, J. (1992) A power primer. PsychologicalBulletin, 112, 155159.

DAmato, R. C., & Dean, R. S. (1988). Psychodramaresearch: Therapy and theory: A critical analysis ofan arrested modality. Psychology in the Schools, 25,305314.

Farmer, C. (1995). Psychodrama and systemic ther-apy. London: Karnac Books.

Fox, J. (Ed.). (1987). The essential Moreno. NewYork: Springer.

Fuhriman, A., & Burlingame, G. M. (Eds.). (1994).Group psychotherapy: An empirical and clinicalsynthesis. New York: Wiley.

Glass, G. V. (1976). Primary, secondary and meta-analysis of research. Educational Research, 5,38.

Goldman, E. E., & Morrison, D. S. (1984). Psycho-drama: Experience and process. Debuque, IA:Kendal/Hunt.

*Goldstein, J. A. (1971). Investigation of doubling asa technique for involving severely withdrawn pa-tients in group psychotherapy. Journal of Consult-ing and Clinical Psychology, 37, 155162.

*Goldstein, S. G. (1967). The effect of doubling oninvolvement in group psychotherapy as measuredby number and duration of patient utterances. Psy-chotherapy: Theory, Research, and Practice, 4,5760.

*Granvold, D. K., & Ollerenshaw, S. J. (1977). In-terpersonal skill training through a data feedbackgroup. Journal of Group Psychotherapy, Psycho-drama and Sociometry, 30, 4959.

Holmes, P. (1992). The inner world outside: Objectrelations theory and psychodrama. London: Tavi-stock/Routledge.

Holmes, P., & Karp, M. (Eds.). (1991). Psychodramainspiration technique. London: Tavistock/Routledge.

Holmes, P., Karp, M., & Watson, M. (Eds.). (1994).Psychodrama since Moreno. London: Tavistock/Routledge.

*Hudgins, M. K., & Kiesler, D. J. (1987). Individualexperiential psychotherapy: An analogue valida-tion of the intervention model of psychodramaticdoubling. Psychotherapy, 24, 245255.

Hunter, J. E., Schmidt, F. L., & Jackson, G. G.(1982). Meta-analysis: Cumulating research find-ings across studies. Beverly Hills, CA: Sage.

Johnson, B. T. (1995). Software for the meta-ana-lytic review of research literatures. Hillsdale, NJ:Erlbaum.

*Johnson, D. W. (1967). The use of role reversal inintergroup competition. Journal of Personality andSocial Psychology, 7, 135141.

*Johnson, D. W. (1970). Efficacy of role reversal:Warmth of interaction, accuracy of understanding,and the proposal of compromises. Proceedings ofthe 78th Annual Convention of the American Psy-chological Association (pp. 385386). Washing-ton, DC: American Psychological Association.

*Johnson, D. W. (1971). Effectiveness of role rever-sal: Actor or listener. Psychological Reports, 28,275282.

*Johnson, D. W., & Dustin, R. (1970). The initiationof cooperation through role reversal. Journal ofSocial Psychology, 82, 193203.

23PSYCHODRAMA META-ANALYSIS

Kellermann, P. F. (1982). Outcome research in clas-sical psychodrama. Small Group Behavior, 18,459469.

Kellermann, P. F. (1987). A proposed definition ofpsychodrama. Journal of Group Psychotherapy,Psychodrama and Sociometry, 40, 7680.

Kellermann, P. F. (1992). Focus on psychodrama.London: Jessica Kingsley.

Kellermann, P. F., & Hudgins, M. K. (Eds.). (2000).Psychodrama with trauma survivors. London: Jes-sica Kingsley.

Kipper, D. A. (1978). Trends in the research on theeffectiveness of psychodrama: Retrospect andprospect. Journal of Group Psychotherapy, Psy-chodrama and Sociometry, 31, 518.

Kipper, D. A. (1981). Behavior simulation interven-tions in group psychotherapy. In G. M. Gazda(Ed.), Innovations to group psychotherapy (2nded., pp. 242278). Springfield, IL: Charles CThomas.

Kipper, D. A. (1982). Behavior simulation: A modelfor the study of the simulation aspect of psycho-drama. Journal of Group Psychotherapy, Psycho-drama and Sociometry, 35, 117.

Kipper, D. A. (1986). Psychotherapy through clinicalrole playing. New York: Brunner/Mazel.

*Kipper, D. A. (1988a). The differential effect ofrole-playing conditions on the accuracy of selfevaluation. Journal of Group Psychotherapy, Psy-chodrama and Sociometry, 41, 3035.

Kipper, D. A. (1988b). On the definition of psycho-drama: Another view. Journal of Group Psycho-therapy, Psychodrama and Sociometry, 40, 164168.

Kipper, D. A. (2001). The theory and practice ofpsychodrama. Sofia, Bulgaria: Damian YakovPublishers.

*Kipper, D. A., & Ben-Ely, Z. (1979). The effective-ness of the psychodramatic double method, thereflection method, and lecturing in the training ofempathy. Journal of Clinical Psychology, 35,370375.

*Kipper, D. A., & Giladi, D. (1978). Effectiveness ofstructured psychodrama and systematic desensiti-zation in reducing test anxiety. Journal of Coun-seling Psychology, 25, 499505.

*Kipper, D. A., & Har-Even, D. (1984). Role-playingtechniques: The differential effect of behavior sim-ulation interventions on readiness to inflict pain.Journal of Clinical Psychology, 40, 936941.

*Kipper, D. A., Har-Even, D., Rotenberg, M., &Dagan, M. (1982). Involvement in role playing asa function of the simulation procedure and level ofimagination. Journal of Group Psychotherapy,Psychodrama and Sociometry, 35, 99110.

*Kipper, D. A., & Ushpiz, V. (1987). Emotional andcognitive responses in role playing. Journal of

Group Psychotherapy, Psychodrama and Sociom-etry, 39, 131142.

Leviton, E. (1977). Psychodrama for the timid clini-cian. New York: Springer.

*Martin, R. B., & Stepath, S. A. (1993). Psycho-drama and reminiscence for the geriatric psychiat-ric population. Journal of Group Psychotherapy,Psychodrama, and Sociometry, 45, 139148.

Moreno, J. L. (1953). Who shall survive? Beacon,NY: Beacon House.

Moreno, J. L. (1964). Psychodrama: Vol. I. Beacon,NY: Beacon House

Mullen, B., Driskell, J. E., & Salas, E. (1998). Meta-analysis and the study of group dynamics. GroupDynamics: Theory, Research, and Practice, 2, 1329.

Mullen, J. L., & Rosenthal, R. (1985). Basic meta-analysis: Procedures and programs. Hillsdale, NJ:Erlbaum.

*Muney, B. F., & Deutsch, M. (1968). The effect ofrole-reversal during the discussion of opposingview points. Journal of Conflict Resolution, 12,345356.

Rawlinson, J. W. (2000). Does psychodrama work?A review of the literature. British Journal of Psy-chodrama and Sociometry, 15, 67101.

*Remer, R., & Betts, G. R. (1998). The differencebetween strict analogue and interpersonal psycho-dramatic simulation (IPS) methods in research onhuman dynamical systems. International Journalof Action Methods: Psychodrama, Skill Training,and Role Playing, 51, 322.

Roine, E. (1997). Psychodrama: Group psychother-apy as experimental theater. London: JessicaKingsley.

Rosenthal, R. (1991). Meta-analytic procedures forsocial research (Rev. ed.). Newbury Park, CA:Sage.

Rosenthal, R. (1995). Writing meta-analytic reviews.Psychological Bulletin, 118, 183192.

Sacks, J. M., Bilaniuk, M. T., & Gendron, J. M.(Eds.). (1995). Bibliography of psychodrama: In-ception to date. New York: Psychodrama Center ofNew York.

*Schramski, T. G., Feldman, C. A., Harvey, D. R., &Holiman, M. (1984). A comparative evaluation ofgroup treatment in an adult correctional facility.Journal of Group Psychotherapy, Psychodramaand Sociometry, 36, 133147.

*Schutte, N., Malouf, J., & ODare, J. (1990). Pre-vention of cigarette smoking: Effect of informationabout the negative social effects of smoking. Jour-nal of Group Psychotherapy, Psychodrama andSociometry, 43, 2734.

*Slawson, P. E. (1965). Psychodrama as a treatmentfor hospitalized patients: A controlled study.American Journal of Psychiatry, 122, 530533.

24 KIPPER AND RITCHIE

*Stallone, T. M. (1993). The effects of psycho-drama on inmates within a structured residentialbehavior modification program. Journal ofGroup Psychotherapy, Psychodrama and So-ciometry, 46, 24 30.

Starr, A. (1977). Rehearsal for living: Psychodrama.Chicago: Nelson Hall.

*Staven, K. E. (1985). Effects of interview and rolereversal on psychiatric patients perception ofbody image. Journal of Group Psychotherapy,Psychodrama and Sociometry, 38, 7583.

Sternberg, P., & Garcia, A. (1989). Sociodrama:Whos in your shoes? New York: Prager.

*Taylor, G. S. (1983). The effect of nonverbal dou-bling on the emotional response of the double.Journal of Group Psychotherapy, Psychodramaand Sociometry, 36, 6168.

Williams, A. (1989). The passionate technique. Lon-don: Routledge.

Yablonsky, L. (1976). Psychodrama: Resolving emo-tional problems through role-playing. New York:Basic Books.

Received March 5, 2002Revision received July 17, 2002

Accepted September 24, 2002

25PSYCHODRAMA META-ANALYSIS