GROUP THERAPY AND INDIVIDUAL THERAPY: A COMPARISON Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY BARRY D. GRAFF 1969 THESis U ILIIIII IIIflIIIIIIIII III III II IIIIIIII 830 I This is to certify that the thesis entitled GROUP THERAPY AND INDIVIDUAL THERAPY: A COMPARISON presented by Barry D. Graff has been accepted towards fulfillment ' of the requirements for flew “£74417 Z I, a'or r0 ess ,, if, 47 L LIB R A R Y Michig.‘ :. \‘tatc UHIVCI slty I I. g NELEI'IUE" "ll , ‘ BINBING 9v - > HUAS&SUNS' I g . 1W! ml: I L05. . ABSTRACT A COMPARATIVE STUDY OF GROUP AND INDIVIDUAL THERAPY by Barry Graff The purposes of this study were to assess the relative effects of group and individual psychotherapy and to explore the suitability of the Argyris Interpersonal Competence Scale as a therapy outcome measure. Forty-four college students in three treatment categories, group therapy (§_= 17), individual therapy (3 = 12), and no therapy (§_= 15), served as Se. Therapy consisted of an average of fourteen sessions for all clients, two-hour sessions for groups, and one-hour sessions for indivi- duals. §s participated in two assessment meetings, one before and one after therapy. In each meeting the S3 completed a Q Sort, the Interpersonal Checklist (101), the I, You, OK-NG measures, and the Openness (0), Data- Seeking (DS), and Self-Disclosure (SD) scales. All §s also participated in two group problem solving sessions, to which their contributions were rated employing Argyris' Interpersonal Competence Scale. Friends of the SS, 31 "others", described the §s on the IC1, and the 0, DS, and SD scales both before and after therapy. Before therapy, 36 Se identified three dimensions on which they wished to change. Significantly (E < .01) higher dropout rates occurred among both the individual therapy and non-therapy §s than occurred among the group Barry D. Graff therapy gs, and this proved a major problem. Firstly, these differential dropouts undermined the meaning of comparisons between the three treat- ment groups. An intercorrelation matrix of pre-treatment scores was examined, and a "self-enhancement" cluster made up of the Q Sort, Interpersonal Check- list dominance-submission dimension, and the I, OK-NG dimension, was most prominent among the self-ratings. Salient among problems with these measures was the frequent occurrence of false positives on the Q Sort, and the strongly favorable bias of Self-Disclosure self-ratings. The hypotheses that group therapy §§ and individual therapy §s would evidence greater positive change than no therapy §s were not supported. Hypotheses that individual therapy would be superior on the Q Sort, and that group therapy would be superior on the Argyris Scale were not sup- ported. An uncertain tendency of the individual therapy remnant to change more positively than the group therapy sample must be considered in the light of the substantial dropout rate differences. Also, considering the number of significance tests run, these few "significant" differences could well be attributable to chance. The inability to secure adequate and comparable group and individual therapy samples precluded any rigorous testing of the central hypotheses. However, several tentative explanations were offered of the near chance level findings which emerged in an attempt to facilitate further research. Thusly, a number of alternative explanations of the failure of group and individual therapy to evidence a clear advantage over no therapy were offered, one being the inappropriateness of'the control group. Also cited Barry D. Graff was the presence of antithetical processes, that is, some clients chang- ing negatively and canceling out positive change in other clients, in therapy experimental groups. Additionally, there was the relative in- experience of the therapists, especially as group therapists, and the rather brief period of psychotherapy, again especially group therapy. Finally, the difference between group therapy process and individual therapy process was discussed. The Argyris Scale was found to have high scorer reliability, but low test-retest stability (r = .10). Perhaps due to the‘fis lack of involvement in the group problem solving sessions, it did not relate meaningfully to therapy outcome. Corrective suggestions were offered. A number of future studies were suggested to explore and control the various problems which appeared in this research. A comparison of premature terminators with clients who remained in therapy was proposed as an attempt to understand the "dropout" problem. Improved research design and control were considered necessary to the solution of this problem. An "own control" design was proposed to control client motivation, and repeated testings were suggested as an effective means for comparing group and individual psychotherapy process. flifi 975/44 GROUP THERAPY AND INDIVIDUAL THERAPY: A COMPARISON By Barry D} Graff A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1969 To Bonnie, my individual, and the interns, my group ACKNOWLEDGEMENTS There are many peOple who are a part of this project, most of whom deserve far more praise than the brief mention I will make of them here. The help of John Hurley has been frequent and significant. It made the completion of this project possible. Bill Mueller, Norm Abeles, and Dozier Thornton contributed greatly to the growth of this dissertation and its author. My wife, Bonnie, contributed the warmth and support to this endeavor that she has contributed to all my endeavors. It is Special. Kees Hofman was a diligent and effective co-rater. His aid was well- received. Many friends' interest and kind words were of inestimable value in keeping the work moving along. Finally, the therapists and SS provided the life blood of this study, giving many hours of their time. I thank all these people for their generous contributions. iii ACKNOWLEDGMENTS . . . LIST OF TABLES . . . . . LIST OF APPENDICES . . . Chapter I. II. III. INTRODUCTION . . TABLE OF CONTENTS Statement of the Problem . Review of the Literature METHOD . . . . Subjects . . . . Criteria for Selection Procedure . Therapy . . Measures . . . Q Sort . . . Interpersonal Checklist Dimension Ratings Argyris Interpersonal Competence Dimensions Chosen by §S Testing Procedures . Hypotheses . EXPloratory Question . Treatment of Data RESULTS Dropout Rate . Hypothesis Testing . Correlations Between Pre-Treatment Measures Among the self-ratings . ratings . Between the self- and Among "others" Argyris Scale Group Therapy VerSus No Therapy "others" Scale ratings . Individual Therapy Versus No Therapy . Group Therapy Versus Individual Therapy: Argyris Scale Individual Therapy Versus Group Therapy. Q Sort . iv The The Page iii vi vii 10 1o 11 11 11 12 12 13 13 14 14 14 16 17 _17 20 20 20 22 22 24 24 24 24 27 28 28 Chapter Argyris Scale . . . . . . . . . . Reliability . . . . . . . . . . Validity . . . . . . . . . . . . EXploratory Question . . . . . . . . Other Findings . . . . . . . . . Ratings on Dimensions Chosen by Expert Criteria . . . . . . . . . IV. DISCUSSION . . . . . . . . . . . . . . Dropout Rate . . . . . . . . . . . . Pre-Treatment Measures . . . . . . . Intercorrelations of Variables . . Among the self-ratings . . . . . Among "others" ratings . . . . . Between the self— and "others" ratings Representativeness and adequacy of measures Self-Disclosure Scale . . . . . Q Sort . . . . . . . . . . . . . Argyris Interpersonal Competence Other Measures . . . . . . . . . ”OVerall Evaluation . . . . . . . . General Findings . . . . . . . . . . The Argyris Scale . . . . . . . . . Dimensions Chosen by §3 . . . . . . Future Research . . . . . . . . . . REFERENCES 0 O O O O O O O O 0 O 0 O O O O O O APPENDICES . . . . . . . . . . . . . . . . . . Scale . Page 29 29 29 29 3O 3O 3O 32 32 34 34 34 34 35 35 ’35 36 38 4O 41 41 47 48 49 54 57 Table LIST OF TABLES Pre-Treatment Means and Mean Differences of the Three Treatments . . . . . . . . . . Correlations Between Pairs of Outcome Measures Using Pre-Test Data . . . . . . . . . . Mean Changes and Mean Between-Groups Differences by the Three Treatments . . . . . . . . . . . . . Comparisons of Proportion Changes by Treatments . . Overall Results - Positive Change for Three Treatment Groups Combining All Outcome Measures Number of Participants Reaching EXperts' Criterion on Each Outcome Measure Before (B) and After (A) Therapy . . . . . . . . . . . Relationship Between Therapist Ratings and Overall Outcome Scores . . . . . . . . . . vi Page 21 23 25 26 27 31 43 LIST OF APPENDICES Appendix Page A. Additional Tables . . . . . . . . . . . . . . . . . . . 57 B. Measures . . . . . . . . L . . . . . . . . . . . . . . 71 C. Additional Forms . . . . . . . . . . . . . . . . . . . 91 vii CHAPTER 1 INTRODUCTION Statement of the Problem The purpose of this study is two-fold. First, to study the relative and differential effects of group therapy, individual therapy, and no therapy. Secondly, to test the reliability and validity of a new therapy outcome measure, an interpersonal competence scale. Review of the Literature The outcome of psychotherapy has been a subject of the research literature for over half a century. From the case studies of Freud (1958), to more recent therapist-centered genuineness, warmth, and empathy studies (Truax, 1963, and Truax and Carkhuff, 1965), the question has often been, "Does psychotherapy work?" unfortunately, the continued research has made it obvious that this is not a very meaningful question. There have been many problems in this research, one of the greatest being the choice of an appropriate outcome measure. A root of this dif- ficulty is that therapeutic outcome remains an ambiguous and elusive concept. Many theoretical attempts have been made, but few therapists, with the possible exception of behavior therapists, believe it useful to name specific criteria for improvement in the individual client. Little wonder, then, that when their clients are used in research studies, the success of their therapy contacts is difficult to ascertain. These problems multiply when individual and group psychotherapy are studied simultaneously. To attack these problems, it seems useful first to inSpect the 2 theoretical and eXperimental literature comparing group and individual therapy. In both cases, articles have been scarce. Authors have seemed to dwell on one or the other, rarely attempting to focus on their paral- lel, although often divergent, aims and characteristics. In an early paper, Shea (1954) compared analytic group and individual therapy, with a focus on the comparison of the client's resistance reac- tions in each. Using case studies, he found that group analytic therapy had the practical advantage of costing less than individual therapy, thus making it available to more pe0ple. In addition, group therapy was found to be a more efficient way of effecting changes in certain resistances. Some were dissolved more quickly than would have been the case in indi- vidual; others disappeared that had not been affected at all by individual therapy. On the other hand, he found group therapy to be inferior in some cases. He cited the relative lack of control over the course of treatment by the group therapist as a cause of these inferiorities. Finally, Shea listed contraindications for group therapy. These were patients with a diminished ability for interpersonal relationships, patients with brittle character structures, and patients with very am- bitious therapeutic goals. Kubie (1958) asked questions about the two therapies rather than v”- making theoretical statements, and he eSpecially wondered about the potentialities of group therapy. For example, he questioned whether a group can communicate a client's unconscious material to him, a task the individual analyst is alleged to do quite well. Kubie further Speculated whether group therapy can produce as deep and as far-reaching changes as individual analysis does. These questions arise, he stated, because group 3 therapy departs from the essence of analysis, the attempt to understand one unconscious at a time. In addition, though, he postulated that group therapy might form a bridge between individual therapy and the "real world", because of the presence of people besides the theoretically uninvolved, unreactive analyst. Although limited to discussions of analysis, these papers do present two of the basic questions that studies of these two techniques must answer: (1) Are they equally effective? and (2) Are there peOple or processes that can be affected by one, and not the other? The few re- search studies in this area begin to answer these questions. Baehr (1954) used a "Discontentment Scale" to compare individual and group therapy with each other and with a combination of both. Sixty- six hoSpitalized World War II veterans served as S8; non-directive therapy was practiced by both group and individual counselors. There were a num- ber of methodological difficulties. For example, the group clients often had a few individual sessions and vice versa. Therefore, the three experimental treatment groups were quantitatively, rather than qualita- tively, differentiated. The combined therapy category had the highest relative movement index, and the group therapy category the lowest, but none of the differences were significant. Nevertheless, when the one treatment categories were brought together statistically, combined therapy was found to be significantly higher than one therapy on the relative movement index at the .02 level. Novick (1965) found that either group or individual therapy proved to be effective in changing nondesirable behavior in children. §s were 4 forty-four, eight to ten year-olds undergoing outpatient therapy at a number of community mental health clinics. All therapists were eclectic, and used a combination of activity play and verbal communication in the sessions. Groups consisted of three to five children. Because of their apparent equivalent effectiveness and practical economy, Novick suggested the increased use of group methods in the treatment of this type of problem. Suinn (1968) used desensitization techniques to treat college students with high test anxiety. Group and individual methods both were effective in producing significant decreases in the test anxieties of §s. In addition, generalization to other anxieties occurred as a result of both methods. Suinn, like Novick, cites the economical use of therapist time as an advantage of group treatment. These three studies combine to make two conclusions quite clear. IFirst of all, the comparative study of group and individual therapy is ill the infant stage; a careful perusal of the reviews of the literature in.the International Journal of Group Psychotherapy from 1954 to the Irresent finds no other experimental studies which directly attack this problem. Secondly, a clear, but obviously tentative, equivalence can be postulated between the two methods for some relatively divergent problems. Three differing populations, child, late adolescent, and adult, treated Withtthree different therapy techniques, non-directive, play, and desen- sitlzization, all reSponded to group as well as individual therapy. However, because of their narrow foci, these studies do not portray the! many complications inherent in comparing two complex treatment methods. Surniberg and Tyler (1962) cite a number of these complications. 5 First is the usual problem of psychotherapy research, that of equating experimental and control groups. Many factors are relevant to this equation, but one of the most difficult to control is the motivation of the §s. Therapy applicants are usually more motivated to change than"‘ non-clients. Even more difficult is the problem of equating therapists. Using the same therapists for both the individual and group therapy §s equates their general ability. However, there is no guarantee that the same therapist has equivalent skills as an individual and as a group leader. Finally, the difference between the processes of group and indivi- dual therapy affect outcome a great deal. Discovering the stages of each process which evidence the most change is crucial if meaningful compara- tive research is to be done. Sundberg and Tyler (1962) present a study which portrays some of these complexities, although it does not solve them. Fairweather et a1. (1960) used a number of outcome measures to assess change in three diag- nostic groups given four different treatments. The diagnostic groups were: a) non-psychotics, b) short-term psychotics, and c) long-term psychotics. The treatments were: a) control or no Special treatment, b) individual psychotherapy, c) group psychotherapy, and d) group psychotherapy and group living. A three by four analysis of variance design was used. The results were varied, and occasionally inconsistent. There were some significant differences between diagnostic categories, some between treatments, and some between interactions, depending on the outcome mea- sure used. For example, there were differences in length of therapy for the various treatments, but length of therapy did not relate to post- hOSpital adjustment. Full time employment was the only follow-up measure 6 which found each of the three treatments to be superior to the control group. It is easy to see that even an ambitious study such as this one is only a beginning step. The present study worked on the same principles as the Fairweather et a1. (1960) study. It attempted to control for the complexities as much as was practically and methodologically feasible (see Method section), while being broad enough to portray those which remained semi- or uncon- trollable. Attempting to portray these complexities returns us to the problem of outcome measures. A number of dissimilar measures must be used, so that more differentiations between the two techniques than those refer- ring to economy can be made. Even with this array of measures, the narrow range of clients (no psychotics, all college undergraduates) will restrict the generalizability of the results. Therefore, this study began with a search for instruments which would reflect the values of the author, his colleagues, and other contem- porary therapists in their work with real clients, in individual and group psychotherapy. The instrument given the major focus in this study is the Argyris Interpersonal Competence Scale (1965). In studies with T-groups, this instrument has shown a high degree of reliability and validity in measur- ing the attribute, interpersonal competence, that its name implies. Its use is enhanced by Argyris' use of three types of behaviors to indicate interpersonal competence: owning, opening up to, and eXperimenting with ideas and feelings. These three types of behavior are often a focus of therapy with an interpersonal emphasis. Because of the newness of this 7 scale, only two studies employing it have been published (Argyris, 1965, I & II), it will have a dual role in this study. It will serve as an outcome measure, as well as having its own reliability and validity further tested. A number of the other instruments we will use attempt to measure behavioral attributes similar or related to those presumably appraised by the Argyris Scale. The Interpersonal Checklist (ICl; Leary, 1957) plots a multitude of categories into two orthological dimensions, love-hate and dominance- submission, which describe the S's interpersonal contacts. In a factor analytic study of the 101, Briar and Bieri (1963) found three main fac- tors: dominance, love, and inferiority feeling. Thus, although their octants were slightly different, their study supported the existence of Leary's dominance-submission and love-hate dimensions. This instrument has been used often in personality research, using self and others' ratings. Armstrong (1958) used two groups of 50 SS each, one normal group and one alcoholic group, to measure its internal consistency. Six different ratings were found to be highly reliable (py<.01) for both groups. The Self-Disclosure dimension was first studied by Jourard (1958), and in its initial form showed inconsistent validity (Hurley, 1967). A newer instrument (Hurley, 1967) oriented to self-disclosure, employing .. both self and others' ratings, seemed to show that self ratings alone .' tend to be invalid. The Openness and Data-Seeking dimensions attempt to measure behaviors similar to those measured by the Argyris Scale, although they have not been studied in research. Openness seems to correSpond to Argyris' owning, and data-se tion catego of the Wind It wil these three tion as its perceptions An in: appears in Haigh Q So interperso- I‘EIiabilit and it has PSYChotics the Change and Butler Suit of it In a Baroff (1‘ therapy a thetapy p ten Once- schizophr two treat group We 1- [W0 expe r 8 and data-seeking corresponds both to Argyris' openness and eXperimenta- tion categories. Together, these dimensions attempt to reflect the size of the window between the individual and the outside world. It will be noticed that the Argyris Scale has the advantage over these three instruments of using behavior in a real interpersonal situa- tion as its source of raw data. The others depend, in great measure, on perceptions, by self and others, as interpersonal data. An instrument having a very different focus, and which frequently appears in individual psychotherapy research literature, is the Butlers _Haigth Sort. The self of the individual is studied, rather than his) interpersonal contacts, through his self and ideal perceptions. Its \ reliability has been found to be high (Frank, 1956, and Nahinsky, 1963),: and it has shown validity in discriminating normals, neurotics, and psychotics (Nahinsky, 1963 and 1966). Its main use has been in assessing the change resulting from individual psychotherapy. Both Rogers (1954) and Butler (1968) showed that self-ideal correlations went up as a re- sult of individual therapy. In a study eSpecially relevant to the present research, Satz and Baroff (1962) attempted to assess the effects of individual occupational therapy alone (thirteen one-hour sessions) and individual occupational therapy plus group therapy (thirteen one-hour individual sessions and ten once-a-week two-hour group sessions) upon sixteen non-paranoid schizophrenics. The self-ideal correlations were slightly lower for the two treatment groups after therapy, and the correlations for the control group were slightly higher. The difference between the control and the two eXperimental treatment groups combined was non-significant at the .10 9 level. The authors suggest a number of plausible explanations for this result, among them the possibility that the experimental period was too short. Williams (1962) reports a quite divergent result. Using a modified Q-technique, sorting into two rather than nine piles, he found that two or three sessions of individual educational-vocational counseling sig- nificantly increased adjustment scores, as measured by self-ideal-ordinary person correlations. Some studies (Phillips et a1., 1965), however, have questioned this instrument's validity in measuring psychotherapy induced changes, and Sundland (1962) has criticized the statistical soundness of it, eSpecially the spurious self-ideal correlations obtained when the individual items are highly intercorrelated with each other. A new 80-item sort (Butler, 1968), which eliminates some of these items, is being used in this study. The I, You, OK-NG dimensions are theoretical constructs (Berne, 1966), and have not seen much research use as yet. Again, self-perceptions are used to assess an individual's attributes; in addition, this instrument, like the Q Sort, makes predictions about an individual from his perceptions of others rather than others' perceptions of the individual. This range of instruments has been chosen for two reasons. First, because of their tentative nature as valid outcome measures, and secondly, because both group and individual therapy are being studied. The first group of instruments, especially the Argyris Scale, seem to be more atuned to group therapy variables; the Q Sort and the self-perception asPects of other instruments seem to be more atuned to individual therapy variables. CHAPTER II METHOD Subjects Forty-four undergraduates at Michigan State University served as SS, sixteen male and twenty-eight female. Seventeen were group therapy clients, six male and eleven female. Twelve were individual therapy clients, three male and nine female. Fifteen were not involved in therapy; seven were male, and eight female. The group clients were drawn from five groups, each containing from six to eight members and two therapists. All therapists were interms or senior staff members at the Michigan State University Counseling Center. Two group clients and one individual client were treated by senior staff members, and the rest were treated by interns. The interns were all Ph.D. candidated in Education or Clinical Psychology. They had had from three to five years of part-time super- vised psychotherapy experience. The senior staff members were Ph.D.‘s in Clinical or Counseling Psychology. They had had from five to seven years of psychotherapy experience. All of the therapists' training began with eXperience in individual psychotherapy, and later included experience in group psychotherapy. They had Spent more hours doing individual therapy than doing group therapy. In addition, they had all received supervision as individuals and in groups, the choice being unrelated to the type of psychotherapy they were performing at the time. Finally, they had all undergone their own personal therapy, individual and group. 10 fl 11 Criteria for Selection: All forty-four §5 were students who initiated contacts with the Michigan State University Counseling Center. The twenty-nine clients sought psychotherapy, the fifteen non-clients sought psychotherapy or other services, such as education-vocational counseling. A11.§s went through the usual screening procedure at the Counseling Center. This consisted of a screening interview, and assignment to therapy by the screening committee. The clients were assigned to group or individual therapy according to their own choices; however, the screen- ing committee had the power to override this choice if it seemed in the client's best interest. None of the clients were psychotic. All SS were volunteers for the research. They were shown letters (see Appendix C) describing the re- search by the screening counselor. When they agreed to participate, the E contacted them by phone to set up an appointment for the pre-testing sessions. Procedure Therapy Psychotherapy consisted of an average of fourteen sessions, with a range of twelve to fifteen sessions, once per week for all clients.* All individual sessions were one hour long; all group sessions were two hours long. The non-client group all attended zero or one session subsequent to the screening interview. Of those who attended one session, six were educational-vocational counseling sessions, and four were individual therapy sessions. The latter four terminated therapy after the first *One group client attended eight therapy sessions; upon consulta- tion with her therapists, she was left in the group therapy sample. She is §_number RB7805. 12 session. The group and individual psychotherapies had an interpersonal em- phasis, although they contained many eclectic elements. The content of the sessions often concerned the clients' present day relationships, although relevant past history data was explored if apprOpriate. Present day relationships included the "here and now" interactions between client and therapist, or, in the groups, between clients. Non-verbal techniques were often part of the therapy. The non-verbal techniques were used as apprOpriate in the course of therapy, rather than as pre-planned exercises. These techniques were similar to those presented by Schutz (1967). Because most of these techniques involve interactions between a number of peOple, they were much more predominant in the group therapy. Spontaneous non-verbal expressions of feelings, such as affection, were an accepted part of the therapy, although client, and, poSSibly, therapist defenses often prevented these expressions from occurring. Verbal discussion and insight often followed non-verbal interactions, whether they involved techniques or Spontaneous expression of feelings. The two psychotherapies thus converged in a number of ways. Theo- retical bases, content of the sessions, and techniques used were all similar. The differences, such as the greater use of non-verbal tech- niques in the group sessions, resulted fnam the differences in number of participants, rather than differing modes of attack by the therapists. assesses 9 Sort: An eighty-item Q Sort recently devised for psychotherapy research (Butler, 1968) was used. Each subject was asked to make self 13 and ideal sorts, and the two sorts were correlated. The eighty items were sorted into nine categories, labeled 0 to 8, with 3, 6, 9, 13, 18, 13, 9, 6, and 3 items placed in each category reSpectively. Interpersonal Checklist: This device contains 134 items which are to be checked true or false. The items are then combined algebraically into two orthogonal dimensions: love-hate and dominance-submission; the love and dominance ends of the scales are theoretically considered posi- tive. The two dimensions are looked at together, rather than individu- ally, when assessing change. Thus, a subjective element is often present when one dimension evidences positive change while the other evidences negative change. Several instances of rater disagreement occurred when the author and the dissertation chairman reviewed these scores. Dimension Ratingg: Five dimensions, Self-Disclosure, Openness, Data- Seeking, I, OK-NG (not OK), and You, OK-NG, will be rated to describe the gs. The Self-Disclosure dimension, originally identified by Jourard (1958), and more recently used by Hurley (1967), is Studied here through use of the Hurley Self-Disclosure Rating Scale (1967), an eight-point scale from not self-disclosing to self-disclosing, with definitions at each scale point. The Openness and Data-Seeking dimensions combine to form a "window to the world", so to Speak, for individuals who exhibit these behaviors. Each are nine-point scales from minimal to maximal, with a definition of each dimension supplied. The I, You, OK-NG dimensions also are rated on nine-point scales, the end points merely being labeled as not OK and 0K. Descriptive ad- jectives are supplied for the end points of each dimension. High or low 14 ratings on these two dimensions combine to form the four basic positions of Berne (1966): 1) I am OK, You are OK; 2) I am not OK, You are OK; 3) I am OK, You are not OK; and 4) I am not OK, You are not OK. Thus, as in the case of the ICl, the two dimensions are looked at together when change is being assessed. Argyris Interpersonal Competence Scale: This scale, developed by Argyris (1965), rates interpersonal competence manifested by individuals in group situations. The SS verbal productions are ranked on the indivi- dual and interpersonal scales, which are then combined to form the inter- personal competence score. The behaviors which facilitate interpersonal competence on the individual Scale are owning, openness, and experiment- ing, all on the ideational or feeling levels. Behaviors which inhibit competence on this scale are notowning, not being open to, and rejecting eXperimenting with ideas or feelings. On the interpersonal scale, the same three qualities and their opposites are used, except that the cri- terion is helping or not helping others to exhibit these qualities. All categories are weighted in the final computation (see Appendix B). The total of the weighted scores is tabulated for each S, and this total corrected if the number of productions of the S is less than the average for his particular problem solving group. Dimensions Chosen by gs: Thirty-six §s were asked to name three attributes on which he wanted to change, and to rank himself on a scale from 1 to 9 schowing where he was functioning on that attribute. Testing_Procedures Data were collected in three ways. First of all, each S picked two friends, one male and one female. These friends were to fill out the 15 Interpersonal Checklist, and the Openness, Data-Seeking, and Self- Disclosure rating forms about the Se. The §_mailed the test forms to the friends with an eXplanatory letter (see Appendix C). The completed forms were then mailed back to the Michigan State University Counseling Center. Five group clients had both friends reSpond before and after therapy, and eight had one friend reSpond. Three individual clients had both friends respond before and after therapy, and six had one friend reSpond. One non-therapy client had both friends reSpond before and after therapy, and eight had one friend reSpond. The last two forms of data collection were two testing sessions in which the SS participated. The first was a group problem solving session, with four §s participating. No more than two of each four were from any one of the three eXperimental treatment groups. The SS worked on one of two problems, the NASA Moon Problem or the George Edwards Case (see Appen- dix B). For each, the SS were asked to read about the problem situation and suggest a solution in the form of ranked alternatives. After their individual rankings were collected, they worked on a group solution and rankings together. This final phase was tape recorded. For the second testing session, each § produced two Q Sorts describ- ing himself and his ideal, completed the Interpersonal Checklist, and ranked himself on the five dimensions. The friend ratings and two testing sessions occurred twice, the pre-tests were before the second therapy session, and the post-tests were after the fifteenth (or last) therapy session. The non-clients had the two testings separated by the same Span, sixteen weeks, as the clients. (This consisted of fourteen weeks of therapy, and two weeks of Spring l6 vacation.) In the post-test group problem solving sessions, each §.worked on the George Edwards Case if he had earlier worked on the NASA Moon Problem, or vice versa. The groups of four were rotated, So that no more than two of the four had worked together in the first session. Typed transcripts of the tape recordings of the group problem solving sessions were rated by two judges independently. The judges were the E, and a fellow graduate Student in Clinical Psychology. They learned the rating system through the writings of Argyris (1965), and personal corres- pondences with him. They used the criteria in Argyris (1965a) to score each production. The reliability of these ratings was calculated. Finally, the two judges conferred on the judgments where they disagreed and tried to reach agreement. These final corrected ratings were used as the basis for the interpersonal competence scores to be used for hypothesis testing. ‘flypgtheses l) The group therapy clients will Show more positive change than the non-clients on all of the outcome measures. 2) The individual therapy clients will Show more positive change than the non-clients on all of the outcome measures. 3) The group therapy clients will show more positive change than the non-clients on the Argyris Interpersonal Competence Scale. 4) The individual therapy clients will Show more positive change than the group clients on the Q Sort data. 5) The Argyris Interpersonal Competence Scale will reliably and validly measure change resulting from group and individual therapy. 17 Exploratory Qeustion How will the group and individual clients compare on the remaining therapy outcome measures? Treatment of Data A number of operations were used to compare the reSultS for the two experimental treatment groups and the control group. First, the mean changes of each group on each outcome measure were compared. Then, the proportions of each group showing positive change on each outcome measure were compared. An overall change measure for each §nwas also calculated. Weights were assigned to each measure; these weights attempt to reflect the amount of data contained in each measure. The measures which produce one datum, the Openness, Data-Seeking, and Self-Disclosure dimensions, were given a weight of one. The I, You, OK-NG, scale produces two dimensions, and, thus, was given a weight of two. The E estimated that the ICl, A Sort, and Interpersonal Competence Scale produce approximately four times as much data as the unit measures; thus, they were given a weight of four. A positive or negative change on each measure was multiplied by that reSpective measure's weight. The weighted scores, ranging from -4 through 0 to +4, were then totaled. The three treatment categories were then compared in two ways. First, the proportion of §s in each category who showed a positive change on the weighted sum of individual measure positive change scores were compared. Secondly, the mean overall positive change scores for each category were totaled, and these three totals compared. 18 Finally, four experienced* therapists suggested levels on six out- come measures that they expected clients to reach at the end of therapy (see Appendix A). These meaSures were the Q Sort, the Interpersonal Checklist, and the Openness, Data-Seeking, Self-Disclosure, and I, You, OK-NG Scales. Then, criteria were averaged, and the number of SS in each eXperimental treatment group who reached the criteria before and after therapy on each meaSure was calculated. Significance tests of the difference between proportions in each group on each measure from pre- to post-testing were calculated. Hypothesis 6 was treated in the following way. The reliability of the Argyris Scale was measured by calculating the percent agreement on all scored units between the two tape recording raters. In addition, the raters discussed the units they scored differently and attempted to reach agreement. A second percent agreement was then calculated. Also, the test-retest reliability was calculated using the pre- and post-test scores. The second part of the hypothesis concerns the concurrent validity of the scale. To measure this, the therapists ranked the clients on a scale from 1 to 5. A score of one meant that the client had changed less than almost all of the clients that that therapist had ever seen; a score of three meant that the client had changed an average amount; and a score of five meant that the client had changed more than almost all of the clients that that therapist had ever seen. Clients with scores of 4 or 5 were put in an "improved" category, and clients with scores of 1 or 2 were *All had at least five years of eXperience in psychotherapy. 19 put in a "not improved" category. These gs were then divided into halves, improved and not improved, using the Argyris Scale scores. Correlations between this division of §5 and the therapist assignments to improved and not improved categories were then calculated, and Significance tests run on the values obtained. CHAPTER III RESULTS Dr0pout Rate Fifty-twoifis completed the pre-testing; 17 in the group therapy treatment group, 20 in the individual therapy group, and 15 in the non- therapy group. One group therapy S dropped out of the therapy after eight sessions, but remained in the research; four individual therapy SS drOpped out of the therapy after one session, but remained in the research, and were switched to the non-therapy group; four additional individual therapy §S dropped out of the research; and four non-therapy §S dropped out of the research. The grave sampling issues posed by these reductions in the original treatment groupings will be recognized by using the term "remnant" in subsequent discussion (but not tables) of the reduced individual treatment sample; "reconstituted" will be similarly applied to the no-treatment sample. Using two-tailed 3 tests (Peatman, 1963), the group therapy treatment group was found to have Significantly fewer (p <.Ol, two-tailed test) dropouts than either the individual therapy group or the non-therapy group. Hypothesis Testing Table 1* presents the pre-treatment means of these three groups on all of the outcome measures. The means of these three groups varied on all measures, but the only statistically Significant differences were between the reconstituted non-therapy group and the other two treatment groups on the self-reported Self-Disclosure Scale. Obviously, the *Reduced NS for the "others" measures in all tables are due to the failure of the Ss' friends to return the test forms. The reduced NS for Argyris Scale resulted from mechanical failure in the recording of one problem-solving group. 20 21 TABLE 1 Pre-Treatment Means and Mean Differences of the Three Treatments Treatments Differences Measures Group(G) Indivi- None(N) G-I G-N I-N dual(I) 5-17 N=12 §;15 Self Q Sort .23 .44 .35 - .21 - .12 .09 101 -6.5 -5.8 -2.1 - .7 -4.4 -3.7 D8 101 5.9 0.5 3.3 5.4 2.6 -2.8 LH 1, OK-NG 5.1 5.3 6.4 - .2 -1.3 -1.1 You, OK-NG 5.5 6.4 5.6 - .9 - .1 .8 Openness 5.8 5.1 6.3 .7 - .5 -1.2 Data-Seeking 5.2 4.8 5.9 .4 - .7 -l.1 Self-Disclo- 5.8 5.5 6.9 .3 -1.1* -1.4* sure Other Ic1DS -1.9(13) -0.9(8) 4.2(9) -1.0 -6.1 -5.1 IClLH 5.9(13) -1.5(8) 3.6(9> 7.4 2.3 -5.1 Openness 5.5(13) 6.2(9) 5.1(9) - .7 .4 1.1 Data-Seeking 5.7(13) 5.8(9) 5.7(9) - .1 0 .1 Self-Disclo- 5.8(12) 5.9(9) 5.6(9> - .1 .2 .3 sure External Argyris .80(16) .65(10) .58(13) .15 .22 .07 Note. "‘ES as indicated at top except when given in parenthesis; include only those "others" who gave both pre- and post-treatment data. *2 <.05, two-tailed 22 lower pre-treatment means provide greater room for change, and these differences must be considered when assessing any significant between- groups findings. Correlations Between Pre-Treatment Measures Table 2 contains the product-moment intercorrelations of the remain- ing 44 SS on the fourteen variables - the ICl has been divided into its dominance-submissiveness (DS) and love-hate (LH) factors and the I, You, OK-NG measure has been divided into its two dimensions for use in Table 2. In five cases, the "other" scores were not available; substituted in these instances was the mean score of the other SS in the same treatment category on each missing outcome measure. Twenty-three of the Ss had one friend respond before therapy, and sixteen SS had two friends reSpond.* Employing the two-tailed test of Significance, twenty-two of the ninety- one correlations were significant at or beyond the .05 level; three of these were negative. Among the self-ratingS - The intercorrelations between the self- rating on the Self-DiscloSure Scale and the other "self" measures were significant at the .05 level or better in five of seven instances, one of these being negative. The self-rating of the Self-Disclosure Scale also forms a cluster with the self-ratings of the Openness and Data- Seeking Scales, each intercorrelation being significant at the .05 level. Finally, the self-rating of the IClD Scale, the Q Sort, and the S *Table 2 cannot be reconstructed from the pre-treatment data given in the Appendix because these latter excluded some cases of unmatched (from different "other" persons) pre- and post-treatment data, while all available data were included in the Table 2 intercorrelations. cusses-030 .400.v .mo assess-030 .H0.v .m assess-030 .00.v .m a 23 m mhfim £00. £00. Ha. N0. #0. HN. 0N. 00. #0.- 0N. 00. N0. 600.- -0s60s0-0a60 Asav -- 00. 00.- man. «an. “a. Na. 0a.- 0m. 00.- 0N. as. 0a.- wasxomm-muma Anne -- 0H.- 00.- 00.- NH. N0. 00. «0.- 0N. HH.- a0. 00.- 006006a0 Auav -- 6N0.- 00.- #0.- mam. 0N. am. 000. 00. 00.- 00.- maaoH ASHV -- nus. n00. 0s. 00.- gas. 0a.- 000. AN. “0.- 00H0H Aoav soauo wow—am --. 000. 0am. 600.- 00¢. 50. 00s. mm. «0.- -030000-0H60 A0 0 -- am. 40.- as. 00. 0N. 00. ms. wassmmm-auma Am C -- 00. 6am. 0N. «N. “a. HH.- 006a0600 An 0 -- N0. HN. 0H.. 0H.- 0N.- oz-M0 .00s Ao V 1111 «a. nod. nm¢. no.1 UZIMO .H Am v as :1:- mH.- no.1 mo. HUH Ad v 00 unu- omm. HH.1 aoH Am v -- Ho.- 000m o a~ v seam 0H NH as 0H 0 0 a 0 m 0 m 1m. H -wuuudduuuuumdo pouuomwuumHom maMNfiMfi. wounded: wounded: moon puma-mum magma ashamed: osoouso mo mufiom coosuom maofiumfiouuoo N mqm 0.2 -0.1 -0.3 Self-Disclo- 0.1(12) 0.2(9) 0.4(9) -0.1 -0.3 -0.2 sure EEEEEESL Argyris 0.01I16) 0.04(10) -0.10(13> -0.03 0.11 0.14 Note. -- SS as indicated at top except when given in parenthesis; include oniy those "others" who gave both pre- and post-treatment data. #2 <.05, one-tailed *TB .44(9) .44(9) .18 .18 .00 Data-Seeking .38(13) .33(9) .56(9) - .05 - .18 .23 Self-Disclo- .42(12) .44(9) .33(9) .02 .09 .11 sure External Argyris .50 .40 .39 .10 .ll .01 Note. -- Ns as indicated at top except when given in parenthesis; include only those "others" who gave both pre- and post-treatment data. *2 0 0(9) 0 2(9) Self-Disclosure 4 5(12) 2 3(9) 2 2(9) Note. -- Ns as indicated at top except when given in parenthesis; include only those "others" who gave both before and after therapy data. *fp < .01, one-tailed CHAPTER IV DISCUSSION Dropout Rate The findings show that SS in the group therapy treatment category were more likely to remain in both therapy and in the reSearch than SS in the other two groups. This might be due to a greater initial commit- ment from clients who agree to enter group therapy, or it could be a result of group therapy exerting a strong hold on its clients. The first seems feasible because committing oneself to two hours and eight pe0p1e a week appears more demanding than committing oneself to one hour per week to only a psychotherapist, let alone to merely two testing sessions. The second also seems feasible; a group therapy client would probably be less likely to fool himself into a premature termination due to the con- tradictory feedback he would receive from his fellow group members. Whatever the eXplanation may be, in this Study group therapy provided the distinct advantage of more frequently keeping its clients. The individual clients who terminated after zero or one session may have fooled themselves into leaving therapy prematurely. This could exemplify the Hathaway "Hello-Goodbye" effect, as cited by Sundberg and Tyler (1962). Briefly, a client wants to believe he is "sick" when he enters therapy, and "well" when he leaves, eSpecially when he is being tested on his progress. The significant dropout rate differences must be considered when interpreting all other findings of the present study. The seemingly slight advantage shown by the remnant individual therapy treatment group must be viewed in the perSpective of its dropout rate of 40% (8 of 20), 32 33 as compared with the 6% (1 of 17) group treatment rate. The shift of four'Ss from individual therapy to the reconstituted no-treatment group also affected the reconstituted no-treatment group results. All four SS showed positive Q Sort changes. These change scores of +.232, +.393, +.464, and +.50 (averaging +.510) were excep- tionally high; the average S remaining in either group or individual treatment averaged a Q Sort gain of only +.090. These data may demon- strate the "Hello-Goodbye" effect, with the SS trying to convince them- selves of their readiness for termination by making their self and ideal images coincide. The three of these four SS with the higher Q Sort change scores had no friends respond in the post-testing. This may have been a result of their need to avoid contradictory feedback which might be in- consistent with their "Goodbye" perception. The overall research design was seriously compromised by both this shift and by the "dropout" of four persons from the original individual treatment group, and four more from the original no-treatment group. Because data from these reclassified and missing SS could conceivably have reversed the few statistically significant between-treatment group findings described, these losses and shifts undermine any clear Statements about the pertinent original hypotheses. Thus, the outcomes of comparisons between treatments administered to such vaguely defined samples as 100% of -the group SS, 60% (12 of 20) of the original individual Ss, and 73% (11 of 15) of the original no-treatment SS can only be viewed as suggestive. In addition to the Specific reasons already cited, a general reason for the failure of SS and "others" to participate in the post-testing was the time of the year. The final testing coincided with the final two 34 weeks of the school year, and, since all SS and most of the "others" were college students, their busy schedules kept a number of them from participating. These busy schedules may have been convenient excuses used to mask resistance, but this cannot be assessed. Pre-Treatment Measures Intercorrelations of Variables Among the self-ratingS - The measure accounting for the most co- variance within the matrix of self-reported pre-treatment scores was the self-rating of the Self-Disclosure Scale. Interestingly, this was the only outcome measure which differentiated between the treatment groups in terms of mean changes. The cluster this measure formed with the self- ratings of the Openness and Data-Seeking Scales may be due to a common "communication" factor, although this common variance could also be a result of the Similarities among the three instruments. The most prominent cluster of variables contained the Q Sort, IClDS dimension, and the I, OK-NG dimension. After the Self-Disclosure Scale, this cluster contributes much of the remaining total covariance. Their common variance might be due to a "self-enhancement" factor. Future studies might aim at better identifying this factor, and the "communica- tion" factor, more directly. A problem would seem to be that the "Hello- Goodbye" effect could easily cause Spuriously high self-ratings, eSpecially during post-testing. An example is the already referred to Q Sort scores of the individual therapy dropouts. Among,"others" ratingS - The importance of the Self-Disclosure Scale is enhanced by the fact that it also provides the greatest portion of the total covariance among the "others" ratings. The IClDS is the second most 35 salient measure, with the other three providing approximately equivalent amounts of the remaining total. Between the self and "others" ratingS - Three of the four pre-test measures rated by both SS and "others" do not correlate significantly at the .05 level. Only the "other" reported IClDs and IClLH, the two dimen- sions which combine to form one measure, the ICl, correlated Significantly with self-reported scores on these same variables. These results are similar to those found in a number of studies in the literature. Rogers and Dymond (1954) found friend ratings to be unrelated to clients' self- ratings. Fairweather, et a1. (1960) discovered two clusters of inter- correlated measures, Subjective self-evaluations and objective interper- sonal evaluations. The latter contained therapists' ratings. This result emphasizes the importance of careful use of outcome measures in psycho- therapy research. Increased familiarity with the Specific value and meaning of various outcome measures is mandatory. Representativeness and adequacy of measures Self-Disclosure Scale - Although the self-rating of the Self-Disclo- sure Scale has gained major focus in this study, there is some question about the meaningfulness of these scores. Forty-six of the eighty-eight ratings are of scores of seven or eight. It seems highly improbably that this is an accurate characterization of these Ss. The self-ratings on the Openness (O) and Data-Seeking (DS) Scales, two measures which corre- late significantly (p <.05) with the Self-Disclosure Scale, support this hypothesis. Well over half of the O and DS ratings (106 out of 176) were below seven on a nine-point scale. Therefore, there seems to be something special about the Self-Disclosure Scale which produced such high ratings. 36 The Special characteristic might be the detailed definitions of each SD scale score point (See Appendix B,p. 89-90). All scores below seven contain some negative descriptive phrase. For example, "rarely reveals own personal feelings" is part of the description of point six. These negative elements may elicit defensiveness in the self-raters, and fre- quent ratings of seven. However, these definitions have not prevented self-ratings lower than seven in another, more sophisticated sample. Ratings lower than seven predominated in the study by Hurley (1967), which used members of a graduate course in group counseling methods as SS. 9 Sort - The Q Sort scores seem somewhat more representative than the Self-DisclOSure Scale scores. The overall pre-test mean of .33 is somewhat higher than the pre-test means, ranging from .16 to .32, reported by Ends and Page, as cited by Truax, Schuldt, and Wargo (1968, p. 53), in their study of group psychotherapy with male alcoholics. However, these hospitalized alcoholics seem like a much more disturbed sample than the present SS, and even lower pre-test scores would be expected. Also, the pre-treatment mean of the present group therapy 83 was almost identical to the pre-treatment mean of all of the alcoholics entering group psycho- therapy in the Ends and Page Study. However, the adequacy of the Q Sort as an outcome measure in this Study is questionable. Perhaps most pertinent are in the high Q Sort again scores of the four persons attending only one individual treatment Siession. No significant positive change appeared on the Q Sort for the Egroup or remnant individual treatment groups. This can be eXplained, as 111 Truax, Schuldt, and Wargo (1968), merely by assuming that the therapies 37 did not produce significant positive change. The absence of change on almost all of the other outcome measures Supports this assumption. This fails, however, to account for the significant positive change which the reconstituted non-therapy group produced on this measure. It has been suggested that the four non-therapy SS who dropped out of individual psy- chotherapy wanted to portray themselves as having gained. Two additional non-therapy clients produced Q Sort scores so divergent that it was dif- ficult to accept their credibility. The self-ideal correlation of one S changed 1.342 from pre- to post-testing (from -.445 to +.897), and another had a post-test self-ideal correlation of +.990. This latter S had also exhibited a good deal of negativistic behavior in the pre- testing session. It seems that these two non-therapy SS wanted to demon- strate that they had changed, either to themselves, to the S, or to both. These data seem to converge with the Truax, Schuldt, and Wargo (1968) findings of a high correlation between high self-ideal correlations and high scores on the Edwards Social Desirability Scale. The Q Sort appears to have produced a number of "false positives" in this study. More im- portant, the instrument may be very susceptible to this form of distortion, eSpecially when clients want to Show that they have changed. Ibelle, as cited by Berry and Miskimins (1969, p. 103), reported that this type of self-concept measure often cannot distinguish schizophrenics from normals. These convergent findings raise questions about the usefulness of the Q Sort as a therapy outcome measure. Frank's (1956) findings seem to offer contradictory evidence. Using ten SS, he found test-retest reliabilities on the Q Sort to range from .93 to .97. However, there is no evidence that his SS were attempting to 38 portray themselves as having changed in the second testing. In fact, they may have been trying to be consistent with their initial results. If they were, this further supports the present evidence that the Q Sort is susceptible to the needs of the Ss. Argyris Interpersonal Competence Scale - The Argyris Scale commanded a major focus in this study. Inspection of the transcript score sheets clarifies one important problem. Of the 2829 scored units, 2820 were scored as ideas; only nine were scored as feelings. This greatly re- stricted the scores, and, in fact, caused this overall sample to be portrayed as relatively incompetent, interpersonally. Data to assess the representativeness of these scores come from two studies. Argyris (1965a) ran sensitivity groups with business execu- tives, with pre- and post-testing. The SS in the present study were comparable to his "low learners" groups, who produced almost no feeling scores.* Hofman (1969) compared "healthy" and clinic married couples. Their interpersonal competence score means were slightly lower, on a corrected (Hofman used a base score of 1, the present Study a base score of 16) scale, than the pre-test means in the present study; the overall mean for his Study was .40, compared to the present overall pre-test mean of .69. However, the most important piece of data again is that his groups produced more feeling scores, 114 out of 4235, as compared to 9 out of 2829, than the SS in the present Study. It might be asserted that the problem was not with the Argyris mea- sure, but rather that this sample was an interpersonally incompetent one. *Exact percentages were not available for the Argyris study. 39 However, it seems highly unlikely that not one.S would produce more than two or three feeling scores, eSpecially in the light of high scores on other measures, most notably therapists' ratings. A lack of ego involvement in the problem-solving tasks employed in the present research seem a more parsimonious explanation of this lack of feeling scores. Evidence for this hypothesis appears if we assume that the ego involvement has two potential sources: the problem itself might have personal relevance, or the individual might be interpersonally involved with other members of the group. In this Study, the problems had little or no personal relevance, and the members of the group were meeting for the first time. The importance of these factors did not become apparent until after the post-testing, when all the tape trans- cripts were scored. Occasional observations of the sessions by the S revealed intellectual involvement; that is, the SS wanted to solve the problems correctly. However, personal feelings did not seem to be a part of the interaction. In the Hofman study (1969), married couples solved similar artificial problems. The results contained many more feeling scores than the present research. Again, the problems had little personal relevance, but the Ss did have a great deal of interpersonal involvement. Finally, in the study reported by Argyris (1965a), members of a business organization participated in sensitivity groups. The problems discussed had a great deal of personal relevance, and SS were also involved with each other. The results again contained an abundance of feeling scores. Thus, it appears that feeling scores, and potentially significant results, might arise from ego involvement on the part of the Ss. Furthermore, the involvement may be related to either the problem or the problem-solving grilriiw. J. . q L. 1.. 40 group. Future research might use tapes from group and individual therapy sessions as their raw data. This would provide the necessary involvement, although the Special nature of a dyadic individual therapy situation might present new complications. Another possibility would be forming the problem-solving groups out of the SS and two or three friends of their choosing. This would be eSpecially helpful if they focused on one of their mutual real-life problems, and would have the additional advantage of providing for the inclusion of non-therapy SS in the sample. However, even if this problem were solved, other deficiencies in the Argyris Scale seem to remain. Its one obvious strength, the high scorer reliability, is offset by the lack of stability it showed in test-retest scores. This finding,possibly attributable to the lack of ego involve— ment, suggests that the Argyris Scale may be highly susceptible to situational factors. Finally, scores on the Argyris measure tended to correlate nega- tively (r = -.375, p <.06) with therapists' ratings of Ss improvement in addition to not correlating positively with any of the other pre-test measures. Since a number of these variables have shown at least some usefulness in past research, this points to another major limitation of the Argyris Scale as used in the present study. Taken together, these problems underline the limitations of this instrument as an outcome measure in this study. It remains to be deter— mined if this measure can be usefully applied to other psychotherapy outcome research. Other measures - Since most of the other instruments have not been used in research before, it is difficult to assess their adequacy at this . .. IQ 41 time. One favorable datum involved the 101. The self and "others" ratings of both the IClDS and IClLH dimensions have significant positive intercorrelations (p <1.05). This may be an indication that the 101 is somewhat less Susceptible to false positives, eSpecially those resulting from the "Hello-Goodbye" effect. This is a distinct potential advantage for psychotherapy outcome research. Overall Evaluation These numerous problems serve to underline the difficulties with psychotherapy outcome measures mentioned early in this paper. However, the dropout problems evidenced in the remnant individual therapy group and the reconstituted non-therapy group prevent the meaSures, eSpecially the Argyris Scale and the Q Sort, from being evaluated completely. These interrelated deficiencies greatly hinder, if not completely eliminate, the possibility of making any definitive statements about even the slight differences found between the three treatment groups. Therefore, results already considered questionable must be inSpected even more tentatively. General Findipgs The possibility of making any clear statements about the hypothe- sized differences among the three treatment groups was eliminated by this combination of "dropout" and measurement compkaxities. Morever, of the 42 mean differences identified between the treatment groups in Table 3, only two - both involving the dubious self-ratings on self-rated Self- Disclosure, were statistically significant. Of the 33 differences among these same treatment groups in "proportions gaining", as listed in Table 4, only one (101) statistically significant difference occurred. These findings that only 3 of 75 differences achieved the .05 level of 42 significance is plainly below the chance level. However, it seems useful to attempt to explain the results that were obtained, if only to stimulate further research thought in this area of study. It must be understood, of course, that these explanations have a weak empirical base. The three treatment categories will be discussed simultaneously when , attempting to eXplain the results. In this way, hypotheses 1, 2, 3, and 4, and the eXploratory question can be attended to at the same time. The remnant individual therapy group, although inconsistently so, evidenced the most positive change. It produced a significantly higher mean change than either of the other two treatment groups on the Self-Disclosure Scale, self-rating (p<3.05), and it showed a significantly greater (p < .05) positive proportion change on the 101, others rating, than the reconsti- tuted no—treatment group. Finally, an average of 61% of the individual group remnant showed positive change on each of the self-rating measures. The group therapy and reconstituted no-treatment categories evidenced an essentially equivalent lack of change, with the group therapy sample show- ing a slight, non-significant tendency toward negative change. An average of only 41% of these SS evidenced positive change on the self-rating measures. There are a number of alternative ways of eXplaining this result. The first, obvious, eXplanation is that in the present case, neither group nor individual therapy was effective, with group therapy being even more ineffective. This is feasible, but seems to be an overambitious inter- pretation of the findings. A common interpretation of equivocal outcome results, initially cited by Bergin (1963), is the presence of antithetical processes in the therapy 43 experimental treatment groups. Some therapists produce negative change in some clients, which cancels out the positive change that they or other therapists may stimulate in other clients. An attempt to test this is portrayed in Table 7. Clients rated as changing more than most clients by their therapists were compared with clients who were rated as chang- ing less than most clients (average ratings were omitted). The overall change score was used for comparison purposes. The clients rated low change by their therapists Showed a tendency (p > .10) to score low on the overall change score. Therefore, there is some support for the existence of Bergin's canceling out effect. TABLE 7 Relationship Between Therapist Ratings and Overall Outcome Scores Therapist Rating Overall Score Low High Totals Low 5.5 5 10.5 High 2.5 8 10.5 Totals 8 13 21 2 _ X — 1.81 ‘2 > .10 The inability of both the individual and group therapy treatments to produce significant outcome measure change may also have been a result of the relative therapist inexperience. Most were psychology interns with three years of part-time therapy eXperience. This could also eXplain the difference between the remnant individual and group categories, since all of the therapists had had most of their previous experience working with individual clients. Most psychotherapy training programs emphasize 44 individual therapy eSpecially in the early stages of the program. There- fore, it seems that research using relatively ineXperienced therapists as their own controls may be an inappropriate design. Only therapists with approximately equal amounts of eXperience in group and individual therapy would be appropriate for such research designs. Another interpretation of the results is that the period of psycho- therapy was too short to produce Significant positive change. A number of studies have cited this explanation, among them the Study by Satz and Baroff (1962) referred to in the review of the Q Sort literature. This interpretation also can offer an explanation for the "differences” be- tween the remnant individual and group therapy groups. DeSpite the "economy" studies of Baehr (1954), Novick (1965), and Suinn (1968), a number of authors feel that group therapy takes longer to produce posi- tive changes. Sundberg and Tyler (1962) cite an article by End and Page which Showed that thirty group sessions in Six weeks produced signifi- cantly more change than fifteen group sessions in six weeks. They refer to End and Page's conclusion that the group therapy clients were just beginning their period of greatest change after fifteen sessions. This, of course, coincides with the maximum number of sessions used in this Study. On the Surface, it might appear that group clients have less therapy time than individual clients Since sessions have to be divided up six to eight ways. However, the focus of the groups in this study often was an interaction between two or more group members, and the techniques used often involved up to all of the group members simultaneously. In the group for which E served as co-therapist, non-verbal techniques 45 accomplished this. For example, in one session the clients used the therapy room as their interpersonal universe, and placed themselves as distant from each other as they felt. They then discussed the eXperience. It remains true that less "therapist time" was invested in the group treatment, on the average, than went into the individual treatment. From that perSpective, the failure of the individual group remnant to exceed the group treatment in gains can be viewed as supporting the view that group treatment is as efficacious, and less eXpensive, than individual therapy. ' can only be settled It seems that this controversy about "economy' by researchers becoming more familiar with the processes of group and in- dividual therapy. This author would agree with End and Page, and would hypothesize that the dormant fifteen sessions actually consists of each group client receiving a great deal of feedback about his interpersonal defenses and postures. This might well cause an initial negative change, at least in self-evaluations; the finding that less than 50% of the group Ss changed positively on the self-rated measures hints at such a process. Once this feedback began to be used apprOpriately, change might occur quickly, because a larger part of the client's environment (7 or 8 rela- tionships) would already be involved in the change. Anecdotal evidence partially supports this hypothesis. Upon consul- tation with the other therapists, the S found the experience with his group to be fairly typical. The group continued for a total of twenty sessions, until summer vacation caused termination. At least five of the six members had started to change, and another ten to twenty sessions would have provided the eXperiences necessary to reinforce their gains. 46 It is further hypothesized that the processes underlying group and individual therapy process are quite dissimilar. The individual client, as a result of a more intense, consistent impact with the therapist, exhibits more facile initial change. However, the major hurdle for the individual client would be transferring this change to the outside world. A large number of new and old relationships must be restructured. These processes are not considered to be all or none. Inidivdual clients cer- tainly work on other relationships during therapy, and group clients may certainly have one EQEE meaningful interaction, either in or out of the group, with the concomitant changes in self-perception. Again, anecdotal evidence is relevant. A number of the individual clients had successfully terminated at the time of the post-testing. None of the group clients had. Some individual clients had apparently exhibited their initial change. The results also provide some evidence that these processes were occurring in the therapy clients in this study. A greater percentage of remnant individual clients than of group clients showed positive change on the "self-rating" meaSures. Two interpretations of this reSult seem logical. Either the individual clients have not yet had a chance to change in relation to others besides their therapist, or Significant others have not had a chance to recognize the change. These need not be contradictory, and both fit in the hypothesized conception of individual therapy process. Another convergent finding is the significant increase in the number of remnant individual clients who reached the eXpert's criterion on the Self-Disclosure Scale, self-rating, after therapy. This is likely to be a result of a "healthy" amount of self-disclosure with the therapist. In 47 the light of previous comments, this can be interpreted as a first necessary step toward overall change, as well as a change limited to one relationship. It was hoped that the Argyris Interpersonal Competence Scale would help portray these hypothesized process differences, especially in con- trasting between "in relationship" changes and "self-perception" changes, such as on the Q Sort. Unfortunately, the limitations of the design and both measures prevented this. Two of the factors mentioned as potential sources of the negative results, therapist eXperience and length of therapy, may be interactive or additive processes. A reasonable hypothesis is that as therapist eXperience increases, length of time needed for Successful therapy de- creases. Unfortunately, the results do not Speak to this issue. The Argyris Scale Hypothesis 5 was Inartially supported when the Argyris Scale Showed high interscorer reliability, although even this 82% agreement was some- what less than that reported by Argyris (1965). When the two raters in this Study inSpected their disagreements, approximately 80% of these seemed attributable to a miscommunication of two minor scoring rules. After correcting these errors, and after discussion of the few remaining disagreements, the raters agreed on all of the items. This would seem to indicate that future scorings by these raters would be able to match the percent agreement score reported by Argyris. This, moreover, can appar- ently be accomplished with only typed transcripts, rather than tape recordings as in the Argyris study, thus greatly speeding up the process. Argyris "gain" scores tended to correlate negatively (r = -.375, 48 tw0dtailed p <1.06) with therapists' ratings of client gain. This result undermines the appropriateness of using the Argyris as an indication of the comparable effectiveness of group versus individual therapy. EXplana- tions for this failure have been offered. Dimensions Chosen by Ss The task of remembering all three dimensions on which SS wanted to change at the pre-testing period proved to be too difficult for thirty- five of the thirty-six‘Ss at the time of post-testing. This might be expected of the non-therapy‘Ss since they probably did not come to the Counseling Center to change these sorts of personal dimensions. But, while it is true that the non-therapy SS had the highest forgetting per- centage, the forgetting rate of the therapy SS is too high to attribute this phenomon to a lack of motivation for change. An alternative eXplana- tion is that the dimensions were unimportant to the Ss, that is, not worth remembering. This interpretation brings up a question about the treatment of Symp- toms by simplistic behavioristic methods. What is the advantage of attack- ing a symptom presented by a client as needy of change, when it may not be important enough to remember if left alone? A more feasible alternative would seem to be a reSponse to the client's need to change, with whatever therapeutic tehcniques seem appropriate, rather than a response to the specific symptom that he identifies as the proposed focus of change. The second result, the remembering of dimensions that changed in a positive direction, suggests an alternative explanation of this forgetting phenomenon. It is simply that SS will remember dimensions on which they have changed positively. However, this second result might also have 49 occurred because SS change positively on self-rated dimensions of their own choosing regardless of whether they can remember them. This might be another example of the "Hello-Goodbye" effect. The following procedure would explicate why remembered dimensions exhibit positive change. In the post-testing, the SS would be asked to remember and rate themselves on their own dimensions. Then, they would be informed of the dimensions they forgot, and asked to rate themselves on these. The difference between the percentage of positive changes in the two groups, remembered and forgotten, would be the crucial statistic. Future Research Because of the diverse problems encountered in this study, an attempt to replicate it by using the same basic design seems inappropriate. There- fore, this section will suggest a number of alternatives, each designed to study one of the Specific difficulties that affected this research. The most serious was the "dropout" problem, relating to the therapy and the research. The meaning of premature termination might better be understood if clients who dropped out were compared with clients who re- mained in therapy. A pertinent finding in this study was the changes in Q Sort scores of the four individual therapy dropouts. The hypothesis that this change was an example of the Hathaway "Goodbye” effect could be checked by asking premature terminators if therapy had changed them. The hypothesis would be that "yes" answers would be associated with high positive change scores. An underlying assumption would be that one session does not produce genuine therapeutic change. A third group to add to this comparison would be clients who wished to remain in therapy, but didn't, for reasons such as therapist absence. This control would help separate 50 the effects of motivation or eXpectation from actual therapy effects. The research dropout problem must be solved through the use of better research design and control. The extrinsic motivation of being paid $3.00 for the post-testing did not prevent dropouts in this study. Neither did the research task, which the SS reported to be interesting. The most appropriate step might be an attempt to obtain a strong commit- ment from the SS before they participate in the pre-testing. Thus might restrict the sample to well-motivated SS, but this probably is already true of this type of research. A second major problem concerns the outcome measures. It appears that even an instrument Such as the Q Sort, which has been the focus of much research, needs a great deal of further study and evaluation before its validity can be assumed with any confidence. This was supported by Ibelle as cited by Berry and Miskimins (1969, p. 103),who demonstrated that the Q Sort often cannot differentiate schizophrenics from normals. In general, this evaluation needs to be done apart from the complexities of research comparing group and individual psychotherapy. The seeming susceptibility of the Q Sort to false positives and to the "Hello-Goodbye" effect demands further study. When comparing pre- mature terminators with clients who remain in therapy, their Q Sort scores should be inspected closely. Another tactic would be to administer the Q Sort pre- and post-test, and use two different §.°r S eXpectancies (or two different sets of instructions) as the eXperimental variables. The first S could eXpect change, as in a psychotherapy study, and the second could exPect stability, as in a reliability study. The extent that the results fulfilled these expectations would reflect the susceptibility of 51 the Q Sort to Ss' needs to portray themselves in certain ways. Another method to inSpect this problem would be a review of the Q Sort literature, with Special attention being paid to factors which might have elicited "false positives." Future research with the Argyris Scale has already been suggested. Tasks to which this scale can effectively be applied must be devised or found; aritifflfial problem solving may be inapprOpriate. The positive results reported by Argyris (1965) may only be replicable within highly engaged sensitivity or psychotherapeutic groups. The pre-treatment mean differences might best be handled through the use of apprOpriate statistical methods, such as an analysis of covariance. Another possibility would be to obtain larger pre-test samples, and then form treatment groups by matching pre-test scores. This matching could, of course, be a complex task when a number of outcome measures were used. A number of lesser difficulties came to light in this Study, and Steps must be taken to control their effects in future studies comparing group and individual psychotherapy. One variable needing control is client motivation. This study used control SS who had come to the Coun- seling Center desiring some change, even though it was not of a psycho- therapeutic nature. An "own control" design seems to be a better approach toward assuring client comparability in this type of setting. This pre— sents a conflict between research and service, in that Ss might have to wait through a relatively long control period before receiving help. This conflict might be overcome by using students who come to the Counseling Center just before summer vacation as SS. These students are not usually seen until fall, and thus, a three-month control period is automatic. 52 Unfortunately, this might be an atypical sample, made up of peOple who could ask for help only at a time when they were unlikely to receive it. A separate study might be needed to determine the representativeness of such a sample. They would have to be compared with clients who sought therapy at other times of the year. It is unclear at this time which measures could best make this comparison. The effect of level of therapist experience must also be considered, especially when eXperience in individual and group work differs greatly. Number of years of experience and number of sessions per year will be helpful as objective measures of this variable. However, less easily quantifiable asPects of eXperience, Such as variety of experience, clien- tele, and techniques will also have to be taken into account. Finally, the interactive or additive effects of therapist eXperience and length of therapy will have to be assessed. This would be done most easily with a therapist eXperience by length of therapy analysis of variance design. Meaningful outcome research may require a more adequate consideration of the divergent processes involved in individual and group psychotherapy. One way of eXploring this area would be repeated testings of the SS through- out the therapy experience. The use of a number of outcome measures would be useful, as they might be alternated to decrease the problems of repeated testing with the same measure. The Self-Disclosure Scale might be appro- priate for this task, as it did differentiate between the treatment groups in this study. Alternatives would be measures which intercorrelated highly with the Self-Disclosure Scale; possibilities would be the Openness and Data-Seeking Scales used in this study. However, it is obvious that the use of these measures depends on their validation in further research. 53 In addition, process testing would have to continue for some time after treatment was terminated. This testing would help assess the validity of the hypotheses made earlier in this paper about individual and group therapy. REFERENCES Argyris, C. Explorations in interpersonal competence-1. Journal of Applied Behavioral Science, 1965, 1, 58-83. Argyris, C. EXplorations in interpersonal competence-11. Journal of Applied Behavioral Science, 1965, 3, 255-269. Argyris, C. Organization and behavior. Homewood, Illinois, the Dorsey Press, 1965. Armstrong, R. G. The Leary interpersonal checklist: A reliability study. Journal of Clinical Psychology, 1958, 14, 393-394. Baehr, G. The comparative effectiveness of individual psychotherapy, group psychotherapy, and a combination of these methods. Journal of Consulting Psychology, 1954, 18, 179. Bergin, A. E. The effect of psychotherapy: Negative results revisited. Journal of Counseling Psychology, 1963, 10, 244-250. Berne, E. Principles of group treatment. New York, Oxford University Press, 1966. Berry, K. L., and Miskimins, R. W. Concept of self and posthOSpital vocational adjustment. Journal of ConSultingpand Clinical Psychology, 1969, 33, 103-108. Briar, S., & Bieri, J. A factor analytic and trait inference study of the Leary interpersonal checklist. Journal of Clinical Psychol- ogy, 1963, 19(2), 193-198. Butler, J. M. Self-ideal congruence in psychotherapy. Psychotherapy: theopy, research and ppactice, 1968, 5(1), 13-17. Ends, E. J., & Page, C. W. A study of three types of group psychotherapy with hoSpitalized male inebriates. Qparterly Journal of Studies on Alcohol, 1957, 18, 263-277. Cited by C. B. Truax, W. J. Schuldt, & D. G. Wargo. Self-ideal concept congruence and im- provement in group psychotherapy. Journal of Consulting and Clinical Psychology, 1968, 32, 47-53. Fairweather, G. W., Simon, R., Gerhard, M. E., Weingarten, E., Holland, J. L., Sanders, H., Stone, G. B., & Reahl, J. E. Relative effectiveness of psychotherapeutic programs: A multicriteria comparison of four programs for three different patient groups. Psychologigal Monogrpphs, 1960, 74, N or 8 (Whole No. 492). Frank, G. H. Note on the reliability of Q Sort data. Psychological Reports, 1956, 2, 182. 54 55 Freud, S. "The Case of Shreber" in The complete_psychological works of Siggund Freud, Vol. 12, 12-82, London, The Hogarth Press, 1958. Hartly, E, & Rosenbaum, M. Cirteria used by group psychotherapists for judging improvement in patients. International Journal of Group Psychotherpp , 1963, 13, 80-83. Hofman, K. Personal communication, 1969. Hurley, S. Self-disclosure in counseling groups as influenced by struc- tured confrontation and interpersonal process recall. Unpublished doctoral dissertaion, Michigan State University, 1967. Ibelle, B. P. Discrepancies between self-concepts and ideal self-concepts in paranoid schiZOphrenics and normals. (Doctoral dissertation, University of Connecticut), Ann Arbor, Michigan: University Micro- filmS, 1960. No. 60-5237. Cited by Berry, K. L., and Miskimins, R. W. Concept of self and posthOSpital vocational adjustment. Journal of Consulting and Clinical Psychology, 1969, 33, 103-108. Jourard, S. M., & Lasakow, P. Some factors in self-disclosure. Journal of Abnormal and Social Psycholpgy, 1958, 56, 91-98. Kubie, L. Individual and group psychotherapy. International Journal of Group Psychotherapy, 1958, 8, 20. La Forge, R. & Suczek, R. F. The interpersonal dimension of personality: III. An interpersonal checklist. Journal of Personality, 1955, 24, 94-112. Leary, T. Interpersonal diagnosis of personalipy, New York, The Ronald Press Co., 1957. Nahinsky, I. D. Q Sort approaches in Studying sex, socio-economic status, and psychopathological variables. Psychological Reports, 1963, 16, 55-64. Novick, J. I. Comparison between Short-term group and individual psycho- therapy in effecting change in non-desirable behavior in children. International Journal of Groungsychotherapy, 1965, 15, 363-366. Peatman, J. G. Introduction to applied statistics. New York, Harper & Row, 1963. Phillips, E., Raiford, A. & El-Batrawi, S. The Q Sort reevaluated. Journal of Consulting Psychology, 1965, 29, 422-425. Rogers, C. Psychotherapy and personality change. Chicago, University of Chicago Press, 1954. 56 Satz, P., & Baroff, A. Changes in the relation between self-concepts and ideal-concepts of psychotics consequent upon therapy. Journal of General Psyphology, 1962, 67, 291. Schutz, W. Joy. New York, Grove Press, 1967. Shea, J. E. Differentials in resistance reaction in individual and group psychotherapy. International Journal of Group Psycho- therapy, 1954, 4, 253. Suinn, R. M. The desentization of test-anxiety by group and individual treatment. Behavior Research and Therapy, 1968, 6(3), 385-387. Sundberg, N. E. & Tyler, L. E. Clinical Psychology. New York, Appleton- Century-Crofts, 1962. Truax, C. B. Effective ingredients in psychotherapy. An approach to un- raveling the patient-therapist interaction. Journal of Counseling Psychology, 1963, 10, 256-263. Truax, C. B. & Carkhuff, R. R. EXperimental manipulation of therapeutic conditions. Journal of Consultipg Psychology, 1965, 29, 119-124. Truax, C. B., Schuldt, W. J., & Wargo, D. G. Self-ideal concept congru- ence and improvement in group psychotherapy. Journal of Consult- ing and Clinical Psychology, 1968, 32, 47-53. APPENDIX A ADDITIONAL TABLES 58 muosuo noon Eoum ouoom mwmuo>< Amv posuo qu8 Boom ouoom sz nonuo mausow Souw ouoom Amy unwamuonu wmmum uofiaom mo pouooua a m.o m.¢ w o.~ «.0 a n m n w m n.¢ a.H - ooh. 0H.o z mswozm as A5 A8 93 o: m m a m.~ ¢.n n m w a m o.m u c.4H- mug. mm.o z «owned Any Any Ame Amv Amy m m o o.m m.m - n m n m m m.w o.m mum. NH.H m SNHem>m so as as as as m.o n m N.m w.~ n q n m m «.5H ¢.o 1 «mm. ao.o m madman A5 A3 A3 as o: a e a H.mHu n.H n o a m o m.aat m.mH mwm. oo.Hu m oomoma Ame A00 Ame Asv Ass 5 n n m.o u n.H n m n m o m.m n.a - mHH. oo.H z mamooh as as as so as m m o m.na o.~H- m N N n m n.mH n.m - mmm.n Hm.o z ammmmH as as A5 A8 AB 5 m.n m.m n.~H o.~ e m m o o ~.mH m.no- «no. mm.H m sumumhm ISVLBILBILS E 0.5 0 _O T. T. On: 0 no no T. T. T. .O "V .1.0 s .d as no 1:3 9 .a 3 - no no 1 8 T.. 3 a T. .l 8 T. 3 a u- .l .l S E .o.s s w n1 n. .01: e u a no mm a. o 1.. . n” as 1.. . u 1 w“ co 1 J O S 3 O S 9 c . .4 1.. 8 3 8 S 3 3 0 N 8 n 9 8 n 8 8 X 5 J N. J M. . a m. a m. m .80 306300 00 00 uuosuo “Hum Hucuouxu consume: ozone unoaumousa ammuoaa nsouo mo mowoom uooauoum w mdmdfi 59 .w: o o o m a m.¢ m.mHu HwN.u NN.o m mmNom< a d N w m ¢.H u 0.0N1 HNm.n om.o m NNommm a a m m m m.oH m.ON- on. ¢¢.o m mowNmm o m N m.w N.¢H1 a N n a N ¢.m o.m~1 NON. NH.H m Nmmomw so as so so as N N o N.m u m.oHn N N N d o ¢.N o.oH1 mag.- w¢.H z oHHm>< Ame Ame Ame A00 Ame m m N m.mN H.0Hu m w w o a m.m m.H mNH. a~.o m ooNQz< as as o: 9.0 9v m.o m m.o m.m N.m 1 a a a m m 0.0 ¢.m u HNe. so.H z mNmQZS Amv Amv ago “no. A00 N m N m.o~ m.m o N o o N v.0 o.¢ oom. 0N.o m quNma as as as as as o m m m.o H.o N N m o o n.0N N.m u can. NH.H m NwNmzm .Iuwmwl. Ame A00 Amy Ame 1 m. m «U m m. m m. m .m m .1 m m .o m.- 8 1.. 3 a T. T. 8 T. 3 a u. T. .l S “v.3 m m mm m“ «.1: s m a n. n. no o MM .1 . . J “A H S J J O S a O S 3 ¢ . .4 T.- 8 a 8 S a S O N 3 n a s n a s u. no 1 H. J M- . a m. a M. Na 08 muoofl s 00 00 m o m .uuonuo maom Hmdhouxm mundane: I“ I" 3600383 0 33s 60 muosoo Loop Eoum ouoom owmuo>< Amv wonuo onE Eouw ouoom sz posuo onEom Eouw opoom Amv umwmmuoau mmmum sowsom No woumoue u- .ouozs N o N w. - N.N o o N o m N.NN N.o was. No.0 m ommlo as as as as as N m w H.o u N.N a m m N m 0.0 N.NH- mom. «0.0 m Nmmqu Am. Ame Nev Ame Ame m N m ¢.¢ N.N e N N m m m.N - m.N - NNm. o¢.o m HmoNNocH mo mouoom once-mum m mqm< Amv uoSuo mama Eoum ouoom sz nonuo onEoN Eoum ouoom Amv .oHdEmm ucoeumouu Hmsvw>fipcfi ecu cw NHHchwNuo mums ow omega .1 .mqu¥ N N 0 0.0 0.0 0 0 m w 0 0.0 H.0 - maq.- mN.0 m Nmthm 00 00 0.0 00 00 m m N NH 0.0 1 0 N N N N N.NH N.N - mom. NH.H 2 0000mm A00 A00 A00 A00 A00 N N 0 N.0H m.N m 0 0 0 N 0.0 0.0 H00. 00.0 m eqNomZ A00 A00 A00 A00 Amv 0.0 m m H.0H 0.H - N N w 0 N N.N 0.H mam. 00.0 h *mNmonh A00 A00 A00 A00 A00 0 N N 0.0 H.m 0 N N m 0 N.N 0.0H- mac. 0m.0 z owaqzm A00 A00 A00 A00 .00 N 0 0 0.0 1 n.0H N 0 N m 0 m.H N.N Haw. Hm.0 m 0H000m A00 A00 A00 A00 Amv 0 c m 0.0 N.N u N m N m N H.m - 0.0 00m. 0N.0 m NNNonm A00 A00 A00 A00 A00 0 m w N.0N1 0H 0 0 N m N 0.m u 0.N - HNO. 00.0 m mHHmZH 00 00 00 00 00 N N m N.N - m.HH N m 0 m N m.N u 0.0 1 H00. 00.H z mNH000 sz A20 sz A20 ‘ .20 ma .0 .0 m. m m... m .0 m .- m. m o 0 S .l 3. a .l .l S .l 3 a u- .l I S o I. B u N1 G o I. B u a 0 NI. 0 0 m m: a. w H s o1. 0. w .5 v... H s u. u. m. .0. .0. m. m N m w s a m” m m” w xo n 0w 0uo O 0 300.030 muonuo «How Hmapouxm monummoz “VHF” msowu ucosumoua NamuoSH-aoz mo mouoom umoH-oum 0H mam¢ Amy nozuo onEom Eoum ouoom Amv uoSuo mama Boum ouoom sz downspono mmmum uoflcom No woumoua a N N 0.0 0 0.N 1 0.0 u N 0 N 0 0 N.¢ 0.0 1 men. N0.H z merZm 00 E 90 00 00 m 0 0 0.0 m. u N.N u N e w 0 0 N.m - H.m - 0N0. 00.0 2 NONNQM :0 A3 00 00 20 a 0 m w 0.0 0.0 1 0 e m N 0 0.0 0.N H00. 0m.H m SNH¢m>m 00 00 00 00 00 m 0 0 0 0.NH N.N u 0 m m 0 0 0.0 0.0 000. 00.0 m 0N¢000 00 E :0 00 00 a N 0 0 N.NHu N.0H 0 m 0 m 0 m.HNu H.0H HNm. mm.0 m 0000mh £0 E 00 00 0.0 a N N 0 0.N u 0.0 u N m m m a 0.N 0.0Hn 000.- 00.0 x mamooh 00 0: 00 00 00 H a 0 m w.m - 0.NH1 a H N 0 m 0.N 0.0H- 000.1 00.H z 000mmH 0.0 E 00 00 0.0 N m m m m.m 0.N a 0 0 m N 0 w.mH N.m u NNO. 0N.0 m *NmNmmm A00 A0 A00 A00 A00 aumamm m 0.3.0.004... m o m .43 SI .4 a T. T. 8.1 .4 a u- T. T. S unpaswm mpqswomm m m m .3.0 o S a o e. a c . 3 1. m. m. w M” m .4. N m m s .4 J H. 1 X. . o m. o m. m .80 Soon 000 00 00 muosuo maom Hmcuouxm monomwoz moose ucosumoua anemone dsouu mo mowoom umoauumom NH mqm< E 0.0 00 A0 S0 0 N 0 0 0.00. 0.0Hu 0 w m m N 0.0 0.N 000. 00.0 m 00N02< 00 00 0.0 00 00 0 0.0 0 0.0 m.0 0.0 u q 0 m 0 0 0.N 1 0.0 u 0N0. nus- : mamozz A00 A00 A00 A00 A00 N N 0 m m.0 0.0 N m N 0 m N.NH 0.N - mNN. NN.0 m «Nona 00 00 00 00 00 0 :11 N 0 0.mH m.m N w m 0 N HN 0.N - MNm. 00.0 m Nquzm Ame A00 A00 A00 .111- mm. m a. m .m m m m w m. m. m w. m m. ..o m- .43 S T. .4 a T. .l S T. .4 a U- T. T. S B 1;. o 1: e u .u .o.s s u a no n. o u 8 .l . . u m S T. . . u 1 X m S 1 1 00d 0 S a O S a s . 14 T.- .1 s .0 s s a s n. n. s s n a s n a s “x no 14 J x. 1 x. . a m. a m. N we muoom 05m 00 00 m . muonuo maom Hmawouxm mandamus 36030003 a . .505 66 mumnuo Luon Eouw muoom mwmum>< Amy umnuo mama Eoum muoom sz umnuo mHmEmw Scum muoom Amv umwawumsu wmmum uofiaom an kummpe * m u u . nu- nun N N m c m w.mHu N.N mNm. 00.0 m *HwNmmm H m m ¢ uuu nun o o N m N N.NH N.N «Nm. om.o m ommlo as as as to g m N o m m.m H.m N m m m o 0.N ¢.m u o¢o. Nm.o m NmmHZA E Q Q E g u o q ¢ N.N m.m N q o N N w.¢ u w.¢ u Noe. u--- m HmoNavcH mo mmuoum ummHuumom NH mama. TABLE 12 Cont inued) SUTJEH nsgdezaqm ' '¢ BJUSOIDSIQ . . 'JIBS Buxnaas-eaaa ' ' U) 3 ssauuado | ' .c U 0 I I H1 ' ' IOI ' ' I l SCI 1 : IDI SIDSOIOSIQ 419$ °° " m 8u1aaas-enea r. a) u a ssauuado .\ g. :33 DN-XO‘Jaqlo xo tn on-xo ‘I N q u.‘ 3 N H m H'T m’ N“ 101 '4 . SO r» ax 101 N °‘ l | Ne «a 3.103 b '3 g H g. :2 i 3 91158.“; 0' : x n: x (9), III-o In 0) U m a g 3 cm a. 67 68 muonuo nuon Eoum muoum mmmnm>< Amy um£uo mama Boum muoum sz “mauo mHmEmw Eouw muoom Amv .mHaEum ucmEuwmuu Hmsufi>wvafi mnu a“ Naamcwwwuo mum3_um mmmze n- .muoz¥ w m N ¢.¢H N.m u m N m a m m.m m.w wa. mm.o m NmNmnm g as as as as N N o m.m N.N u N m N N m m.NH w.w u mam. NN.0 z mm¢om< Amv Amv Amy Amy Amy 0 m N m.NN a.N w m m o m N.m ¢.HH ¢ow. n--- m quomz Amv Amy Amv Amv Amy 0 m.o m.¢ o.¢ N.N u o o o q N N.N m.w m¢N. o~.o- m *mNmonH Any Any Amy Amy Amy 0 q o m.m . m.H m N m o N m.m N.oHu ooH. qw.o z owaqzm cc as as as as N m m m.m a.oH N ¢ N m o e.o ¢.H mmo. om.o m oaooom Amv Amv Amy Amy Amy m N N N.N : N.m u o N o o N m.w . m.H New. Nfi.o m NNNonm A5 to as a: Ev o w w o.w~u o.oH m N N ¢ N m.m u m.oHu ¢m¢. wo.o m mHHmZH as as 95 as as m m m m.m m.NH o o N m N m.N u m.HH 0mm. «0.0 2 mNHmoh as as as as as Ilm S & I I S lmlrfl 0 I I I .0 [W m. m. n J... m m. an. n a m .. m m s 1 p 1.. ... m m m p... m. w m w m m m m o c. a o c. a c . 3 I. s a 8 8 a S N 8 n 8 S n 9 S 0 9 a m a m a one. aw nNu xmm 30$.an mumnuo «Ham Hanumuxu mandamus asouu ucweumoua anumsancoz mo mmuoum ummHuumom ma mamm£on Hmnnm>ucoa nonuo no acnmmmnnxm Nb mNo>coo .nmom paw Noonxam mo unamoaxm ecu nounmmm omammop m mm Nun>nmmmo mom: can c0nuumnmuan Hmcom unannouan Eonm azmnpnuns paw uonsv macaw .nmwcm no mmocnxnam .Eoponon mo onenmmmnmxo nonno>ucoc an poxo>coo dunno on mucmnmmmnpcn no Nunnnumom .GOnqu nnonunmo HmHUOm unanno o» munommo mumnmom .mnmsuo cun3 namEm>Ho>an umcnmww wmcowmp m on Nun>nmmma mom: was cenuomnounn fiasco unannouan Eonm czmnpsun3 paw nonsv macaw mq<0m NV UZHmoqomHnumAMm Roz .mfiOnumunHmenumn nonuo suns denunmon o>nwcowop soc m mosmnnnmummuon Nnxonsv nan .Nnnnmuamaoe mnmSuo mo weensndo umooom paw now: On madam .opmomm new Immune m Np monomoaoo Nnnmnunmm on wcnxcnnu mo Nunpnwnm .n oaNo coco m20n>no moon on .q owmenumnom ponnmmp m noononm cu unemwo may .mnmnuo mo omennmmm no mason unmoum no now: On nos madam somnom .cnwnn Nnm> macaw enouosnum NunHmuomnmm .mnamuonvona coomo one con53 nauseoumum m>nmaomov mxme no mNnHmGOHumn .chNanao mmmnmxm Nnnmscnucoo cu mEomm .mwmannmnmm vmnnmop .m oEOm nownonm cu unommm o50n>no am moxmz 02H 84m MMDmQHUmH a: 5mm V/ 9O .nmow was“ no.3 mnnm nunomEoocs mnmonaw Eovnom paw wcnmoHoonu nunom Manon pnmsou pmuw>nuos Nnnwn: on On madam conned one .uamnmamcmnu manage no: :wSOSun< .menuomnmncn Hmaomnmmnmucn an mnmcuo mam unmm usonm mwsnnomm paw muswsonu wannmsm an po>no>an NHo>nuom mH .mnonuo cu meanuommn coco anew ozone mnoE mommonaxm .Hmow mnnu suns announONEoocs mnmmmnm moEnumEom use .mnonuo nuns ponmnm on coo nonns Nummcon nunom vnm30o pmum>nuos Namanscow mEmom .osnu cu menu Eonm Emnu mnwo>mn can mwcnnoom :30 nun3 nomucou voow an «Boom mMDmoqomHnumAmm k\\ m N / .maxu once mounEOOHQo moans: ozone nm>m Nanmn non nouns: ozone Nausosvmnm 0:3 somnoa one .mwannoom nwcomnma $30 mnm0>mn Nnmnwn nan mEoHnona paw mwannmom .mnmnuo now pmdnoucoo paw Uo>no>cH NaoansSow mEomm was cOnuom .w unoucn Hmcomnoanoucn an mmumanonunma amumo .conmonmxo nmconuoem an m>nm ncmmop coco pounnnscn mnoE msmmm .moanu oumanucn no mwcnnmmw Swan nonumn monQOu ammo can women cu Nnownmn denuano>aou manuonnumon oboe Honoom m on NunEnomcoo Na pounenn on non Annamwcoo can wcnowuso on Na: .wnmm mnwo>wn Naonmn nan somnma .N Nfipcmnnm NnnmSOnuco>coo m we anon man mNmHm \. . fl Namnanosoov maaom u2nn