TOWARDS A TAXONOMY OF EFFECTIVE GROUP LEADER BEHAV-iOR Fhesés for the Degree of M. A, MECHIGAR SEAT E UNWERSW ELLYN S. BAUER 1 ‘97 3 Michigan §t3w Universlty amomc av HUM; & 8096' 800K BINDERY INC. LIBRARY among (Ill-n... Illa... . ABSTRACT TOWARDS A TAXONOMY OF EFFECTIVE GROUP LEADER BEHAVIOR By Ellyn S. Bader The purpose of this study was to determine effective behaviors of group therapists. Effectiveness was measured by two methods. First, patients were asked to rate their group leaders on a five point scale and second, the observers rated the therapists after they had observed all of them. Ten therapists participated in the study. Each of them conducted three groups. Three observers were trained to observe the therapists and code their behaviors on a twelve category scale. After each group, the patients in the group rated the therapist on a five dimension scale. This was then called the patients' rating of therapist effectiveness. At the conclusion of all their observations, the observers ranked the therapists in order of whom they felt to be most and least effective in conducting group therapy. RESULTS Although different patients rated the therapists each time, they agreed in whom they selected as effective Ellyn S. Bader therapists. Thus, if a therapist obtained a high score for his first group, he tended to receive a high score from the patients in his second and third groups. Also, the observers agreed among themselves in whom they selected as effective therapists. Furthermore, the patients and observers agreed with each other on whom they selected as the most and least effective therapists. Three categories of behavior were found that seemed to differentiate the most effective therapists from the others. These were praises, touches and special techniques. These results coincide with some earlier findings and also provide new information towards developing a taxonomy of effective group leader behavior. TOWARDS A TAXONOMY OF EFFECTIVE GROUP LEADER BEHAVIOR By Ellyn S. Bader A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1973 TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . . . . . . . . iii LIST OF APPENDICES . . . . . . . . . . . . . . . . . . iv CHAPTER I. INTRODUCTION . . . . . . . . . . . . . . . . . 1 II. HISTORY . . . . . . . . . . . . . . . . . . . . 3 Major Problems . . . . . . . . . . . . . . . 9 Hypotheses . . . . . . . . . . . . . . . . . 9 III. METHOD ‘. . . . . . . . . . . . . . . . . . . . 10 Brief Summary . . . . . . . . . . . . . . 10 Selection of Subjects . . . . . . . . . . . . . 10 Selection of Observers . . . . . . . . . . . 11 Therapist Rating Scale . . . . . . . . . . . 11 Patient Rating Scale . . . . . . . . . . . . 12 The Training Procedure . . . . . . . . . . . 13 IV. RESULTS . . . . . . . . . . . . . . . . . . . . 15 Reliability among Patients' Ratings . . . . . lS Reliability among Observers' Rankings . . . . 17 Agreement between Observer Rankings and Patient Ratings . . . . . . . l7 Behavioral Differences in Therapists . . . . l8 Regression Analysis . . . . . . . 21 Correlation of Significant Categories . . . . 22 V. DISCUSSION . . . . . . . . . . . . . . . . . . 25 Implications of the Results: New Directions. 27 VI. SUMMARY . . . . . . . . . . . . . . . . . . . . 31 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . 33 APPENDICES . . . . . . . . . . . . . . . . . . . . . . 35 ii 10. 11. 12. LIST OF TABLES Pre- and post-observation measures of reliability . . . . . . . . . . . . Scores for each therapist by patient rating and observer rank . . . . . . . . . Reliability of patient ratings nested occasions . . . . . . . . . . . . . Reliability of Observers' rankings of therapists . . . . . . . . . . . . Kendall and Spearman correlation coefficients in for patient rating with observer rank . . Spearman correlation coefficients for praises . . . . . . . . . . . . . . Spearman correlation coefficients for tOUCheS .0 O O I O I O I O O O I O Spearman correlation coefficients for Special techniques . . . . . . . . Comparison of top four therapists vs. four therapists on category praises Summary table of step-wise multiple regression . . . . . . . . . . Pearson correlation coefficients for important variables . Spearman correlation coefficients for important variables . . . . . . . iii category category category bottom Page 14 16 16 17 18' 20 20 20 21 22 23 24' LIST OF APPENDICES APPENDIX Page A. THE THERAPIST RATING SCALE AND CATEGORY DEFINITIONS . . . . . . . . . . . . . . 35 B. THE PATIENT RATING SCALE . . . . . . . . . . . 38 iv CHAPTER I INTRODUCTION Over the past twenty years, group therapy has become an increasingly popular form of mental health treatment. Many inpatient psychiatric units across the country use group psychotherapy as a regular adjunct to their therapy programs. Group therapy has also become more widely used as a means of outpatient care. And, with the spread of the encounter movement, group therapy has become more widely known to the public. As group therapy has become a more popular form of treatment, considerable research has been accumulated which supports the conclusion that group psychotherapy can facil- itate constructive personality or behavior change in persons described as mentally ill or emotionally disturbed. Peyman (1956) found group therapy to be very effective with a group of chronic schizophrenic patients and Tucker (1956) used group therapy very successfully with chronic psychotic soiling patients. Baehr (1954) and Cadman (1954) also asseSsed group therapy to be a very effective means of facilitating constructive personality change. Very little research, however, has been conducted to determine whether successful outcome in group therapy is related to orientation of the therapist or to specific therapist characteristics. Today, orientation of group therapists may be very diverse. While psychoanalytic therapists focus generally on psychodynamics and past history, gestalt therapists focus predominantly on the here and now. Recently, Yalom and Liberman (1972) have researched the area of therapist characteristics and have uncovered some results which suggest that certain behaviors of group therapists are more important for positive outcome than is the orientation of the therapist. Their work provided the impetus for this study of characteristics of group therapists in three short-term inpatient psychiatric units. Ten thera- pists were observed by trained observers and rated by the patients in their groups. The problem was then to determine whether therapists who were rated most highly by the patients and the observers used any behaviors more or less frequently than did the lower rated therapists. CHAPTER II HISTORY Hobbs and Pascal (1946) were the first to investi- gate therapist characteristics. They made use of tapes of group sessions and classified therapist responses as either client-centered, eclectic or didactic-authoritarian. They classified patient responses as either therapeutically posi- tive or negative with insight or abreaction being scored positively and passive-dependent or hostile-agressive responses being scored negatively. Their results indicate that client-centered and eclectic responses by the thera- pist are more highly associated with positive therapeutic statements by the patients than are didactic-authoritarian responses by the therapist. Ends and Page (1957) explored variation in outcome in group therapy with alcoholics by focusing on differences related to the therapeutic orientation of the therapist. Their results strongly suggest that activities of the thera- pist are significantly related to variations in outcome rather than being related to the particular orientation of the therapist. As a result of the Ends and Page (1957) and Hobbs and Pascal (1946) studies, other researchers decided to explore further the area of therapist characteristics. Rogers (1957) suggested that therapeutic personality change occurs in proportion to the amount that the patient experi- ences certain qualities in his therapist. He called unconditional positive regard, empathic understanding, genuineness, and willingness to be known by the client the four necessary conditions for constructive personality change. Barrett-Lennard (1962) researched these further and obtained results for all these characteristics except willingness to be known by the client. Truax (1961) designed some research to evaluate statistically the relationships between intrapersonal explor- ation by patients in group therapy and the specific char- acteristics of their therapists. He used Open-ended groups of hospital inpatients with three different types of leaders and tested seven hypothesized therapist conditions, includ- ing three of those suggested by Rogers (1957). He found the following therapist conditions to be significantly related to intrapersonal exploration: accurate empathy, uncondi- tional positive regard, self-congruence/genuineness, assumed similarity of self and patients, responsivity, leadership. Truax and Carkhuff (1967) showed that helpful or harmful aspects of therapeutic encounters depend primarily on the therapist and his characteristics. They described accurate empathy, nonpossessive warmth and the therapist's self- congruence to be the major ingredients of helpful therapy. In this research, these therapist characteristics were defined more behaviorally than in the past. Accurate empathy was defined to be reflection of the patient's feelings, particularly when the therapist's response expanded the client's awareness by going beyond the patient's current verbalization. Self-congruence was defined by a lack of phoniness and defensiveness and an openness to experience. Nonpossessive warmth was defined by the thera- pist showing love and positive affect in conjunction with being nonjudgmental. This was not based on unwarranted optimism or mere reassurance. As a result of this research Carkhuff and Truax suggested that future research should be conducted on therapist characteristics and that current teaching should emphasize antecedents to effective thera- peutic outcome rather than emphasizing personality dynamics and psychopathology. Recently, Yalom and Lieberman (1971, 1972) conducted a very comprehensive study of group therapy. They made use of eighteen leaders who conducted groups for a total of thirty hours. They found the most competent leaders they could to represent each therapeutic orientation. Two hundred and nine undergraduate students participated in the study. The leaders were of the following orientations: Rogerian (2) Gestalt (2) Psychodrama (2) Sensitivity Training (2) Transactional Analysis (2) Leaderless Tape (2) Synanon (2) Marathon Groups (2) Psychoanalytic (1) Sensory Awareness (1) Trained observers attended each group and rated how fre- quently the leaders displayed each of twenty-eight behaviors. These twenty-eight variables describing leader behavior were intercorrelated and then factor analyzed. Four clusters emerged (emotional stimulation, caring, meaning attribution, executive behavior) which accounted for 75 per cent of the variance. Two clustering methods were then used to identify six types of leaders. This method of dividing leaders pro~ duced discriminations among leaders of highly varied orienta- tion. In another part of the same study, Yalom and Lieberman used many methods to determine the high learners and the casualties from the 209 group participants. A casualty was defined as enduring, significant, negative out- come, which was caused by the subject's participation in the group and which persisted even eight months after the group ended. Casualties were identified by the following means: 1. Request for emergency psychiatric aid 2. Dr0pouts from groups 3. Peer evaluation 4. Self-esteem drop on Rosenberg Self-esteem measure 5. Subject's own testimony 6. Entrance into psychotherapy 7. Leaders' ratings All of the names gotten in this way were studied further and sixteen casualties were identified. All of these students had undergone psychological decompensation that had persisted and there was evidence that the group was the responsible agent. The high learners were identified in many ways among which were a large battery of self-report questionnaires which each student filled out before beginning the group, after each meeting, at the end of the group experience and again, for a final follow-up six months later. Of the six leader types earlier identified, the second type was by far the most effective in producing posi- tive changes, while minimizing the number of participants who were negative changers, dropouts or casualties. These leaders were high on caring and meaning-attribution with only moderate use of emotional stimulation and executive function. These leaders were individually focused leaders who gave love as well as information about how to change. They all subscribed to some systematic theory about how individuals learn and change and they all shared this with their groups. It was also discovered that overall success for the participants was related linearly with the amount of caring displayed by the leader. The second type of leaders, those identified to be the most successful, subscribed to three different orienta- tions. One was an eclectic marathon leader, one was a sensitivity group leader and the third was a leader in trans- actional analysis. Further analysis was done to determine whether leaders of the same orientation were similar in their behaviors. Except for the two gestalt leaders, similarities based upon orientation were weak or non-existent. The find- ings were indisputable that conventional categories of leader orientation were poor predictors of leader behavior. These findings reinforce earlier findings and suggest that it is fruitful to examine behaviors of group therapists in hopes of some day developing a clear taxonomy of effective group leader behavior. Major Problems In this study, differences in therapist character- istics were to be identified. The major difference between this study and that of Yalom and Lieberman was that the groups were not ongoing and that the patient pOpulations changed rapidly. Most of the patients remained hospitalized for a short time, usually on an average of seven days. Then, most were discharged to other means of treatment. Therefore, measures of outcome success and failure were difficult to determine. It was decided to use ratings by the patients and observers to determine behavioral differences in therapists. The patients rated the therapists on a five point scale after each group. The observers rated the therapists after they had observed all ten of them. These rating measures will be described further in Chapter III. Hypotheses l. Therapists who will be rated high on the Therapist Rating Scale by one group of patients will be rated high by a second and third group of patients. 2. A high amount of agreement will exist between the three Observers' ratings of the ten therapists. 3. A high amount of agreement will exist between the Observers' ratings of the therapists and the patients' ratings of the therapists. 4. Therapists rated high by the patients and observers will differ in some behavioral characteristics from the other therapists. CHAPTER III METHOD Brief Summary In this study, ten therapists were observed conducting groups. Each therapist was observed by three different observers for an hour each time. The therapists were observed on an average of once every two weeks. The observers used the Therapist Rating Scale to observe the therapists during the group and immediately after the group gave the patients in the group a rating scale to fill out. In this way the patients could express their feelings about the therapist immediately afterthe group. After all of the observations were Completed, each observer was asked to rank the ten therapists in order of whom he felt to be most to least effective in conducting group therapy. Selection of Subjects Ten different group therapists were selected from three inpatient psychiatric units. All of the hospital units were short-term care facilities. Therapists were selected on the basis of their willingness to participate in the study, as this entailed having a non-participating 10 11 observer visit their groups. Furthermore, wherever possible, therapists were selected who had little patient contact out- side of the group setting. The final group of ten therapists consisted of two psychiatrists, two psychologists, one social worker, two psychiatric nurses and three psychiatric techni- cians. None of them subscribed to any rigid theory of conducting groups. Most had had some psychoanalytic training, one had been trained in reality therapy and one had a very strong Rogerian background. However, all of them described themselves as eclectic therapists. Selection of Observers Several people volunteered to participate in the study to enhance their own knowledge of psychiatry and group therapy. Three observers were selected primarily on the basis of their previous work experience. Two men and one woman were selected. The men were employed by a hospital; one as a psychology research assistant and the other as a psychiatric technician. The woman was employed at a nearby community crisis center. It was thought by the researcher that these three peOple could be trained to a high degree of reliability the most rapidly. Therapist Rating Scale The Therapist Rating Scale consists of twelve behavioral categories that may be performed by a therapist .u n\ u an» I m' n‘; .q. 5.. a: C». n\~ L» Y‘ 51.. 12 in the process of a group. These include such behaviors as asking a direct question or giving praise to a patient. Each category may be clearly defined and accurately observed by a trained observer. The scale was developed by David Freeman, PhD, Larry Harper and Ellyn Bader after observing many groups at El Camino Hospital. Some of the coding categories were also derived from those developed by Moustakas and Schalock (1955). For the observations of the categories on the Therapist Rating Scale, a thirty second interval time schedule was used. Therefore, no category could be scored more than once within each thirty second interval. However, if more than one therapist behavior occurred within a thirty second interval, this could be recorded. An auto- matic timing mechanism emitted an audible beep every thirty seconds through ear phones worn by the observers. Since observations lasted for one hour, the maximum number of responses on any one category was 120 and the minimum was 0. A copy of the Therapist Rating Scale with a description of the twelve categories is found in Appendix A. Patient Rating Scale At the completion of each group, all patients in the group were asked by the observers to fill out the Patient Rating Scale. This scale consists of five questions designed to tap the patients' feelings about each therapist. 13 Since all of the observations of each therapist were done approximately two weeks apart and since the observations were done in short-term care facilities, the patients rating the therapists were usually a totally new group each time. A c0py of the Patient Rating Scale is found in Appendix B. The Training Procedure The three observers were trained in two weeks with six three hour training sessions. Prior to the training, the researcher made three video tapes of group therapy sessions in which the video was always focused on the therapist. The first training session involved a general discussion about the research project including a clear explanation about the definition of each category on the Therapist Rating Scale. The observers were also shown the Patient Rating Scale and taught how to administer it. For the next three training sessions the observers worked intensively with the video tapes. First they worked without the timer and later with it. For the last hour of the third video session, a strict observation with no col- laboration was done. A reliability check was done on this data. Next, in pairs of two, the observers practiced by doing observations of real groups at El Camino Hospital. Scott's Pi (Holsti, 1969) was used to compute the overall 14 reliability between pairs of observers. When the reliability figures were above .80, it was decided that the observers could begin the actual observations. After each rater had observed his ten groups for the project, overall reliability was again computed between pairs of observers. The pre- and post-observation measures of reliability are shown in Table 1. Table 1. Pre- and posteobservation measures of reliability. Pre-observation Post-observation Observers reliability reliability Observer I and II .865 .892 Observer II and III .816 .826 Observer I and III .823 .925 Before the actual observations began, the researcher acting as 'expert' coded one group with each of the three observers. Reliability of each category was then computed as a correlation coefficient between the Observers' ratings and the author's. These ranged from .72 to .99 with the. mean being .83. Reliabilities for each category are found in Appendix A. CHAPTER IV RESULTS At the completion of the observations, each observer was asked to rank the therapists from l-lO in order of whom he felt to be the most to least effective in conducting group therapy. The Patient Rating Scale scores for each group were then computed. The scale was scored on a 5-4-3-2-1 basis. As the number of patients in each group varied, an average score was computed for each group. Then, the three average scores for each therapist were added together to obtain a total score and ranking for each therapist. These totals varied from 71.1 to 45.5 with a mean of 59.9. Table 2 contains the score for each therapist, as well as the rank given to that therapist by each observer. It is interesting to note that the three therapists rated most highly by the patients were also selected in that order by the observers. Reliability amonggPatients' Ratings With the scores obtained from the patients' ratings for each of the three groups for each therapist, an analysis of variance was conducted. This was used to determine the 15 16 Table 2. Scores for each therapist by patient rating and observer rank. Therapist Patient Rating Observers' Ranks Therapist 1 71.1 1 l l Therapist 2 67.2 2 2 2 Therapist 3 64.5 3 3 3 Therapist 4 63.6 7 5 9 Therapist 5 62.5 8 9 4 Therapist 6 61.6 4 8 5 Therapist 7 59.4 6 4 6 Therapist 8 52.4 9 10 10 TherapiSt 9 51.3 5 7 7 Therapist 10 45.5 10 6 8 consistency of scores by patients over occasions. A relia- bility coefficient of .78 was found from Guilford's (1956, p. 300) formula for intraclass correlation. Table 3 shows these computations. These results suggest that even though therapists were rated by a different group of patients each time, there still existed agreement among patients in whom they rated high and in whom they rated low. Table 3. Reliability of patient ratings nested in occasions. Sums of Mean r Variation df Squares Squares kk Therapist 9 186.0585 30.6732 Occasion 2 11.8047 5.9023 Residual 18 81.4537 4.5252 Total 29 279.3169 .78 l7 Reliability among Observers' Rankings Next, an analysis of variance was conducted to determine the amount of agreement between the observers' rankings of the ten therapists. Again Guilford's (1956) formula for intraclass correlation was used to determine the reliability coefficient. Here a reliability coeffi- cient of .89 was found. Table 4 reports these results. This suggests a high amount of agreement among observers in whom they selected as the most and least effective therapists. Table 4. Reliability of Observers' rankings of therapists. Sums of Mean Variation df Squares Squares rkk Therapists 9 202.8333 22.5370 Observers 2 0.0 0.0 Residual 18 44.6667 2.4815 Total 29 247.5000 .89 Agreement between Observer Rankings and Patient Ratings Kendall and Spearman correlation coefficients were computed to determine the amount of agreement between the Observers' rankings of the therapists and the patients' total ratings. The results appear in Table 5. These results show a high amount of agreement between the way 18 patients and observers rated the therapists. This suggests that even though the method for rating was different, the results were nearly identical. Table 5. Kendall and Spearman correlation coefficients for patient rating with observer rank. Kendall correlation Spearman correlation coefficient coefficient Patient rating -0.5843 Patient rating -0.7416 with N (10) with N (10) Observer rank Sig. .009 Observer rank Sig. .007 Behavioral Differences in Therapists After all the observations were completed, a total number of times each category behavior occurred was computed for each therapist. Every therapist used asking a direct question more than any other behavior. In looking at the data visually, three categories looked like they might dif- ferentiate the higher rated therapists from the lower rated ones. These were praises, special techniques and touches. The occurrence of touches and special techniques was somewhat unusual. Special techniques were used by only two therapists. These were the therapists rated first and second by both the patients and the observers. Therapist 1 made use of special techniques ten times in his three groups. He conducted several role playings within the group, had one group take 19 a blind walk and did some gestalt work with a pillow. The second therapist used special techniques twice. Both times it involved some gestalt work with two chairs in which the patient acted out both sides of a conflict. None of the other therapists used special techniques in any of their groups. The occurrence of touches was also unusual. This behavior was used only by therapists rated one, three and seven by the patients and those rated one, three and four by the observers. Therapist 1 used it twenty times, thera- pist 3 used it ten times and the third therapist used it twice. No touching occurred in any of the other groups. By looking at the data, it seemed as if the category praises might differentiate therapists. Therapist 1 used praising behavior a total of twenty times in his three groups, while therapist 10 used this behavior only twice. Except for one therapist, it seemed as if a near linear relationship might exist between the amount of praising behavior shown by a therapist and the way he was rated by the patients. At this point, Spearman's rank correlation coeffi- cients were computed between each of the twelve categories and the patient ratings and between the twelve categories and the observer rankings. Significance levels of .05 or .01 were only found for three categories: special tech- niques, touches and praises. These results appear in Tables 6, 7 and 8. 20 Table 6. Spearman correlation coefficients for category praises. Patient Rating 0.3988 Observer Rank -0.7570 with N (10) with N (10) Praises Sig. .127 Praises Sig. .006* Table 7. Spearman correlation coefficients for category touches. Patient Rating 0.576 Observer Rank -0.724 with N (10) with N (10) Touches Sig. .05 Touches Sig. .05 Table 8. Spearman correlation coefficients for category special techniques. Patient Rating 0.746 Observer Rank -0.748 with N (10) with N (10) Special Tech. Sig. .01 Special Tech. Sig. .01 Next, an analysis of variance was run in which the top three therapists were compared with the bottom three based on the patients' ratings. This was done on all of the categories except Special techniques and touches. It was hoped that other significant variables might be identified in this way. None of the categories reached a level of significance at the .05 or .01 level. However, praises 21 came the closest to being significant. Then, an analysis of variance on praises was run in which the top four therapists were compared with the bottom four. The results of this appear in Table 9. Table 9. Comparison of top four therapists vs. bottom four therapists on category, praises. Sums of Mean Variation df Squares Squares F ratio Between Groups 1 217.25 217.25 6.65* Within Groups 6 195.96 32.66 *Significant at the .05 level. Regression Analysis Next, a step-wise multiple regression was run in which nine category variables were entered freely. This was done to account for the variation in the total patient rating by the category scores. Touches entered first, as it was the single most important category to account for the varia- tion. It alone accounted for 32 per cent of the variance. Next, direct question and special techniques entered and then touches was removed. This occurred because direct question and special techniques seemed to incorporate touches and account for about the same amount of variance. Next, praises entered and at this point 67 per cent of the 22 variance had been accounted for with a standard error of estimation of 5.54. None of the other category variables after this added enough to the variance to be worth reporting. The final functional equation was: Patient Rating = 36.68 + .l7(DQ Score) % .65(Pr Score) + 1.85(ST Score) + e A summary of the regression analysis appears in Table 10. Table 10. Summary table of step-wise multiple regression. Variable Multiple Increase Beta Partial Entered Removed R RSQ in RSQ Coeff. Correl. 10 .5668 .3213 .3213 .668 .567* .6742 .4546 .1333 .166 .500 9 .7296 .5323 .0777 1.848 .565* 10 .7295 .5321 .0002 6 .8186 .6701 .1380 .653 .535* *Significant at the .05 level. Correlations of Significant Categories With the data obtained from the regression analysis, Spearman and Pearson correlation tables were done on the significant variables. The category interprets was included as it was the next after praises to be brought into the regression analysis. Even though it accounted for an increase of only .022 in the variance, it seemed important 23 to look at its correlation with the other variables. The results of the correlations appear in Tables 11 and 12. From these tables, it is clear that the categories praises, special techniques and touches all seem to cor- relate with each other consistently. They also correlate with the patient ratings and the observer rankings. In the Spearman table, the category direct question correlates with touches. This might help to explain how direct ques- tion and special techniques replaced touches in the regression. From the data, it appears that group therapists who use a lot of verbal praising behavior, also use touching and special techniques. It seems that patients and observers respond most favorably to therapists who make use of these three categories of behavior. A further discussion of these results may be found in Chapter V. Table 11. Pearson correlation coefficients for important variables. Patient Direct Special Inter- Rating Question Praises Techniques Touches prets P.R. --- .500 .535* .565* .567* -.006 DQ .500 --- .109 .071 .261 .140 Pr .535* .109 --- .853** .735** -.206 ST .565* .071 .853** --- .852** -.492 T .567* .261 .735** .852** -—- -.387 I -.006 .140 -.206 -.492 -.387 --- *Significant at the .05 level. **Significant at the .01 level. 24 Table 12. Spearman correlation coefficients for important variables. Patient Direct Special Inter- Observ. Rating Quest. Praises Tech. Touches prets Rank P.R. --- .249 .399 .746** .576* .079 -.742* D0 .249 --- .043 .207 .636* -.103 -.310 Pr .399 .043 --- .703* .552* .152 -.757** ST .746** .207 .703* --- .642* .145 -.748** T .576* .636* .552* .642* --- .067 -.724* I .079 -.103 .152 .145 .067 --- -.067 Obs. -.742* -.310 -.757** -.748** -.724* -.067 --- Rank *Significant at the .05 level. **Significant at the .01 level. CHAPTER V DISCUSSION What are the implications of the results? First, let us summarize the findings, and then discuss what the results imply concerning the broader problem to which this study was addressed: that of developing a clear taxonomy of effective group leader behavior. The most clinically significant finding was the differences in three behaviors found in therapists who were responded to favorably by their patients and by the observers. It was discovered that therapists who used special techniques, touching and praising behavior were responded to most favorably by the patients and observers. Since the observers were asked to rank the therapists in terms of effectiveness, we may say that the therapists who used these three types of behavior were judged to be the most effective by the observers. However, whether or not we may call the patient ratings a measure of effectiveness is questionable. The patients responded to five questions about the therapists. These included questions about how much the patient felt cared about and understood, as well as questions about the therapist's competence. For the 25 26 purpose of this study, we will call the patient's total rating on this scale a measure of therapist effectiveness. Reinforcement for this assumption is provided by other results of this study. A comparison of patient ratings with observer rankings showed a very high amount of agreement between these two measures. This suggests that both were measuring a similar component. This is further reinforced by the results of Observers' agreement with each other and by patients' agreement with each other. The patients who rated each therapist were almost totally different each time. This was a result of observa- tions being spread over a number of weeks. However, the results of the patient ratings are very interesting. None of the therapists ranged more than three points in their average scores for each group. As an example, Therapist l scored approximately 22, 23 and 24 as his average scores by the patients, while Therapist 10 scored 14, 14 and 17. This suggests that patients were able to make discriminations among therapists. Also, the observers agreed among themselves in whom they selected as the most and least effective therapists. All of the observers visited the therapist's groups at dif- ferent times. Since the patients and observers agreed with each in whom they selected, it seems safe to state that a common element was being measured. We will call this thera- pist effectiveness. 27 Implications of the Results: ’New Directions How do we explain the importance of the three categories which seem to differentiate therapists? This question can be most easily answered by comparing the results of this study with those of the Yalom and Lieberman (1972) study. They found that the most effective leader- ship style combined high caring with the use of meaning- attribution. The least effective leaders in their study were low in caring and did very little meaning-attribution. Effectiveness was measured by attitude measures, value measures, self-ratings of change, self-esteem and self- ideal discrepancies, conceptions of others, measures of the person's propensity to use adequate or inadequate c0ping strategies, leader evaluations, judgments by the participants' social network and interviews. From these measures each participant was judged to be either a high- learner, a moderate-changer, a non-changer, a dropout or a negative changer. The leaders who used high caring with meaning-attribution were by far the most effective in pro- ducing positive changes while minimizing the number of participants who were negative changers, dropouts or casualties. Two types of behavior were subsumed under the dimension caring as defined by Yalom and Lieberman. One was offering friendship, love and affection, while the 28 second was giving support, praise and encouragement. These two types of behavior are almost synonymous with the cate- gories, touches and praises, which were found to be significant in the present study. Therefore, although the types of groups were different in the two studies, similar categories were found to differentiate therapists. Also, the methods for differentiating therapists were different; yet, the results were the same. It seems that patients respond most favorably in a very supportive atmosphere. The category special techniques was also found to be significant in the present study. This may be related in some ways to meaning-attribution as defined by Yalom and Lieberman. Meaning-attribution represents the naming function of leader behavior, wherein the leader gives meaning to experiences that members undergo. It refers to the translation of feelings and behaviors into ideas. Members perceived leaders who did not assume this function as more like peers. In this study, the leaders who used special techniques, directed gestalt work or role playing. Perhaps, in doing So, the leaders were perceived by the patients more as leaders and less like peers. Also, meaning- attribution referred to sharing a theory about how change takes place with the group. The use of special techniques may have given the same feeling to the patients in the group. Role playing is a means of practicing a behavior that a patient might want to change. Gestalt two-chair work for 29 conflict resolution was another special technique used. The therapist who used this might have told the group that it was necessary to resolve conflict for any change to be able to occur in the individual. Therefore, the finding of special techniques as a significant category, does not seem to be unrelated to earlier findings. Since no outcome measures were readily available in this study, it would be helpful to repeat the study in on- going therapy groups. Perhaps, this could be done by using groups in an outpatient clinic. In this way information from a patient population could be obtained in the same way as the present study with the addition of outcome measures similar to those used by Yalom and Lieberman. This new study could also provide valuable information about the relationship between patient ratings and outcome in group therapy. In other words, do patients who rate therapists highly receive the most benefit from group therapy? The results of the present study provide valuable information towards developing a taxonomy of effective group leader behavior. These results along with earlier findings have implications for new approaches in training group therapists. Usually a therapist is trained in a particular orientation such as psychoanalytic therapy or gestalt therapy. Rarely is any emphasis placed on effective characteristics of a good therapist. Research has shown that orientation is not as important as certain therapist 3O characteristics are in producing positive change. This is not to suggest that therapists not be trained in a particular orientation; but to emphasize the need for additional elee ments in the training. Therapists should be taught what types of behavior are effective in creating change. In this way patients will have a better chance of realizing their goal in therapy: that of constructive personality change. CHAPTER VI SUMMARY The purpose of this study was to determine effective, behaviOrs of group therapists. Effectiveness was measured by two methods. First, patients were asked to rate their group leaders on a five question scale and second, the observers rated the therapists after they had observed all of them. Ten therapists participated in this study. Each of them conducted three groups several weeks apart. The patients in each group varied as all of the groups were conducted in short-term inpatient hospital units. After each group, the patients in the group rated the therapist on a five point scale. Each question on the scale was scored on a 5-4-3-2-1 basis. An average score for each group was then computed. This was called the patients' rating of therapist effectiveness. Three observers were trained to observe the thera- pists on a twelve category behavior scale. All of the observers had previous work experiences in psychiatry. At the conclusion of all their observations, the observers ranked the therapists in order of whom they felt to be most and least effective in conducting group therapy. 31 32 All of the categories of behavior were correlated with each other and analyzed to uncover differences between the highest and lowest rated therapists. Three types of behavior were found that seemed to differentiate therapists. These were praises, touches and Special techniques. These results seem strong when viewed in relation to the other findings. Patients and observers agreed with each other on whom they selected as the most and least effective thera- pists. Thus, the three significant categories of behavior correlated with both theipatients and observers ratings of the therapists. Also, patients agreed over time in whom they liked and did not like as therapists. This was true even though different patients rated the therapists each time. Finally, the observers agreed with each other in whom they selected as the most effective therapists. The results of this study were very similar to earlier findings from different types of groups. This was especially encouraging since the methods for obtaining the same results were somewhat different. The results of this study can provide new information for training effective therapists. In the future, therapists should be trained to offer praise, warmth and encouragement to their patients. In this way patients will be able to move towards their goal of constructive personality change. BIBLIOGRAPHY , . "6'1!“ BIBLIOGRAPHY Baehr, G. O. The comparative effectiveness of individual psychotherapy, group psychotherapy, and a combina- tion of these methods. J. Consult. Psychol., 1954, '18, 179-183. Barrett-Lennard, G. T. Dimensions of therapist response as causal factors in therapeutic change. Psychol. Monogr., 1962, 16, No. 562, 1-35. Bion, W. R. Experiences in groups. New York: Basic Books, 1959. Cadman, W. M., L. Misbach and D. V. Brown. An assessment of round-table psychotherapy. Psychol. Monogr., 1954, 68, (13, whole no. 384). Ends, E. J. and C. W. Page. A study of three types of group psychotherapy with inebriates. Quart. J. Stud. Alcohol., 1957, 18, 263-277. Feitel, B. Feeling understood as a function of a variety of therapist activities. Dissertation Abstracts, 1968, .29 (3B), 1170-B. Guilford, J. P. Fundamental statistics in psycholo and education. New York: McGraw-Hill, Inc., 1556. Hays, W. L. Statistics for psychologists. New York: Holt, Rinehart, 8 Winston, 1963. Heckel, R., G. Holmes and C. Rosecrans. A factor analytic study of process variables in group therapy. J; Clinical Psych., 1971, 21, 146-150. Hobbs, N. and G. R. Pascal. A method for the quantitative analysis of group psychotherapy. Amer. Psychol. 1946, l, 297. Holsti, O. R. Content analysis for the social sciences and humanities. califOrnia: Addison-Wesley, 1969. 33 34 Lieberman, M. A. Behavior and impact of leaders. In L. Solomon and B. Berzon (eds.), New perspectives on encounter grou s. San Francisco: Jossey-Bass THC.,41972, pp.135-l70. Lieberman, M. A., I. Yalom and M. Miles. Impact on partici- pants. In L. Solomon and B. Berzon (eds.), New perspectives on encounter groups. San FranciSEo: Jossey-Bass Inc., 1972, pp. 119-134. Moustakas, C. E., and H. D. Schalock. An analysis of therapist-child interaction in play therapy. Child Develop., 1955, 26, 143-157. Peyman, D. A. B. An investigation of the effects of group psychotherapy on chronic schizophrenic patients. Group Psychother., 1956, 9, 35-39. Rogers, C. R. The necessary and sufficient conditions of therapeutic personality change. J. Consult. Psychol., 1957, 2_1, 95-103. Slavson, S. R. The fields ofggroup psychotherapy. New York: International Universities Press, 1956. Truax, C. B. The process of group psychotherapy. Psychol. Monogr., 1961, 15, (7, whole no. 511). Truax, C. B., and R. R. Carkhuff. Toward effective counsel- ing and psychotherapy: traifiing and practite. Chicago: AIdine Publishing Company, 1967. Tucker, J. B. Group psychotherapy with chronic psychotic soiling patients. J. Consult. Psychol., 1956, 20, 430. Walker, H. M. and J. Lev. Statistical Inference. New York: Holt, Rinehart 8 Winston, 1953* Yalom, I. and M. Lieberman. A study of encounter group casualties. Archives of Gen. Psychiat., 1971, 253 223-254. APPENDICES APPENDIX A THE THERAPIST RATING SCALE AND CATEGORY DEFINITIONS mmmuDOH mm: Hzmumh Ammmmo mmmmzmz abomu mmmho ZH mmHH mm>ommmHhummHmuzoz onhmmD m>HeummHm mmoH> 38 "I111111117111AllS