‘K 13" . 1‘21‘15155“: .' fi:§fi may; . I . ' €961“? I‘F‘VL “3:: 1:31: C1. 1‘1.“ . .131‘1‘113‘11‘; ’ 1‘1: IX‘XJ: - ‘ . .Lfljfi‘l“ ,‘1 It.“1 3 3: «J‘L J'“ H‘g‘x ‘13}: 13;;- 131;;1is‘;' 1:211:13: "In; I. . M: - 3711‘1 ‘ .141‘ “‘1‘; 1371: v-‘i‘ ‘m. . x “we“, 1"; m 1323.- “:‘52‘13313: ‘ ‘ R1“ ”13.1. 3. . “{ 1.3.1 u I.“ I A h I 1; 3 ' m1 3 . 3 .31.. 4 H1“ ;;‘ m.;‘1‘“"‘3~' $21.18 {.1 '1'.” H1313: 1 '1!" ’1‘ ‘31;;I;11;.. “11% I “IE‘LI % 5‘1; £33.31 awfin‘: FM»? 3 1.31:“ um}: finll‘ , I“: “I £3311“? 1.: Q3153. 31%;sz 1 ‘ 131. 3:3? 31133:: F133;“ “3311-" u. h:- “‘Ii‘UX‘gE ‘3: 13w”. 1:32:11; wIIt‘fi‘l‘ “3‘1; 1.“:‘1 W ‘1; “3:33:32“ 11 v 3!} ;:,;:-I‘I...i;;-; ' m L, .J H “‘3‘ a. r 1.7. I: ‘s‘fiwx‘: 1"“.3‘3 1‘:.“ :‘I Lin 3:33.533. 3. 4.“ ;:‘ 1/- I... ‘c: ; ‘3‘ }:::II;;‘:‘ “1:51:15: \ “Lac “5;: fig; $3 .1.“ . .111 “‘3‘: 3:31. {311‘ 11.1% I I 3.:{11‘ 1&{1‘35- . ,1: 311%.; . : “1““ “3% (fig: ,Ig‘I .1354 L“ g‘fi‘ "‘ ‘IIlL #2:;‘Nhg‘ "5:33.33 ‘12:: ..11 khfiiikaxn '1. 11:33:31}? 13.13:“; a 1:331: xiii}: .4...“ :33. ‘3 "T 5% 3:59;} R0121? .. 3: :51“. .‘3 3333: :v“ M: 13;} 1:? :fi‘ '1‘: 3.131% um- “It? 1 1;: 11%;... 9342'“? +33: 1. a. 5 .. t ‘1. x ‘ II ' . ,‘5. R‘fi 1.13; 3133 3‘13: 1.1, mm; 31%" LA, '3; 113%: X. 111:1...‘1; 11391351. ‘T‘: :35 :‘Q‘: {‘3‘ 1“: 1.; ‘3‘; 3133 3:1: ‘I. Km: 21;. 13:5 .ml‘fif Ih.‘ 3 3:3 ‘ a. 3.35:3.va 13‘“ 1.1““ :L‘.‘4 ‘16, I a v "up 1 1.. 1 :‘1" ”W “"31“. ”a“. ' [q {$351k I‘IJII‘ 2‘” \N'V‘V;1‘|':l 3.. .2‘ :“:""v ‘ .1‘; 1 I‘m“: iplg; “‘I‘. WI" 1' ”W131“ “1%! Q“ 1;! .‘ t‘wf‘iié; 111:“ 'I‘;‘.:. ‘I‘lfl H J.“ J. h “91;"; IH'I ‘3“;‘L‘31Ifi‘l‘m‘u ;;.,I"‘I.n ‘3‘; . 1‘1“" 1‘3: ‘1;h3“"‘1 . 33:351. 3‘ 33;.‘3‘ 4.3:." $3M!“ I W: ‘. L 3; 13‘ 3:; ~3- ‘ | III ‘3‘ l .I , :‘HW 1|I... ;i,‘::‘ ';,‘ :1 IL. . 1'; .3 .l ... .‘x: LX 1.1 3.31 . 3 3:35;! 1333‘?" @133; 2351;. 13311 ‘3. ““1“”: “WM“: 3; 331.. 1-81.31 3%ka13é7; .... . 13.33115“ "L .33.: ”1‘3? \ '\O :1; 3:... .1 113332 1%}, I. Eff. I": 33511]; ‘3‘": ‘ lllflflfllllllllTlllflTlllll 5 293 00647 2447 l;fjfi ’ ?_II University This is to certify that the thesis entitled THE EFFECT OF THERAPIST—CLIENT PERSONALITY SIMILARITY ON THE PROCESS AND OUTCOME OF SHORT-TERM PSYCHOTHERAPY presented by ROBERT STEVEN COLEN has been accepted towards fulfillment of the requirements for 12h D degreein COUNSELING PSYCHOLOGY r professor Date May 5, 1978 0-7 639 mum 292' THE EFFECT OF THERAPIST-CLIENT PERSONALITY SIMILARITY ON THE PROCESS AND OUTCOME OF SHORT-TERM PSYCHOTHERAPY BY Robert Steven Colen AN ABSTRACT OF A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Counseling, Personnel Services and Educational Psychology 1978 G, 11:20:52? ABSTRACT THE EFFECT OF THERAPIST-CLIENT PERSONALITY SIMILARITY ON THE PROCESS AND OUTCOME OF SHORT-TERM PSYCHOTHERAPY by Robert Steven Colen The major purpose of the study was to examine the effect of therapist—client personality similarity on the process and outcome of short-term psychotherapy. While previous research in the area has generally offered con— tradictory findings, the importance of the similarity/ dissimilarity hypothesis for psychotherapy highlights the need for further exploration. The study was conducted at three outpatient community mental health centers in the western Michigan area. The client sample was drawn from these clinics during an ll— month period from October, 1975, to August, 1976. Out of one hundred forty clients who volunteered to participate, seventy-four eventually comprised the client sample. Twelve experienced psychotherapists from these clinics served as the therapist sample. Therapist—client personality similarity/dissimilarity was determined post hoc on the basis of therapist and client scores on the Judging—Perceiving dimension of the Myers-Briggs Type Indicator — Form F, Robert Steven Colen administered to both groups prior to the start of therapy. Therapists treated clients in a routine manner for a mini— mum of three therapy sessions. Dependent measures were obtained at the conclusion of this third session. Several psychotherapy process variables were evaluated by client and therapist ratings on the Barrett—Lennard Relationship Inventory — Therapist and Client Forms. Therapeutic outcome was determined by pre—post—treatment gain scores on the Gough-Heilbrun Adjective Check List and by client post— treatment self-ratings on a Survey Questionnaire. Collection of data continued until each of the twelve sample therapists (six Judging, six Perceiving therapists) saw a minimum of two similar and two dissimilar clients. A multivariate repeated measures design was employed in the study. The independent variable was therapist type, with two levels (Judging and Perceiving therapists), while client type, also with two levels (Judging and Perceiving clients), served as the repeated measures factor. Since therapists in the study treated both Judging and Perceiving clients, the assumption of indepen— dence across the client groups was violated. A repeated measures design was thus utilized to control for this dependency across client types. Three research hypotheses were tested using four multivariate repeated measures analyses of variance. Statistical significance was set at the a = .05 level. Hypothesis I was not rejected as no significant difference Robert Steven Colen was found between Judging and Perceiving therapists on the dependent measures of therapy success. Hypothesis II was also not rejected as Judging and Perceiving clients did not significantly differ on the therapy process and outcome measures. Hypothesis III showed no significant interaction between therapist and client types based on the MBTI Judging—Perceiving similarity measure. Similar and dissimilar therapy dyads displayed no significant difference on the dependent variables. No support was thus obtained for the similarity/dissimilarity hypothesis (Hypothesis III). Supplementary analyses explored the relationship between global measures of personality similarity/ dissimilarity and measures of therapy success. Cronbach's D and D2 techniques, using therapist-client difference scores on the MBTI, were used to assess global personality similarity. Multiple regression analyses failed to show significant predictive capacity for the measures of therapy success based on the global similarity procedures. Further, bivariate plots did not suggest either a linear or a curvilinear relationship between these variables. Personality similarity/dissimilarity (measured by discrete personality dimensions or global indices) and measures of therapy process and outcome thus appeared to be unrelated. Nonetheless, various methodological issues were presented which suggested the need for further exploration in the area. Such research appears justified before the similarity hypothesis is rejected. TO SUNNIE The Sunshine of My Life ACKNOWLEDGEMENTS Many people have graciously assisted in the preparation of this dissertation. I sincerely acknowledge the following individuals for their help, guidance, and encouragement: Alex Cade, chairman of the doctoral committee, for his steadfast interest and wise tutelage during my six years of graduate training. Cecil Williams, Director of the Michigan State University Counseling Center, research advisor, friend, for his rare patience, dedication and humanness. Dave Wenger, practicum supervisor, member of the doc- toral committee, for his continual encouragement and responsiveness to my needs. Dick Johnson, committee member, for his assistance with the statistical aspects of the research and his accessi— bility throughout the research project. Nancy Martin, statistical consultant, for teaching me the rigors of experimental research and for showing me a great deal of patience and interest during several trying months. Judi Trott, superb typist and editor, for her capacity to perform graciously and supportively under pressure. Her work speaks for itself. Alice Feingold and Judy Spratling, for effectively iii typing early drafts of the dissertation. Haskell Mermelstein, grammarian par excellence, for long hours spent editing, proofreading, and perfecting the present manuscript. Rose Pecker, for her assistance and fortitude during the last frantic hours prior to mailing deadlines. Friends, colleagues, therapists and clients at the Muskegon County Community Mental Health Service, Muskegon, Michigan, and the Oceana County Community Mental Health Service, Hart, Michigan, for volunteering their time and energy during the early stages of the research. Special “ thanks go to Dr. Gary Vreeman and Dr. Ralph Ford, Clinic Directors, for allowing the study to be performed at their respective agencies. Kathy Reider, Terrace Center supervisor, for making the facilities of the Terrace Center available for the research and for her constant support and friendship. Louann Dabrowski, for her important and efficient assistance in collecting the data during the first year of the study. Likewise, Cheryl Bergren and Marsha Polsen, for their aid in data collection. Friends and family, for their strength and encouragement during the long years of the research. My parents for continually supporting my endeavors with patience, caring and love. Mom, this one's for you. Above all, Sunnie, wife, fellow psychologist, best friend, for her love, devotion and self—sacrifice throughout the course of this often trying experience. iv TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . . . . . . CHAPTER I: THE PROBLEM . . . . . . . . . . . . Need for the Study . . . . . . . . . . . . Purpose of the Study . . . . . . . . . . . Definition of Terms Used in the Study . . . Research Hypotheses . . . . . . . . . . . . Theory . . . . . . . . . . . . . Overview of the Study . . . . . . . . . . . CHAPTER II: REVIEW OF THE LITERATURE . . . . . Therapist—Client Personality Similarity . . Studies Indicating a Positive Linear Relationship Between Therapist-Client Similarity and Therapy Process and/or Outcome . . . . . . . . . Studies Indicating a Curvilinear Relationship Between Therapist- —Client Similarity and Therapy Process and/or Outcome . . . . . . . . . . . . . Studies Indicating No Relationship Between Therapist—Client Similarity and Therapy Process and/or Outcome . Studies Indicating a Negative Linear Relationship Between Therapist— Client Similarity and Therapy Process and/or Outcome Or a Facilitating Effect of Therapist—Client Compatibility Upon Therapy Process and/or Outcome . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . CHAPTER III: METHODOLOGY . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . Setting . . . . . . . . . . . . . . . . . Client Sample . . . . . . . . . . . . . . . Therapist Sample . . . . . . . . . . . . . Procedure . . . . . . . . . . . . . . . . . Page .viii . l O l . 3 . 3 . 5 . 6 . 12 . l4 . l4 . l6 . 27 O 30 . 34 . 41 . 43 . 45 . 45 . 46 . 46 . 47 . 48 Instrumentation . . . . . . . . . . . . 52 Myers- Briggs Type Indicator (MBTI) . . . . . 53 Reliability and Validity of the MBTI . . . . . . . . . . . . . . 55 Similarity Dimension Used in the Study . . . . . . . . . . 59 Barrett— —Lennard Relationship Inventory (BLRI) . . . . . . . . . . . . 63 Reliability and Validity of the BLRI . . . . . . . . 66 Gough— Heilbrun Adjective Check List (ACL) . . 69 Reliability and Validity of the ACL . . . . . . . . . . . . . . 72 Survey Questionnaire (SQ) . . . . . . . . . 74 Reliability and Validity of the Survey Questionnaire . . . . . 75 Design . . . . . . . . . . . . . . . . . 76 Research Hypotheses . . . . . . . . . . . . . . . 79 Analysis . . . . . . . . . . . . . . . . . . 82 Summary . . . . . . . . . . . . . . . . . . . . . 83 CHAPTER IV: ANALYSIS OF RESULTS . . . . . . . . . . . 85 Hypothesis I: Therapist Effect . . . . . . . . . 87 Hypothesis II: Client Repeated Measures Effect . . . . . . . . . . . . . . . . . . . 89 Hypothesis III: Therapist—Client Interaction Effect . . . . . . . . . . . . . . . . . 91 Global Measures of Personality Similarity . . . . 92 Summary . . . . . . . . . . . . . . . . . . . . 100 CHAPTER V: SUMMARY AND DISCUSSION . . . . . . . . . . 103 Summary of the Study . . . . . . . . . . . . . . 103 Discussion . . . . . . . . 108 Limitations and Suggestions for Future Research . . . . . . . . . . . . . . . . . . . 124 APPENDICES A. Summary Data for Clients Used in the Study . . . . . . . . . . . . . . . . . . . . 129 B. Summary Data for Clients Not Used in the Study . . . . . . . . . . . . . . . . . . . . 130 C. Summary Data for Clients Who Terminated Early in Treatment (Less Than Three Sessions) . . . . . . . . . . . . . . . . . . 131 D. Summary Data for Therapists Used in the Study . . . . . . . . . . . . . . . . . . . 132 vi Explanatory Material for Prospective Client Subjects . . . . . . . . . . . . . . Memorandum Presented to All Intake Therapists . . . . . . . . . . . . . . . . Initial Instructions for Client Subjects . Client Consent Form . . . . . . . . . . . . Therapist Data Sheet . . . . . . . . . . . Number of Similar and Dissimilar Clients Based on the MBTI Judging-Perceiving Dimension for Each of the Twelve Sample Therapists . . . . . . . . . . . . . . . . MBTI Type Table for the Therapist Sample . MBTI Type Table for the Client Sample . . . MBTI Type Table for the Original Population of Therapists . . . . . . . . . . . . . . . PHI Coefficient Intercorrelations Between MBTI Type Categories . . . . . . . . . . . Survey Questionnaire . . . . . . . . . . . Cell Means for Sum and Difference Variables . . . . . . . . . . . . . . . . . Step-Wise Regression to Analyze the Contribution of Each of the Four MBTI Dimensions in Predicting Therapy Success . Supplementary MANOVAS . . . . . . . . . . . Univariate Analysis of Variance with a Repeated Measures Factor (Clients) for the Barrett-Lennard Relationship Inventory Subscale (Client Form) — Unconditionality of Regard . . . . . . . . BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . 133 134 135 138 139 140 141 142 143 144 146 148 152 156 157 LIST OF TABLES Table Page 3.1 Possible Therapist—Client Dyadic Combinations Based on the Judging—Perceiving Dimension of the MBTI . . . . . . . . . . . . . . . . . . . . 51 3.2 Sixteen MBTI Personality Types . . . . . . . . . 55 3.3 Correlations Between MBTI and Gray—Wheelwright Psychological Type Questionnaire for 47 Male College Students . . . . . . . . . . . . . . . . 57 3.4 Corrected Split—Half Reliability Coefficients of the Barrett—Lennard Relationship Inventory Scales, from Data Gathered after Five Therapy Sessions . . 67 3.5 Test—Retest Reliabilities for the Four Adjective Check List Scales over a Ten—Week Period . . . . 73 3.6 Pearson Product—Moment Correlation Coefficients Among the Major Dependent Variables . . . . . . . 77 3.7 Design of the Study . . . . . . . . . . . . . . . 78 4.1 Multivariate Analysis of Variance with a Repeated Measures Factor (Clients) for the Barrett-Lennard Subscales (Client Form)~Level of Regard, Empathic Understanding, Congruence, Unconditionality of Regard . . . . . . . . . . . . . . . . . . . . . 88 4.2 Multivariate Analysis of Variance with a Repeated Measures Factor (Clients) for the Barrett—Lennard Subscales (Therapist Form)—Level of Regard, Empathic Understanding, Congruence, Uncondi— tionality of Regard . . . . . . . . . . . . . . . 88 4.3 Multivariate Analysis of Variance with a Repeated Measures Factor (Clients) for the Barrett-Lennard Total Relationship Scale (Client Form), Barrett— Lennard Total Relationship Scale (Therapist Form), and Survey Questionnaire . . . . . . . . . . . . 89 viii Table Page 4.4 Multivariate Analysis of Variance with a Repeated Measures Factor (Clients) for the Adjective Check List Subscales — Number of Favorable Adjectives Checked, Number of Unfavorable Adjectives Checked, Self-confidence, Personal Adjustment . . . . . . 89 4.5 Multiple Regression Analysis and Canonical Correlation Analysis Using Therapist- -C1ient MBTI D- Scores to Assess Global Personality Similarity . . . . . . . . . . . . . 94 4.6 Multiple Regression Analysis and Canonical Correl tion Analysis Using Therapist— —Client MBTI D —Scores to Assess Global Personality Similarity . . . . . . . . . . . . . . . . 95 4.7 Step—wise Regression to Analyze the Contribution of Each of the Four MBTI Dimensions in Predicting Therapy Success . . . . . . . . . . . . . . . . . 98 4.8 Multiple Regression Analysis Using Therapist—Client MBTI Difference Scores on the J—P Dimension to Predict Therapy Success . . . . . . . . . . . . . 99 5.1 Client and Therapist Barrett-Lennard Relationship Inventory Mean Scores for Judging and Perceiving Therapists . . . . . . . . . . . . . . . . . . . 109 5.2 Client and Therapist Barrett—Lennard Relationship Inventory Mean Scores for Male Judging, Female Judging, Male Perceiving, and Female Perceiving Therapists . . . . . . . . . . . . . . . . . . . . 112 5.3 Multivariate and Univariate Analysis of Variance for Client MBTI Continuous Scores for the Three Clinics . . . . . . . . . . . . . . . . . . . . . 117 5.4 Multivariate and Univariate Analysis of Variance for Therapist MBTI Continuous Scores for the Three Clinics . . . . . . . . . . . . . . . . . . 118 ix CHAPTER I THE PROBLEM The present investigation is an attempt to study the effect of therapist-client personality similarity on thera— peutic process and outcome. The need and purpose of the study, along with relevant theoretical material, are de- tailed in this chapter. Also included are a description of terms used in the text and a statement of the research hypotheses. An overview of the study concludes the chapter. Need for the Study Research in psychotherapy has repeatedly demonstrated that the factors underlying constructive therapeutic change are indeed complex. A plethora of research over the past twenty years has explored those variables that are felt to be crucial to both therapeutic process and outcome. Some of the variables which have occupied the attention of researchers are various client variables, therapist varia- bles, therapist—client interaction variables, and technique variables. Each of these variables has at one time or another been shown to affect therapy process and/or outcome (Whiteley, 1967). A key factor which seems to supersede the importance of each of these variables, however, is the 1 ¥___ nature of the relationship between therapist and client. There appears to be a general consensus among psycholo- gists that the therapist-client relationship is of major importance for the ultimate success of psychotherapy (Kell & Burow, 1970; Bordin, 1968). One important variable which many researchers feel has a marked effect on the nature of the therapist-client relationship is the similarity or dissimilarity of the per- sonality structures of therapists and clients (Carson & Heine, 1962; Mendelsohn & Geller, 1963). It may then be hypothesized that the greater the degree of personality similarity between therapist and client, the more positive will be the therapeutic alliance and the more likely it will be that constructive therapeutic change will take place. Conversely, it may be hypothesized that the greater the degree of personality dissimilarity between therapist and client, the weaker will be the therapeutic alliance and the less likely it will be that constructive therapeutic change will take place. The importance of this similarity/ dissimilarity hypothesis for psychotherapy is reflected in a statement of Meltzoff and Kornreich (1970): The similarity-dissimilarity hypothesis is an intriguing one with profound implications for case assignment. It is a tempting thought that success ratios could be improved by informed and systematic matching of the personalities of therapists and patients. (p. 312) While several research studies have focused on the similarity/dissimilarity hypothesis over the past two decades (e.g., Mendelsohn & Geller, 1963, 1965, 1967; Wogan, 1970; Braun, 1971; Whittlesey, 1972), results of these studies have generally shown discrepant findings. The employment of a wide diversity of methods, personality variables, therapist and client samples, and process and outcome measures in some of the studies makes a meaningful comparison of findings extremely difficult. While results thus far have generally been inconclusive, the importance of the similarity/dissimilarity hypothesis for psycho- therapy highlights the need for further exploration. Purpose of the Study The current study has been designed to provide addi- tional research data concerning therapist-client personality similarity/dissimilarity. An attempt has been made in the study to correct some of the methodological weaknesses of earlier research (See Chapter II). In so doing, a more definitive statement can be made concerning the utility of the similarity hypothesis for psychotherapy. The purpose of the present study will be to examine more closely the topic of therapist-client personality similarity/dissimilar— ity and to study its effect upon therapeutic process and outcome. Definition of Terms Used in the Study In the pages that follow, several terms are employed which require clarification for the reader. An explana- tion of each of these terms appears below: Similarity Hypothesis (Similarity/Dissimilarity Hypothesis): The hypothesis which states that personality similarity between therapist and client will facilitate therapeutic process and outcome. Judging Types: Term for individuals whose characteris- tic style or "preference" is to form quick judgments about phenomena which have been perceived. Judging types are operationally defined as those individuals who score toward the Judging end of the dichotomous Judging—Perceiving scale on the Myers-Briggs Type Indicator. Perceiving Types: Term for individuals whose charac- teristic style or "preference" is to become aware of phenomena without forming quick judgments. Perceiving types are operationally defined as those individuals who score toward the Perceiving end of the dichotomous Judging- Perceiving scale on the Myers—Briggs Type Indicator. Therapeutic Process'and Outcome: Therapeutic process refers to the nature of the therapist-client interaction (i.e., hgw therapeutic change occurs) while therapeutic outcome refers to the measurement of specific therapy change variables (i.e., what_change has occurred as a result of treatment). Some writers maintain that there are essential differences between process and outcome measures (Luborsky, 1959; Farnsworth, 1966; Paul, 1972). Nonetheless, other investigators hold the View that process and outcome are both aspects of therapeutic change and that the two should be considered co—equally in assessments of therapy success (Myers, 1966; Wellman, 1967; Kiesler, 1966). The latter position is favored in the present study. Measures of Therapy SuCCess Or Therapy Gain: Terms which refer to all dependent measures of therapeutic pro— cess and outcome used in the study. Therapy; Therapist: Abbreviated terms used in the text as synonyms for psychotherapy and psychotherapist, respectively. Research Hypotheses The following research hypotheses have been formulated for the present investigation. A more detailed presentation of these hypotheses appears in Chapter III. Hypothesis I: There will be no significant difference between Judging and Perceiving therapists on the various measures of therapy success. Hypothesis II: There will be no significant differ— ence between Judging and Perceiving clients on the various measures of therapy success. Hypothesis III (Similarity Hypothesis): Therapists and clients whose Myers—Briggs types are similar on the Judging—Perceiving dimension will score significantly higher on the measures of therapy success than therapists and clients whose Myers-Briggs types are dissimilar on the same dimension. Theory The similarity/dissimilarity hypothesis has its roots in theoretical formulations advanced by two seemingly independent psychology specialties — social psychology and clinical psychology. The influence each of these disci— plines has had on the similarity hypothesis is described below. Most of the "similarity" research in the field of social psychology has appeared in the context of the broad area of interpersonal attraction. Interpersonal attraction has been shown to be an important ingredient of psycho— therapeutic success (Rogers, 1961). Byrne (Byrne & Griffitt, 1973) defines interpersonal attraction as "a construct refer- ring primarily to an individual's affective evaluation of another individual" (p. 318). The nature of a person's "affective evaluation" of another person has been a major research concern of Newcomb (1953, 1961). Newcomb has pro- posed the theory that peOple with similar "orientations" or attitudes are attracted to one another. When an individual encounters another person who has conflicting attitudes, a state of strain arises in the relationship. Since this strain is uncomfortable, the first individual will seek to resolve the strain by forming an attachment to a different person who holds similar attitudes. This "strain toward symmetry," similar to the balance theory of Heider (1958), is the basis of Newcomb's theory of interpersonal attraction. According to Newcomb, symmetry (i.e., state of balance or equilibrium) in relationships is important in two ways. First, symmetry allows a person to predict another person's behavior with more certainty. Second, the more a relation- ship approaches symmetry, the more confident the individuals can be of their own "orientations" or attitudes. Individuals thus attempt to "consensually validate" their own attitudes through the interaction (Newcomb, 1953). Thibaut and Kelley (1959) present a theoretical model of interpersonal attraction which complements the model pro— posed by Newcomb. While Newcomb postulates a “strain toward symmetry" to explain interpersonal attraction, Thibaut and Kelley use a behavioral approach to explain the reinforcing value of such symmetry. Thibaut and Kelley's exchange theory of attraction explains interactions in terms of rewards exchanged and costs incurred. Reward is defined as any activity on the part of an individual which contri- butes to the gratification of another individual's needs. On the other hand, cost is conceived as the punishment incurred in carrying out that activity, such as anxiety or fatigue. Thibaut and Kelley postulate that a person is rewarded for engaging in interactions with another person who holds similar beliefs or values provided the first per- son is in need of social support for opinions or attitudes held. Opinion support is then considered to have learned reinforcement value. Thus, two individuals with similar values will provide rewards for each other merely by expressing those values. Since it is relatively easy to express values to another person who will be receptive to those values, this can be considered a low-cost operation. While Thibaut and Kelley employ a reward-cost model to explain the facilitative effect of value similarity on interpersonal attraction, they use the same model to explain the facilitative effect of personality similarity on interpersonal attraction (Thibaut & Kelley, 1959). Therefore, people who are similar on certain personality "traits" will find their interactions to be the most rewarding while incurring the lowest costs. Three research studies by Byrne and his associates (1967, 1967, 1973) add further support to the work of Newcomb and that of Thibaut and Kelley. Byrne (Byrne & Griffitt, 1973) maintains that similarity of attitudes between two people is usually preferred to dissimilarity of attitudes. He states that interpersonal attraction is a positive linear function of the proportion of shared simi- lar opinions and attitudes between two individuals. In explaining why similar attitudes influence attraction, Byrne, like Newcomb, postulates that similarity is a posi- tive experience in that two people will View the world in a similar fashion. This tends to be reinforcing as the individuals are then provided with a social validation of their views (Byrne & Clore, 1967). Thus, the theoretical formulations of Newcomb, Thibaut and Kelley, and Byrne explain the facilitative effect of similarity on interpersonal attraction. Since interpersonal attraction has been found to be an important element of therapeutic success, these formula- tions highlight the importance of similarity in psychotherapeutic relationships. The "clinical" roots of the similarity hypothesis can be traced back to theoretical formulations of clinical psychologists and psychiatrists who stress the importance of the relationship in psychotherapy. Such eminent theor— ists as Fromm—Reichmann (1950), Sullivan (1953), and Rogers (1951, 1961) all are in agreement that the therapeutic relationship between therapist and client is central to the success of therapy. In the words of Fromm—Reichmann (1950): We know now that the success or failure of psychoanalytic psychotherapy is, in addition, greatly dependent upon the question of whether or not there is an empathic quality between the psychiatrist and the patient. A careful evaluation regarding the personalities of both patient and doctor and of the psychopathology of the patient should be helpful in the decision. (p. 62) Fromm-Reichmann thus believes that the therapeutic experi- ence is actually an interpersonal exchange between the patient and the therapist. A positive, empathic relation— ship between the two then facilitates the process of psychotherapy. This interpersonal approach to psychotherapy has been further advanced by the writings of Sullivan (1953). Sullivan views psychotherapy as a special kind of interpersonal process which is similar to the social psycho— logical concepts of interpersonal attraction. Sullivan states: 10 The general science of psychiatry seems to Jme to cover much the same field as that which is studied by social psychOlogy, be- cause scientific psychiatry has to be defined as the study of interpersonal relations .... (pp. 367-368) Rogers (1951, 1961), perhaps more than any other theorist, has written of the profound importance of the relationship for successful psychotherapy. In his work, Rogers stresses the facilitative effect of a positive, supportive relationship between therapist and client. Factors such as positive regard, unconditionality of regard, and empathic understanding are all held to be crucial determinants of effective therapeutic relationships (Rogers, 1961). Several contemporary writers have also emphasized the significance of positive therapeutic relationships for therapy success. Kell and Mueller (1966) stress this view as they feel the relationship is the key to therapeutic growth. Furthermore, they state that "the potency of a relationship resides in the commitment and affective involvement of the client and his counselor" (p. 20). In a later work, Kell and Burow (1970) describe their prefer- ence for viewing the relationship as responsible for therapy success. They state: The first aspect of our set, not so simple as it may sound, is the conviction that an interpersonal relationship must be formed. The nature of the relationship is, of course, determined considerably by how the personali- ties of the therapist and the client or clients interact. (p. 11) 11 Bordin (1968), in his evaluation of the counseling process, also supports the importance of a strong thera- peutic relationship. In his words: In most cases, both therapeutic progress and therapeutic impasses tend to polarize around the relationship between counselor and client. (p. 402) Finally, Patterson (1966) after summarizing the major therapeutic approaches, concludes that the common element in all methods and "schOols" of psychotherapy is the rela- tionship between the therapist and client: All approaches, then, appear to involve a relationship between a counselor and a client each of whom contributes to the relationship certain characteristics which lead to client change. (p. 498) The similarity/dissimilarity hypothesis seems to follow logically from the above theoretical formulations from the clinical field which streSs the importance of the relationship for psychOtherapy. Similar therapist-client dyads may facilitate positive therapeutic relationships as they may allow for greater understanding and ease of relating for members of the dyad. That is, therapists and clients who have similar personality styles are likely to have experienced common feelings and conflicts and, therefore, may effectively empathize with each other. In addition, these common feelings and conflicts may serve to confirm each person's own perception of self. Finally, members of similar dyads may show common affect needs 12 which might serve to draw the two individuals closer together. In the words of Izard (1960a): More dynamically, they have similar ways of expressing and receiving affect, and this actual similarity of affective characteristics might reasonably be expected to facilitate mutually satisfying interactions and experiences. For one thing, the same individuals, groups and interpersonal situations that evoke positive affect in one member of an affectively similar pair would tend to evoke positive affect in the other. Actual personality similarity should also increase the accuracy of interpersonal perception and communication, particularly the perception and communication of attitudes, values and other affective aspects of experience. This is almost tantamount to saying that §§ can perceive and communicate that which is similar to their own experience more readily than the dissimilar, the familiar more readily than the unfamiliar. (p. 50) In summary, the similarity/dissimilarity hypothesis has evolved from the separate theorizing of social psycholo— gists and clinical psychologists and psychiatrists. While the similarity hypothesis is based largely on logic and reason in the clinical arena, the hypothesis receives more firm experimental support from the field of social psychology. Nonetheless, the hypothesis has been a popular one in both fields, one that has generated considerable research as will be described in the next chapter. Overview of the Study Introductory material concerning the problem, need, and purpose of the current research has been presented along with the theoretical justification for the simi- larity hypothesis. The remaining chapters contain an 13 investigation of the problem in its entirety. In Chapter II, the literature relevant to therapist—client personality similarity is reviewed. Chapter III contains the sample, procedure, instrumentation, design, research hypotheses, and analysis techniques used in the study. Results of the tests of the research hypotheses and results of additional statistical analyses are presented in Chapter IV. Finally, Chapter V consists of a summary and discussion of the study with suggestions given for future research in the area. CHAPTER II REVIEW OF THE LITERATURE The purpose of this chapter is to review the literature pertaining to therapist-client personality similarity/ dissimilarity. Concluding this review is a discussion of some of the methodological weaknesses of past research and the attempts which have been made in the current study to correct these weaknesses. A summary of the review of the literature concludes the chapter. Therapist-Client Personality Similarity Research concerning the effect of therapist—client personality similarity on therapeutic process and outcome has generally offered contradictory findings. In some studies, high similarity has been associated with therapy success (Vogel, 1959; Welkowitz gt 21" 1967; Holzman, 1962; Tessler, 1973; Kessel, 1967; Hebert, 1968; Carkhuff & Pierce, 1967; Spivak, 1962; Axelrod, 1952; Tuma & Gustad, 1957; Izard, 1960a, 1960b; Halpern, 1955; Braun, 1971; Mendelsohn & Geller, 1963; Mendelsohn, 1966; Praul, 1970) while, in other studies, the relationship has been found to be curvilinear, with medium similarity associated with therapy success (Gerler, 1958; Schopler, 1959; Carson & Heine, 1962; Mendelsohn & Geller, 1965). In addition, some 14 15 studies have found either no relationship between the degree of similarity and therapeutic success (Mihalick, 1970; Moses, 1970; Ourth, 1964; Powell, 1970; Whittlesey, 1972; Lichtenstein, 1966; Carson & Llewellyn, 1966; Jones, 1968) or a negative relationship between the same variables (Fosshage, 1969; Cutler, 1958; Snett, 1973; Lesser, 1961; Bare, 1967; Mendelsohn & Geller, 1967; Wogan, 1970). Finally, some studies have found that therapist—client compatibility, rather than similarity, has a facilitative effect on therapeutic success (Sapolsky, 1965; Hetrick, 1970; Gassner, 1970; Seidman, 1971; Mendelsohn & Rankin, 1969). A review of the literature pertaining to therapist- client personality similarity follows below. To add some order and clarity to the large volume of research which has been summarized, studies will be organized into four major categories: 1. Studies indicating a positive linear rela— tionship between therapist-client similarity and therapy process and/or outcome 2. Studies indicating a curvilinear relation- ship between therapist—client similarity and therapy process and/or outcome 3. Studies indicating no relationship between therapist-client simIlarity and therapy process and/or outcome 4. Studies indicating a negative linear rela— tionship between therapist-client similarity and therapy process and/or outcome or a facilitating effect of therapist—client cqmpatibility upon therapy process and/or outcome 16 It should be noted that while therapist-client personality similarity studies are emphasized in the litera— ture review, studies which have assessed similarity from other perspectives (e.g., similarity of values, attitudes, needs) are included for comprehensiveness. Studies Indicating a Positive Linear Relationship Between Therapist—Client Similarity and Therapy Process and/or Outcome Several studies have found a positive linear relation— ship between therapist—client attitude or value similarity and therapeutic success. Vogel (1959) examined the hypo- thesis that similar patient—therapist attitudes toward therapy would strengthen the therapeutic relationship. In his study, Vogel utilized a total of forty—nine therapists and sixty-two of their patients from two outpatient clinics. Patients and therapists were administered the California F Scale to measure authoritarianism and, in addition, the Authoritarian—Equalitarian Therapy card sort to assess patient and therapist descriptions of the ideal therapy relationship. The criterion measures included a Therapist Rating Scale and an Observer Rating Scale which were devised by the author to assess the therapeutic relationship. Results indicated that similarity of patient-therapist descriptions of the ideal therapeutic relationship was related to higher criterion ratings. Vogel did not find a relationship, however, between patient—therapist similarity on F Scale scores and the various criterion measures. 17 A study by Welkowitz gt 31. (1967) likewise suggests that therapist—client value similarity is related to therapy gain. The researchers administered the Ways to Live Scale and the Strong Vocational Interest Blank to measure the value systems of thirty—eight therapists in psychoanalytic training and forty—four of their patients. Therapist outcome ratings of patient improvement served as the dependent variable in the study. Results showed a significant relationship between therapist—patient value similarity and therapist ratings of patient improvement. Holzman (1962) examined the relationship between increasing similarity of value judgments between therapist and patient and therapeutic outcome. Inpatients and out— patients who had been involved in intensive psychotherapy for seven months were employed by Holzman as subjects in the study while first— and second-year psychiatric residents served as the therapist sample. Change in value similarity of therapists and patients was determined on the basis of a 133—item value inventory which was administered prior to therapy and after seven months of treatment. The dependent measure, improvement in social adaptation, was measured by hospital discharges for the inpatients and by pre-post—therapy ratings of the outpatients' ability to carry out basic life functions (e.g., employment, social ties). Holzman found that improvement in social adaptation for outpatients, but not for inpatients, was significantly related to therapist—patient increased value similarity, 18 provided the patient had a similar life situation to that of the therapist (e.g., age, education, race, occupation, intelligence). The study thus further suggests that therapist-patient similarity of values facilitates therapeutic success. Several studies have shown that not only actual value similarity, but perceived value similarity can positively affect the therapeutic process. Using a counseling analogue study, Tessler (1973) related counselor-client perceived value similarity to various outcome measures. Tessler had ninety-six subjects each role-play a college student seeking help for a personal problem from a school counselor. Counselor—client perceived value similarity was manipulated by providing‘§§_with pre—information con— cerning their counselor. Among the dependent variables in the study was client satisfaction with the counseling relationship. Results showed significant differences in client relationship satisfaction between high and low value similarity groups. These findings suggested that perceived similarity of values was a meaningful variable which facili— tated the counseling relationship. Along the same lines, Kessel (1967) found that perceived value dissimilarity reduced interpersonal attrac— tion between an interviewee and his interviewer. Using a psychotherapy analogue study, Kessel asked a group of 108 male undergraduates to discuss various topics with an inter— viewer. The Study of Values was used to assess actual 19 subject values while a collaborator was responsible for inducing an attitudinal set in each interviewee regarding the similarity or dissimilarity of his interviewer's values. It was found that those subjects who were led to believe that their interviewer had valuesdissimilar to their own gave a lower rating to their interviewer than those subjects.who were led tobelieve that their iEter- VieW¢¥h Q Yalues similar to their own. In addition, subjects from the dissimilar values group were signifi- cantly less likely to want to pursue a counseling relation- ship with their interviewer. Kessel's results thus suggest that perceived value dissimilarity between a patient and therapist may hinder the formation of a positive therapeutic relationship. In addition to value and attitude similarity, other researchers have explored alternative kinds of therapist— client similarity. Hebert (1968) studied the effect of counselor-client need and sex similarity on the quality of the counseling relationship. Hebert utilized nineteen counselor trainees (twelve male, seven female) who were involved in educational-vocational counseling with a total of seventy-six clients (two male, two female clients per counselor). Counselor—trainees and clients were administered the Edwards Personal Preference Schedule prior to counseling to measure personality needs while the Communication Rating Scale, completed by each participant after the second interview, was used to asseSs the counseling relationship. 20 Results showed a low positive correlation between similarity of personality needs and trainee and client perceptions of the counseling relationship. Furthermore, similar trainee- client sex pairings were associated with the highest ratings of the counseling relationship. Hebert‘s study thus sug- gests that counselor—client need and sex similarity are positively related to the quality of the counseling relationship. A study by Carkhuff and Pierce (1967) found similarity of race and social class of counselors and clients to be related to counseling gain. Four female lay counselors (one upper-class Caucasian, one upper-class Black, one lower-class Caucasian, one lower—class Black), having simi— lar training, experience, and expertise in counseling, were each asked to interview sixteen female schizophrenic patients (four upper—class Caucasians, four upper—class Blacks, four lower—class Caucasians, four lower—class Blacks). Excerpts from the counseling interviews were rated by two experienced raters on depth of patient self— exploration. Results showed that patients who were most similar to their counselors with regard to race and social class tended to explore themselves the most in counseling while those patients who were most dissimilar to their counselors tended to explore themselves the least. Finally, Spivak (1962) explored the effect of perceived personality similarity on a therapist's willingness to pursue treatment with a patient. In his study, Spivak 21 asked forty psychiatrists to complete a questionnaire for each of fifteen consecutive initial interviews with new patients. The questionnaire was specifically designed by the authOr to measure the psychiatrists' perceptions of their patients. Among other results, Spivak found that the psychiatrists were more likely to initiate intensive psychotherapy with patients whOm they perceived as being similar to themselves. These results thus suggest that perceived personality similarity positively affects the therapeutic process. While the above similarity studies highlight the facilitating effect of similarity on interpersonal inter- actions and therapeutic relationships, other studies have specifically focused on the relationship between actual therapist-client personality similarity and therapeutic success. Perhaps the earliest study which attempted to assess this relationship was a study by Axelrod (1952). Axelrod had ten psychiatrists from an outpatient clinic select two of their "most improved" and two of their "least improved" patients. Rorschach inkblot tests were then given to both psychiatrists and patients. Three judges independently rated the Rorschachs on twelve personality characteristics and then determined on the basis of these protocols which two patients were most and least like their psychiatrists. Results confirmed the authOr's hypothesis that similar psychiatrist-patient personality characteristics were related to therapy gain. 22 A later study by Tuma and Gustad (1957) examined the relationship between amount of counselor-client similarity on selected personality traits (California Personality Inventory) and the amount of client learning about self (Self-Knowledge Inventory). Fifty-eight male undergradu- ates who had applied for vocational counseling at a university counseling center were seen by one of three counselors. Counselors were doctoral candidates in psychology with one to three years of counseling experience. Results indicated that client learning about self was best when counselors and clients were similar on the measures of dominance, social presence, and social participation. The authors concluded that similarity between clients and counselors on selected personality variables fosters better client criterion performance. In a non—therapeutic context, Izard (1960a, 1960b) designed a series of studies to explore whether personality similarity is a significant factor in interpersonal attraction. Focusing on the relationship between personality similarity and friendship behavior, Izard (1960a) asked 200 high school and college students to rank order their Closest personal friends. Subjects were then administered the Edwards Personal Preference Schedule (EPPS) to assess personality similarity. Thirty pairs of best friends were chosen and were then matched with sixty control subjects who had been paired at random. An analysis of EPPS profiles revealed that mutual friends had similar 23 personality profiles and were significantly more similar than the control §§° Izard concluded that personality similarity facilitates the expression of positive affect, the latter being a key determinant of interpersonal attraction. In a follow-up study, Izard (1960b) clarified whether personality similarity among friends in the first study had existed prior to their friendship or whether personality similarity had resulted from their friendship. For this study, Izard employed a sample of college freshmen who had no prior acquaintance with each other before attending college. All students were administered the EPPS shortly upon entering college. After six months of college, forty- seven female students were asked to list their three most likeable and three least likeable classmates. Twenty-five students finally comprised the subject sample, twelve of whom were randomly paired with their "most likeable" choices, thirteen of whom were paired with their "least likeable" choices. As expected, EPPS personality profiles were significantly similar for subjects and their socio- metric choices but not for subjects and their sociometric rejections. The findings of Izard's second study further supports the hypothesis that personality similarity facili- tates interpersonal positive affect and, as a result, interpersonal attraction. Consistent with Izard's results are the findings of an earlier study by Halpern (1955) in which personality 24 similarity was found to be a vital part of the empathic process. Halpern administered an 80—item version of the Guilford-Martin Inventory of Factors to measure the personality similarity of thirty-eight female student nurses. Each student nurse was then asked to predict the performance of five other student nurses on the Guilford inventory. Results indicated that the students were able to predict ("empathize") with greater accuracy about those students who were similar to themselves than about those who were dissimilar. Halpern hypothesized that individuals can only recognize feelings and behavior patterns in others if they have experienced these same feelings and behavior patterns in themselves. While it may be questioned whether Halpern actually measured the construct "empathy," a later study by Braun (1971) partially supports Halpern's findings. In this study, Braun investigated the relationship between therapist—patient personality similarity and therapist empathic ability. Experienced VA psychotherapists and clinical psychology trainees served as the therapist sample while male "psychoneurotic" patients comprised the patient sample. Personality similarity was assessed on the basis of therapist and patient scores on the Myers—Briggs Type Indicator (MBTI). Therapist and patient Q sorts, taken between the tenth and fourteenth therapy sessions, were used to measure therapist empathic ability. Braun found that therapist-patient similarity on the MBTI 25 Thinking-Feeling dimension was significantly related to therapist empathic ability, but only for female therapists. This relationship did not hOld for the male therapists or for the combined therapist sample. Perhaps the main support for the similarity hypothesis comes from a series of studies by Mendelsohn and his colleagues conducted at the Counseling Center of the University of California, Berkeley (Mendelsohn & Geller, 1963; Mendelsohn, 1966). In the first investigation, Mendelsohn and Geller (1963) had seventy-two vocational clients and ten counselors complete the Myers-Briggs Type Indicator. Global personality similarity was measured by difference scores on the MBTI dimensions (Extraversion- Introversion, Sensing—Intuition, Thinking-Feeling, Judging-Perceiving) for each counselor-client dyad using Cronbach's D technique (sum of the differences between counselor and client scores of each MBTI dimension). The outcome variable in the study, length of stay in counseling, was the number of client sessions prior to termination. Results clearly indicated thatthe greater the MBTI simi- larity score, the greater the number of counseling sessions. Mendelsohn and Geller interpreted these findings as indica- ting greater client commitment to counseling when counselor and client have similar personality characteristics. In an attempt to replicate his first study, Mendelsohn (1966) modified his methodology and employed a new counselor— client sample. In this study, Mendelsohn used eleven 26 counselors and a sample of 201 undergraduates who had applied for educational—vocational counseling at a uni- versity counseling center. While counselors and clients again were administered the MBTI, global personality similarity was this time assessed on the basis of Cronbach's D2 procedure (square root of the sum of the squared differ- ences between counselor and client scores on each MBTI dimension). Duration of counseling again served as the dependent variable in the study. Results for this investigation replicated Mendelsohn's earlier finding in that counselor-client similarity was related to a greater duration of counseling. However, Mendelsohn found that similarity produced a greater variability in the number of counseling sessions than did dissimilarity. That is, dura— tion of counseling was short when there were dissimilar dyads, but counseling was of short or long duration when there were similar dyads. It should be noted that Mendelsohn also analyzed the contribution of each of the four MBTI dimensions but found no effect attributable to the individual dimensions. Mendelsohn's research thus provides additional evidence for the facilitative effect of therapist—client personality similarity on therapeutic outcome. Finally, a study by Praul (1970) is deserving of men- tion at this point. While Praul was concerned chiefly with counselor-supervisor relationships, her results have rele- vance for the similarity hypothesis. In her study, Praul 27 administered the MBTI as an index of personality similarity to twenty-nine counselors-in-training and their supervisors. The dependent measure, counseling effectiveness, was assessed by supervisor and peer ratings of counseling success, level of trainee empathic understanding (Accurate Empathy Scale), and trainee change in self-concept (Butler- Haigh Q Sort). Results indicated that supervisor-counselor trainee personality similarity on the MBTI Sensing- Intuition dimension was significantly related to counseling effectiveness as measured by supervisor ratings. Personality similarity was not found, however, to influence either trainee empathic understanding or trainee change in self— concept. Studies Indicating a Curvilinear‘Relationship Between Therapist-Client Similarity and Therapy Process and/6r Outcome While the research cited above suggests that a posi— tive linear relationship exists between personality similarity and therapeutic success, other studies indicate that a curvilinear relationship is operative with medium similarity producing the greatest therapeutic gain. Gerler (1958) explored this relationship in his study of five therapists and fifty-seven clients who had applied for personal-social counseling at the University of Illinois Student Counseling Service. The Ewing Personal Rating Form, administered to both therapists and clients, was used to assess personality similarity. Difference scores on the Ewing between clients and therapists were 28 classified as high, medium, or low and were then compared with judged therapeutic improvement. Gerler found that a medium amount of similarity on related personality traits was more conducive to therapy gain than a high degree of similarity, although the medium and low similarity groups were not found to be significantly different from one another. Gerler's findings thus offer only partial support for the notion that a curvilinear relationship exists between therapist-client personality similarity and therapy gain. A curvilinear relationship between similarity and outcome was also found in a study by Schopler (1959). In this study, Schopler administered the Leary Interpersonal Check List to a sample of fifteen therapists and fifty- eight patients. Similarity was defined as a correspondence between therapist and patient self-descriptions on the Leary Check List. Therapist ratings of success and the number of patient interviews served as the outcome measures. While no significant results were obtained, a trend was found suggesting a C-shaped curvilinear relation- ship between similarity and therapy success. Low therapist— patient personality similarity was associated with moderate success, while high similarity was associated with either high or low success. No relationship was found, however, between similarity and the number of therapy interviews. Consistent with the findings of Gerler and Schopler is a study by Carson and Heine (1962). Using supervised 29 senior medical students as therapists, Carson and Heine selected sixty outpatients who had been referred to a local psychiatric clinic as the patient sample. Therapists and patients were both administered the Minnesota Multi- phasic Personality Inventory (MMPI) prior to therapy and rank-order correlations of MMPI scores were then used to assess global personality similarity/dissimilarity. The sixty therapist-patient pairs were then placed into five groups of twelve pairs each on the basis of increasing similarity. Therapeutic gain was determined using a criterion of supervisor outcome ratings made at the end of treatment. Results indicated that therapy gain was associated with similarity in a curvilinear fashion, with extreme similarity or extreme dissimilarity impeding therapeutic success. While Mendelsohn and Geller's first study (Mendelsohn & Geller, 1963) found a linear relationship between simi- larity and therapy success, a later study by the same researchers (Mendelsohn & Geller, 1965) revealed a curvi— linear relationship between similarity and client attitudes toward the counseling experience. For the 1965 study, Mendelsohn and Geller used the same client and counselor sample as in the 1963 study, but made several modifications in their methodology. Cronbach's D2 global similarity procedure was substituted for the original method of matching with subjects now placed in high, middle, and low similarity groups on the basis of Myers-Briggs Type 30 Indicator scores. Questionnaires were mailed to clients three to twelve months after termination of counseling to measure client attitudes toward their counseling experience. The authors felt that the questionnaire would add substan- tial new information that was not available as a result of their first outcome measure (i.e., length of stay in counseling). While there was significant variability in the findings based on the specific criterion variable used (Evaluation, Comfort-Rapport, Judged Counselor Competence), results generally supported a curvilinear relationship between similarity and counseling success. Studies Indicating No Relationship Between Therapist- Client Similarity and Therapy Process and/or Outcome A study by Mihalick (1970) offers little evidence to support the notion that therapist—client value similarity facilitates therapeutic success. Mihalick administered the Ways to Live Scale to assess the values similarity of a sample of forty—one clients and eleven therapists at a university psychology service. Therapeutic success was determined on the basis of therapist ratings of therapy gain. Although results were in the predicted direction, client-therapist value similarity was not significantly related to the success of psychotherapy. A significant negative relationship was found, however, between client— therapist value disparity and the duration of therapy. While Mihalick's study examined the effect of tested value similarity on therapeutic outcome, a study by Moses 31 (1970) explored the relative influence of tested versus perceived value similarity on outcome criteria. Moses utilized ninety-nine female clients and four female counselors from a vocational-educational counseling center. Tested value similarity was measured by client-counselor performance on the Whitehorn—Betz A-B Scale, The Values Profile, and The Work Values Inventory. Perceived value similarity was determined by having clients complete these measures as they felt their counselors would complete them. Results generally failed to support the main hypothesis in that perceived and tested value similarity were not related to client and counselor ratings of counseling outcome. Ourth (1964) tested the similarity hypothesis in a study of forty-two outpatients and six psychotherapists at a university mental hygiene clinic. Similarity was based on therapist-patient performance on an Internal-External Orientation test which measured each participant's mode of personal validation (internally or externally oriented). Ourth found that initial internal-external similarity between therapists and patients did not influence either therapist ratings of patient improvement or duration of psychotherapy. Powell (1970) examined the effect of therapist- patient complementarity and similarity on various measures of therapy effectiveness. Using the Whitehorn—Betz A-B Scale, Powell matched ten type "A" and ten type "B" thera- pists with ten type "A" and ten type "B" patients. Senior 32 medical students were used as therapists in the study while clinic outpatients served as the patient sample. Among the outcome measures obtained after the first interview were patient perceptions of the interview, degree of patient self-exploration, and degree to which the patient related intimately with the therapist. Powell's main finding was that there was no significant difference between the complementary and similar therapist—patient dyads with respect to the dependent measures of therapy effectiveness. A study by Whittlesey (1972) again explored the effect of counselor-client personality similarity on mea- sures of counseling outcome. Whittlesey utilized fifteen college counselors at a university counseling center and 100 of their clients who had received counseling during a selected three-month period. Counselor—client personality similarity was based on Holland's personality types as assessed by the Vocational Preference Inventory. Counseling outcome was determined by several measures, among which were counselor ratings of goal achievement and counselor ratings of liking for clients. Results showed no relationship between counselor-client personality similarity and the counseling outcome measures. While Carson and Heine (1962) found a curvilinear relationship to exist between similarity and therapy outcome, Lichtenstein (1966) was unable to find such a relationship in his replication of much of the original 33 Carson and Heine study. Using third-year medical students as therapists, Lichtenstein employed a sample of fifty- four outpatients for his study. The MMPI was again used to assess global personality similarity while outcome ratings made by supervising psychiatrists once more served as the criterion of therapy gain. Results indicated that no relationship existed between therapist-patient per- sonality similarity and the measure of therapeutic success. In a similar attempt to replicate the Carson and Heine (1962) study, Carson and Llewellyn (1966) were also unable to support a curvilinear relationship between similarity and therapy outcome. In keeping with the earlier study, sixty-five outpatients and twenty-two experienced therapists from a public psychiatric clinic were administered the MMPI to assess global personality similarity. Therapist outcome ratings and total number of interviews were used as the criterion measures. Statistical tests failed to find a significant relationship between therapist-patient personality similarity and therapeutic outcome. The authors concluded that global personality similarity does not seem to be a very "fruitful" concept and that future researchers should employ more precise, analytical procedures rather than a global index to deter— mine personality similarity. Finally, Jones (1968) investigated the counseling relationships of five counselors and sixty community college students who had volunteered for two vocational 34 counseling sessions. Clients were matched with counselors on the basis of Myers-Briggs Type Indicator scores into two groups of high and low similarity (global index of similarity). The Barrett-Lennard Relationship Inventory was used to measure several counseling process variables while a semantic differential checklist was employed to assess counseling outcome. Jones found no significant differences between the high and low similarity groups that could clearly substantiate either a linear or curvi- linear relationship between similarity and counseling process and/or outcome. Studies Indicating a Negative Linear Relationship'Between Therapist-Client Similarity and Therapy Process and]or Outcome Or a Facilitating Effect of Therapist-Client Compatibility Upon Therapy Process and/or outcome Fosshage (1969) found counselor-client interest simi- larity to be positively related to a counselor's initial liking for a client. Nonetheless, additional results of the study failed to show a relationship between interest similarity and counseling outcome. Fosshage used the Strong Vocational Interest Blank to assess the interest similarity of eight counselors and fifty-two clients from four college counseling centers. Counselor and client rating scales were employed to measure counseling process and outcome. While results did not support a positive relationship between interest similarity and outcome, a trend suggested that counselor-client interest similarity was negatively related to counseling outcome. Fosshage 35 concluded that interest similarity may make for a more comfortable counseling relationship, but may impede client change. Cutler (1958) examined the relationship between therapist-patient conflict similarity and therapy process in his study of two psychotherapists and five of their patients. The two therapists were chosen from a pool of ten therapists based upon the dissimilarity of their conflict areas. Cutler defined therapist conflict areas as those personality traits which showed a discrepancy between therapist self-ratings and independent judges' ratings. Patient conflict areas were rated on the basis of tape recorded therapy sessions. In addition, therapist responses to patients were labeled as "task-oriented" (facilitative for psychotherapy) or 'ego-oriented" (defen- sive responses, non-facilitative for psychotherapY). Results showed that therapists were less effective thera- peutically (more "ego-oriented" responses) when their patients' behavior was similar to behavior which was conflictual for the therapist. Snett (1973) also examined the relationship between therapist-patient conflict similarity and therapeutic process. In this study, conflict similarity was based on correlations of therapist and patient scores on the Blacky Pictures Test. Five male therapists and forty patients served as subjects while various therapist ratings of the therapy relationship were used as the dependent 36 measures. No support was found for the hypothesized nega- tive relationship between therapist-patient conflict similarity and therapist ratings of patients. Nonetheless, when defensive similarity was partialed out of the index of conflict similarity, a negative relationship was found between conflict similarity and the therapy process measures. Lesser (1961) administered a Q-sort technique to eleven counselors and twenty—two clients at a university counseling center to measure counselor—client self- perceptions and client ideal self-perceptions. After several counseling sessions (range of three to twelve), each client was retested on the Q-sort to measure changes in self and ideal self-perceptions (measure of counseling. progress). Additionally, clients rated counselors and counselors rated themselves on empathic understanding. Lesser found that similarity of self-perceptions between clients and counselors was negatively related to counseling progress and was unrelated to ratings of empathic understanding. Bare (1967) investigated the effect of counselor personality characteristics, as well as counselor—client personality similarity, on measures of counseling success. Bare had forty-seven graduate student counselors and 208 counseling center clients complete the Gordon Personal Profile, the Gordon Personal Inventory, and the Edwards Personal Preference Schedule. Counseling proceeded for 37 ten consecutive weeks, one hour per week. At the conclu- sion of this ten-week period, counselors and clients were asked to rate their feelings and attitudes about their counseling experience. Results indicated that dissimi— larity on several personality traits (e.g., original thinking, vigor, responsibility), not similarity, was associated with counseling success. In the last of a series of similarity studies, Mendelsohn and Geller (1967) found results which suggested a negative relationship between similarity and counseling outcome. Investigating the counseling relationships of 201 counseling center clients and eleven counselors, Mendelsohn and Geller discovered that a client's failure to keep a scheduled counseling session was associated with global counselor-client similarity on the MBTI. Nonetheless, clients who returned after a missed session remained in counseling longer than those who never missed a session. The similarity hypothesis likewise failed to receive support in a study by Wogan (1970). Wogan factor analyzed the MMPI scores of twelve psychiatric residents and eighty- two inpatients who were involved in psychotherapy. The crossproducts of the MMPI factor scores of each therapist and patient were used to arrive at a rather sophisticated index of similarity. After two weeks of treatment, thera- pists and patients completed rating scales which assessed the nature of their therapy relationships. In addition, 38 at the end of therapy, therapists ranked their patients from most to least improved. Results indicated that personality similarity between therapists and patients was detrimental to the therapeutic relationship. For example, similarity on the Repression factor detracted from the patient's liking for a therapist, while simi- larity on the Subtlety factor was associated with a slower rate of patient perceived progress in therapy. Wogan concluded that similarity of "defensive styles" may have an adverse effect on therapy since this may interfere with the therapist's ability to objectively understand and effectively communicate with patients. While the research cited above has been concerned with the similarity of therapists and clients, other studies have focused on elements of similarity and complementarity as they affect the therapeutic process. Such studies have been classified as "compatibility" studies and have generally employed the Fundamental Inter— personal Relations Orientation Behavior (FIRO—B) scale as the measure of compatibility. The FIRO-B scale identifies three areas of interpersonal need (inclusion, control, affection) which can be used to form compati- bility scores in dyadic interactions. Sapolsky (1965) employed the FIRO-B scale in his examination of the thera— peutic relationships of twenty—five hospitalized female patients and three psychiatric residents. Sapolsky found therapist—patient compatibility of needs to be positively 39 related to both therapy outcome ratings and patient attitudes towards their therapists. Hetrick (1970) analyzed the effect of therapist— client compatibility on communication within therapy dyads. Employing a counseling analogue design, Hetrick selected twelve high— and twelve low—compatibility dyads based on FIRO—B total compatibility scores. Each therapy dyad consisted of college students who were designated as either "therapists" or "clients." Along with other mea- sures, vocal stress units (counted after each of six counseling sessions) provided an index of number of words spoken by the participants. Results showed that com— patible therapist—client dyads talked at higher rates than incompatible dyads and that therapists identified the compatible clients as being easier to communicate with than incompatible clients. Gassner (1970) investigated the relationship between therapist-patient personality compatibility and variables related to therapy success. The FIRO—B scale was used to assign a high- and a low—compatibility patient to each of twenty-four therapists. Therapists in the study consisted of pastoral counseling trainees while the patient sample was comprised of psychiatric inpatients at a state hospital facility. While results showed no differences between the compatibility groups on measures of therapy outcome (ward nurses' assessments of patient behavior change), a 40 significant relationship was found between compatible dyads and attraction in the therapy relationship. In a psychotherapy analogue study, Seidman (1971) asked twenty Type A and twenty Type B student-therapists to respond to videotapes of an intropunitive-neurotic patient and a schizoid patient (two theater arts majors enacted both patient prototypes). Subject-therapists' responses to the videotapes were rated according to the facilitative conditions of respect, empathic under- standing, and congruence. Results indicated that com- patible dyads (i.e., Type A - schizoid, Type B — neurotic) showed higher ratings than incompatible dyads on the respect and empathic understanding scales. Seidman's findings thus offer further support for the facilitative effect of personality compatibility on therapeutic success. Finally, Mendelsohn and Rankin (1969) studied the effect of sex differences on the relationship between compatibility and counseling outcome ratings. Using 162 clients (104 males, 58 females) and eleven counselors (six males, five females), the researchers found that counselor-client compatibility, measured by the FIRO-B scale, was an excellent predictor of counseling outcome for female clients, but not for male clients. While compatibility in the inclusion and affection need areas was negatively associated with counseling outcome, a 41 positive relationship was found between compatibility in the control need area and counseling gain. Discussion The research studies enumerated above offer con- flicting findings concerning the similarity-dissimilarity hypothesis. The employment of differing samples, pro- cedures, personality variables, and dependent measures in many of these studies serves to confuse the issue regarding the true effects of similarity. The exact nature of the relationship between personality similarity and therapeutic success thus remains obscure. The attempt of the present study was to explore systematically this relationship and, in the process, to correct several methodological flaws of previous studies. Specific attention was given to the following: 1. In vivo research: While some similarity research has employed psychotherapy analogue studies with role— playing of therapist—client interactions (Kessel, 1967; Tessler, 1973), the present study was an in vivo investiga— tion of actual therapist—client psychotherapeutic relationships. Since role-playing is a threat to both internal and external validity (Spencer, 1978), the current in vivo investigation was seen as a marked improve— ment over some earlier research designs. 2. Therapist sample: Experienced psychotherapists were used as subject therapists in the current investi— gation. The use of inexperienced, untrained "therapists" 42 in some previous studies has limited the generalizability of results [e.g., the use of medical students as thera- pists in the Carson & Heine (1962) and Lichtenstein (1966) studies; psychiatric residents in the Holzman (1962) and Wogan (1970) studies; therapist trainees in the Hebert (1968) and Braun (1971) studied]. 3. Type of counseling: Only clients with personal- social concerns were included in the client sample. The lack of differentiation between vocational counseling and personal-social counseling in some studies (Mendelsohn & Geller, 1963, 1965; Jones, 1968; Moses, 1970) has been a source of confusion and has again limited the generaliza— bility of results. 4. Similarity/Dissimilarity determinations: Specific personality variables rather than global similarity indices were used to determine therapist-client similarity/ dissimilarity. The use of global indices of similarity in some studies (Carson & Heine, 1962; Mendelsohn & Geller, 1963, 1965; Lichtenstein, 1966; Carson & Llewellyn, 1966; Jones, 1968) may have obscured findings along individual dimensions. Bare (1967), following the lead of Cronbach (1955), states that global similarity indices are of limited value as "they tend to mask differential personality interactions on specific dimensions" (p.420). 5. Outcome measures: Outcome measures which have good psychometric properties were used to assess client therapy gain. Criterion measures of questionable validity 43 have been employed in some previous research studies [e.g., length of stay in counseling in the Mendelsohn & Geller (1963) and Mendelsohn (1966) studiee]. Summary A review of the literature concerning therapist- client personality similarity was presented. Studies were cited which have shown a positive linear, negative linear, curvilinear, or no relationship between therapist-client similarity and therapy process and/or outcome. The main research which supported the similarity hypothesis was a series of studies by Mendelsohn and his colleagues con- ducted at the University of California, Berkeley (Mendelsohn & Geller, 1963; Mendelsohn, 1966). Using a sample of counseling center clients and counselors, Mendelsohn and Geller (1963) found a positive linear relationship between counselor-client global personality similarity, as assessed by the Myers-Briggs Type Indicator, and counseling outcome. Further work by these researchers, however, found a curvilinear relationship to exist between similarity and outcome (Mendelsohn & Geller, 1965). In an attempt to replicate a curvilinear relationship between these variables, Lichtenstein (1966) and Carson and Llewellyn (1966) employed medical students to serve as therapists with various outpatients. Using the MMPI to assess global personality similarity, the investigators found no relationship between personality similarity and 44 therapeutic outcome. Additional research was cited which has found either a negative effect of similarity (Bare, 1967; Wogan, 1970) or a facilitating effect of therapist-client compatibility on therapy success (Sapolsky, 1965; Seidman, 1971). Concluding the chapter was a discussion of several of the methodological flaws of some previous studies, e.g., failure to employ in vivo research designs, use of inexperienced therapist subjects, inadequate methods of determining similarity, use of questionable outcome measures, and failure to make a distinction between vocational and personal-social counseling. CHAPTER III METHODOLOGY The chapter contains a detailed description of the sample, procedure, and instrumentation used in the study. The design, research hypotheses, and analysis techniques are also included. The chapter concludes with a summary section. Overview Therapist—client dyads were examined to study the effect of personality similarity on therapeutic process and outcome. Twelve experienced psychotherapists and seventy— four of their clients served as subjects in the study. Pre- therapy testing for clients consisted of the Myers—Briggs Type Indicator and the Gough-Heilbrun Adjective Check List while pre-testing for therapists consisted solely of the Myers-Briggs Type Indicator. Personality similarity/ dissimilarity was determined on the basis of therapist— client similarity or dissimilarity on the Myers-Briggs Judging—Perceiving dimension. Post-treatment measures were obtained at the conclusion of the third psychotherapy session. At this time, therapists and clients completed the Barrett—Lennard Relationship Inventory to measure several therapy process variables. In addition, clients 45 46 were administered the GoughrHeilbrun Adjective Check List and the Survey Questionnaire to assess therapeutic outcome. Data collection continued over an 11-month period until each of six Judging and six Perceiving therapists treated a minimum of two similar and two dissimilar clients. Setting The study was conducted at three outpatient community mental health centers in the western Michigan area. The clinics generally serve a similar client population of white, low-middle income blue—collar workers. Client diagnoses at the three clinics encompass a wide range of treatment disorders. Therapy is typically short-term as approximately 60% of the clients terminate prior to the seventh session. Client Sample The client sample was drawn from the three clinics during an ll—month period from October, 1975 to August, 1976. Clients had to meet the following criteria to be selected as subjects in the study: 1. At least seventeen years of age 2. Of apparent normal intelligence 3. Not diagnosed as psychotic or organic 4. Not classified as an "emergency" client, (i.e., absence of severe stress which would contraindicate pre-testing) One hundred forty clients volunteered to participate. Out of this total, approximately half were later eliminated 47 from the client sample due to voluntary withdrawal, early termination from treatment, or a failure to meet the four— point criteria listed above. The remaining seventy—four clients comprised the client sample. The mean age of these clients was 31.1 years with a range of seventeen to fifty-eight years. Seventy—three percent were females and the ratio of Caucasians to Blacks was 71 to 3. All clients had at least some high school education with 69% either high school graduates or college educated. The most frequently occurring diagnoses were anxiety neurosis and depressive neurosis. Summary data for clients in the study appears in Appendix A. Descriptive information for clients not included in the final client sample can be found in Appendix B. Therapist Sample The therapist sample was chosen from a pool of twenty therapists at the three mental health centers. Twelve of these therapists met the eligibility requirements for participation in the study. To be eligible, therapists had to treat a minimum of four adult clients of specified personality types (two MBTI Judging types and two MBTI Perceiving types) for three sessions each. The twelve eligible therapists generally reflected the entire thera- pist population, i.e., white, middle income, Masters-level psychologists or counselors. Of the twelve sample thera- pists, male therapists exceeded female therapists seven to five. The mean age of these therapists was 31.8 years 48 with a range of twenty-four to fifty—two years. The average number of years of professional experience was 6.5 years. Since these therapists generally showed a diversity of education and training, along with several years of professional experience, they were considered to be an experienced, qualified therapist sample. Appendix D contains relevant descriptive information for each of the twelve sample therapists. As indicated in Appendix D, an eclectic approach to treatment was utilized by all twelve therapists. The therapists had been asked to label their primary mode of psychotherapy as either: 1) non-directive, 2) psycho- analytic, 3) behavioral, 4) eclectic, or 5) other. There was universal agreement as to an eclectic orientation despite the fact that therapists showed varying professional training. Procedure Clients were asked to volunteer for the study during their regular intake interview. While interviews typically took place on the telephone, some face-to-face contacts were also made. The purpose of the research was described at this time in general terms (i.e., to learn more about the personality characteristics of therapists and clients). Clients were informed that participation was on a voluntary basis and that their decision to volunteer would not affect the services they were to receive. Strict confidentiality was to apply at all times. Research data were to be coded 49 and therapists would not have access to client responses. Explanatory material for prospective participants can be found in Appendix E. Clients who expressed an interest in the study were asked to complete the pre-test measures prior to their first therapy session. These measures consisted of the Myers—Briggs Type Indicator-Form F and the Gough—Heilbrun Adjective Check List. Total testing time for the tests was approximately one hour. While most clients took the pre-tests at their respective clinic one hour prior to their first scheduled appointment, some chose to complete the measures at home. In such cases, test materials were mailed to volunteers in advance of their first therapy session. Procedural flexibility was necessary to insure client cooperation. Therapists at the three clinics were informed of the research study during regular staff meetings. At those meetings, therapists were told the purpose of the study in general terms (i.e., to learn more about therapist-client interactions). They were informed that the study could be useful in defining ways to maximize therapeutic process and outcome. As with the client sample, participation was to be on a voluntary basis. Therapist data would be held in strict confidence. The twenty therapists who expressed an interest in the study were asked to complete the Myers— Briggs Type Indicator — Form F during this initial stage of the study. 50 Clients were assigned to these therapists shortly after their intake interview. Systematic matching of clients and therapists by personality type was not possible due to clinic policy. The chief consideration for client assignment, therefore, was therapist-client schedule compatibility. An assignment bias was not evident as therapists generally had full case-loads with no apparent discrepancies in schedule availability. It would thus appear that essentially a random procedure was used in client assignment. The personality similarity/dissimilarity of the therapist-client dyads was determined post hoc on the basis of therapist and client scores on the Judging— Perceiving dimension of the Myers-Briggs Type Indicator. Personality similarity/dissimilarity was then operationally defined as therapist-client similarity or dissimilarity on the Judging—Perceiving dimension of the MBTI. Table 3.1 depicts the four possible therapist-client dyadic combina- tions. The distribution of similar and dissimilar clients for each of the sample therapists can be found in Appendix J. Myers—Briggs types for subject therapists and clients can be found in Appendices K and L, respectively. Therapists in the study were instructed to proceed with therapy as usual. Both therapists and clients were aware of their subject status, but were not aware of the specific variables under scrutiny. Subjects were generally 51 instructed to disregard research considerations as much as possible. Table 3.1 Possible Therapist - Client Dyadic Combinations Based on the Judging—Perceiving Dimension of the MBTI Therapist Client Dyadic Similarity/ MBTI Type MBTI Type Combination N Dissimilarity Judging- Judging Judging Judging 18 Similar Judging- Judging Perceiving Perceiving 20 Dissimilar Perceiving- Perceiving Judging Judging 14 Dissimilar Perceiving- Perceiving .Perceiving Perceiving 22 Similar Post-testing for all subjects was begun after the third session in psychotherapy. The criterion of three therapy seSsions was selected in response to the wishes of clinic supervisors to avoid a high client mortality rate. This criterion enabled a greater percentage of clients to be included in the study within the time constraints of the research project. For client subjects who met the three session criterion, post—testing consisted of the Barrett-Lennard Relationship Inventory, Client Form, the Survey Questionnaire, and the Gough-Heilbrun Adjective Check List. Total testing time for these tests was approximately one-half hour. While most clients completed these measures at their respective clinics immediately 52 after the third session, some clients, by necessity, were compelled to complete the measures at home. In such cases, clients were instructed to mail back the tests the follow— ing day. Post—testing for the therapists consisted of the Barrett-Lennard Relationship Inventory, Therapist Form. While some therapists completed this instrument immediately after the third therapy session with subject clients, most therapists completed the measure within a three—day period. Data collection continued over an ll-month period. At the conclusion of this time period, twelve of the origi- nal twenty therapists who took the MBTI met the eligibility requirement of treating two similar and two dissimilar clients for three sessions each. The final subject sample then consisted of six Judging and six Perceiving therapists and their similar and dissimilar clients. While a minimum of four clients was seen by each therapist (two similar and two dissimilar), the average number of clients seen was six (range of four to eleven). Instrumentation Several instruments were employed in the current investigation. Therapist—client personality similarity/ dissimilarity was determined on the basis of therapist and client scores on the Myers-Briggs Type Indicator — Form F. The Barrett-Lennard Relationship Inventory, Client and Therapist Forms, was used to measure several psychotherapy process variables while pre—post—treatment scores on the Gough-Heilbrun Adjective Check List were used to assess 53 therapeutic outcome. In addition, a brief Survey Questionnaire was used to measure client perceptions of therapy outcome. Each instrument will be discussed in turn. Myers—Briggs'Type Indicator (MBTI) The Myers-Briggs Type Indicator (Myers, 1962) was chosen as the measure of therapist-client personality similarity since it has been widely used in previous similarity studies (Mendelsohn & Geller, 1963; Mendelsohn, 1966; Jones, 1968). The MBTI scale dimensions seem to tap variables which have a wide applicability for personality research. The MBTI has also shown good psychometric properties with sufficient reliability and validity. The test is self-administering and requires no more than a high school reading ability. The MBTI is a personality inventory which is designed to measure variables from Carl Jung's personality typology (Jung, 1923). The MBTI yields preference scores on four discrete dimensions:* 1. Extraversion (E) or Introversion (I) 2. Sensing (S) or Intuition (N) 3. Thinking (T) or Feeling (F) 4. Judging (J) or Perceiving (P) *While only the MBTI Judging-Perceiving dimension was used to assess personality similarity/dissimilarity, subse- quent analyses utilized all four MBTI dimensions (See Chapter IV). It is for this reason that the four MBTI di- mensions will be included in the discussion which follows. 54 The E — I scale measures an interest in people and things (E) versus an interest in concepts and ideas (I); the S - N scale measures the tendency to perceive through usual sensory processes (S) or indirectly by way of the unconscious (N); the T — F scale probes tendencies to evaluate phenomena impersonally and logically (T) versus personally and subjectively (F); and finally, the J - P scale measures the tendency to form quick judgments about phenomena (J) or to become aware of them (P). Items on the MBTI scales consist of various value judgments, behavior reports, and word pairs. Questions appear in a forced choice form in which items have two opposite alternatives. One alternative reflects a particu- lar function or attitude (e.g., Judging) and the other alternative reflects the opposite function or attitude (e.g., Perceiving). Different weights are assigned to each alternative to reflect its relationship to the scale classification and to set the scale's zero point. The test thus yields four dichotomous type dimensions (E—I; S—N; T-F; J-P). Table 3.2 depicts the sixteen possible combinations of these preferences. While MBTI scores are usually treated as dichotomous data based on type classifications, the MBTI Manual (Myers, 1962) gives specifications to convert dichotomous data into continuous quantitative data. In this way, the relative strength of each type of preference may be measured. Although a type approach was favored in the 55 present study, analyses based on MBTI continuous data also appear in Chapter IV. Table 3.2 Sixteen MBTI Personality Types ISTJ ISFJ INFJ INTJ ISTP ISFP INFP INTP ESTP ESFP ENFP ENTP ESTJ ESFJ ENFJ ENTJ E = Extraversion I = Introversion S = Sensing N = Intuition T = Thinking F = Feeling J = Judging P = Perceiving Reliability and Validity of the MBTI: The MBTI has shown good internal consistency reliabilities for the Extraversion-Introversion (E—I), Sensing-Intuition (S-N), and Judging-Perceiving (J-P) scales. Reliability estimates for these scales using dichotomous data generally range from .70 to .85 (Stricker & Ross, 1963). The Thinking- Feeling (T-F) scale seems to be the only scale with a questionable reliability. Stricker and Ross (1963) found a low reliability coefficient of .43 for the T - F scale in a sample of 300 college males. Higher reliability esti- mates for this scale have been obtained, however, when 56 MBTI scores are treated as continuous data (Stricker & Ross, 1963). The stability of MBTI scores over time has also been generally confirmed. Test-retest reliabilities over a 14 month period using MBTI continuous scores yielded correla- tions of approximately .70 for the E - I, S — N, and J - P scales (Stricker & Ross, 1964a). Although the T - F scale again showed a lower correlation of .48 in the Stricker & Ross (1964a) study, a more recent work by Levy pp 31. (1972) found test-retest correlations of .82 and .73 for the T - F scale over a two-month period. After reviewing all reliability data, Mendelsohn (1965) has concluded that the reliability of the MBTI compares favorably with similar self-report inventories. Evidence which supports the construct validity of the MBTI is found in a study by Bradway (1964). Bradway had twenty—eight Jungian analysts classify themselves according to Jung's psychological types. Bradway then compared the analysts' self-typing with their scores on the MBTI. Results showed 100% agreement between the analysts' self-typing and MBTI scores on the E — I dimen- sion; 68% agreement on the S — N dimension; 61% agreement on the T - F dimension; and, 43% agreement on all three dimensions. These findings thus suggest that the MBTI is a valid measure of psychological type. Additional evidence for the construct validity of the .NBTI has been obtained by comparing MBTI dimensions with 57 the same dimensions on the Gray—Wheelwright Psychological Type Questionnaire (Gray& Wheelwright, 1946). Like the MBTI, the Gray-Wheelwright inventory was developed to measure Jungian personality types. The E-I, S-N, and T-F scales are included in this instrument although no pro- visions are made for the J-P scale. Correlations between MBTI and Gray-Wheelwright scales using the continuous scores of a sample of 47 male college students can be found in Table 3.3. Table 3.3 Correlations Between MBTI and Gray-Wheelwright Psychological Type Questionnaire for 47 Male College Students Gray—Wheelwright MBTI Scale Scale E-I S-N T-F J-P E—I .79** .00 —.37** -.16 S-N -.24 .58** .15 .41** T-F -.20 .17 .60** .33* SOURCE: Stricker and Ross (1964b) *p=.05 **p=.01 It can be seen from Table 3.3 that the E-I scales correlate .79, the S-N scales correlate .58, and the T-F scales correlate .60 (Stricker & Ross, 1964b). It would seem then that the MBTI and the Gray-Wheelwright are essentially measuring the same type dimensions. 58 The validity of the MBTI has been further demonstrated by correlating MBTI scores with a wide variety of standard— ized measures - Minnesota Multiphasic Personality Inventory (MMPI), California Psychological Inventory (CPI), Maudsley Personality Inventory (MPI), Personality Research Inventory (PRI), Edwards Personal Preference Schedule (EPPS), Allport-Vernon-Lindzey Study of Values (AVL) and the Strong- Vocational Interest Blank (SVIB). The E-I scale has been found to correlate highly with the Social Introversion scale of the MMPI and the Extraversion scale of the MPI (Stricker & Ross, 1964b). The E—I scale seems to measure the degree of ease in and liking for social situations rather than the Jungian notion of an interest in external as opposed to internal events. The S—N scale shows moder— ate correlations with measures of independence and flexibility on the CPI and EPPS (Stricker & Ross, 1964b; Myers, 1962, respectively). The S—N scale is also nega- tively correlated with the SVIB business detail and business contact scales (Stricker & Ross, 1964b). These relationships suggest that the S—N scale taps an interest in tangible things versus an interest in abstract ideas. The T—F scale has been found to correlate positively with the Theoretical values scale of the AVL (Myers, 1962) and negatively with the Nurturance and Affiliation scales of the EPPS (Myers, 1962). This would indicate that Feeling types show an interest in humanistic values while Thinking types display an impersonal, logical approach to life. 59 Finally, the J—P scale yields positive correlations with both the Flexibility scale of the CPI (Stricker & Ross, 1964b) and the Order scale of the EPPS (Myers, 1962). The items on the J—P scale seem to reflect either a preference for organized and planned activity or spontaneous and novel activity. Positive correlations with a rating of Has No Self-Understanding (Stricker & Ross, 1964b) and with the PRI Self-Insight scale (Myers, 1962) suggest that the J—P scale also measures an awareness of phenomena. In summary, the MBTI has shown high reliability for three of its four scales. Only the T—F scale has produced relatively low reliability estimates. The construct validity of the MBTI mainly rests on the test's associa- tion with other standardized measures. Moderate to high correlations have been found which attest to the general validity of the instrument. Nonetheless, additional validation studies are called for, particularly with regard to the validity of the scales' conceptual definitions. Similarity Dimension Used in the Study: While a typological approach to personality assessment was used in the current study, some difficulties were encountered which discouraged the use of a global concept of typology. In the screening of therapists from the three clinics prior to the initiation of the research project, it became clear that the therapists did not show a balanced 60 distribution across the sixteen MBTI type categories. In fact, most of the therapists displayed similar type preferences in favor of "NF" types (Appendix M). These findings were consistent with the data found by previous investigators (Myers, 1962; Stricker & Ross, 1964b) in which a preponderance of "NF" types were found among psychologists and other individuals in the helping professions. A perusal of the MBTI types for the thera- pist population, however, suggested that the Judging— Perceiving dimension showed a striking variability across the population of therapists. The J—P dimension was the only dimension which seemed to show differences which could adequately define two distinct therapist groups. It was for this reason that the Judging—Perceiving dimension was chosen as the measure of therapist—client similarity/ dissimilarity. Since the J-P dimension is a general dimension which assesses various ways of dealing with the outside world, the scale seemed well suited to delineate personality differences among therapists and clients. In fact, the underlying premise of the MBTI itself is that people differ in a systematic fashion in the way in which they use perception and judgment. In the words of Myers (1962): "Perception" is here understood to include the processes of becoming aware, — of things or people or occurrences or ideas. "Judgment" is understood to include the processes of coming to conclusions about what has been perceived. If people differ systematically in what they perceive and the conclusions 61 they come to, they may as a result show corresponding differences in their reactions, in their interests, values, needs and motiva- tions, in what they do best and in what they like best to do. (p. 1) It would appear then that the Judging-Perceiving dimension most clearly represents the essence of the MBTI and, in addition, typological theory. Furthermore, the J-P scale's moderate correlations with the other MBTI scales suggests an interrelationship with the other dichotomous dimensions. Intercorrelations between MBTI type cate- gories can be found in Appendix N. It can be seen from Appendix N that Stricker and Ross (1963) found significant intercorrelations of .31 for J-P/S—N; .14 for J-P/E—I; and, .14 for J-P/T-F. This relatedness thus further facilitates a differential classification of therapists and clients based on the Judging-Perceiving dimension. A doctoral dissertation by Carlyn (1976) outlines the major characteristics of the J—P dimension. The J—P preference seems to reflect a person's idiosyncratic way of dealing with the world. The Judging individual attempts to regulate and control life in a systematic, organized fashion. Such an individual tends to reach conclusions quickly, deciding what needs to be done and then accom— plishing the needed tasks. The Judging person reaches these conclusions by shutting off perception temporarily. Conversely, the Perceiving individual delays judgment to allow more evidence to accumulate. Such a person tends to be open—minded and flexible and tries to understand 62 and adapt to life. The Perceiving individual thus lives a spontaneous, curious existence. It can be speculated that a Judging client would respond to a therapist who could meet a need for quick solutions to problems. Such a client would probably want to know what has to be done to solve such problems and how to go about doing it. A long—term treatment process probably would not satisfy this client. The Perceiving client, on the other hand, would most likely gravitate towards a therapist who would allow a free exploration of alternatives. Since Perceiving clients wish to under— stand their existence, they will tend to dwell on their concerns in order to seek enlightenment. A long—term, dynamic form of psychotherapy would probably be most appealing to such clients. It can be further hypothesized that Judging thera— pists, like Judging clients, would show a similar need to come to quick solutions to the problems of their clients. They would most likely display little patience for a dynamic, insight-oriented form of psychotherapy. Perceiving therapists, in contrast, would probably delay solutions to problems in the search for various alternative courses of action. Such therapists would most likely thrive on long—term, intensive therapy cases. The above formulations suggest that the Judging- Perceiving dimension is a significant variable with applicability for personality similarity research. While _¥— 63 some researchers (Mendelsohn & Geller, 1963; Mendelsohn, 1966; Jones, 1968) have utilized a global MBTI index to measure personality similarity/dissimilarity (i.e., summation of therapist—client difference scores for the four MBTI dimensions), the present study favored the use of a discrete personality dimension (J-P) to assess similarity. It was felt that a global index of similarity might serve to obscure differences which seemed to exist along the Judging-Perceiving dimension. Barrett-Lennard Relationship Inventory (BLRI) Several prominent psychologists have indicated the importance of a positive therapist-client relationship for therapeutic success (Rogers, 1961; Patterson, 1966; Kell & Mueller, 1966). Relationship or "process" variables are seemingly an essential part of the psychotherapy experience. The Barrett-Lennard Relationship Inventory (Barrett-Lennard, 1962) was used in the present study to measure several “process" variables under conditions of therapist-client personality similarity/dissimilarity. The BLRI was chosen as it: 1) has been used in similar psychotherapy studies to measure relationship variables (Jones, 1968; Tosi, 1969; McNally, 1972); 2) has shown good empirical validity; and 3) permits a relationship to be viewed from two different perspectives (therapist and client). The Barrett-Lennard Relationship Inventory is a self—report inventory which measures an individual's 64 perception of various dimensions of the therapeutic relationship. The test consists of 92 items which purport to measure the following process variables: 1) Empathic Understanding, 2) Level of Regard, 3) Unconditionality of Regard, 4) Congruence, and 5) Willingness To Be Known. Each_test item is rated in the affirmative or negative and given weights of +3, +2, +1, —1, —2, or -3 according to the rater's strength of feeling about the relationship. Scores for the five variables are then summed to yield a Total Relationship score. Positive scores indicate a good therapeutic relationship while negative scores suggest a poor relationship. The BLRI is available in four parallel forms. The two Client forms are identical except for the gender of the third person pronouns (i.e., corresponding to male and female therapists). The two Therapist forms differ from the Client forms in that the positions of the first- person and third-person pronouns are reversed (e.g., "He is interested in me" becomes "I am interested in him"). In this way, the therapy relationship may be viewed from different vantage points using essentially the same test items. A modified version of the BLRI was used in the present study according to specifications by Barrett—Lennard (1962). This version differs from the original BLRI only in that the Willingness To Be Known scale is omitted. The questionable validity of this scale prompted Barrett- 65 Lennard to modify his instrument. The present version of the BLRI thus contains the original four scales with 69 test items. These four scales are defined below: Empathic Understanding (EU): The EU scale reflects the degree to which one person is conscious of the immediate awareness of another individual. It is an active process of reaching out to an individual, of understanding a person's feelings, concerns and attitudes. While this awareness and understanding is usually generated on a verbal level, other aspects of awareness are equally significant (e.g., non—verbal). Level of Regard (LR): The LR scale can be conceived as the totality of affective expression of one person for another. It includes both positive and negative feeling reactions. Positive feelings may include love, liking, and appreciation while negative feelings may include hate, dislike, and impatience. Unconditionality of Regard (UR): The UR scale measures the amount of variability of one person's affective response to another. The scale reflects the degree of constancy of regard which is felt by one person in a relationship. Congruence (Cg): Barrett—Lennard (1962) defines congruence as "the degree to which one person is functionally integrated in the context of his relationship with another, such that there is absence of conflict or inconsistency between his total experience, his awareness, and his overt 66 communication...." (p. 4). The positive aspect of this concept suggests that a person's verbal communication will be consistent with non—verbal expressions (e.g., gestures, tone of voice). Reliability and Validity of the'BLRI: Information concerning the reliability and validity of the BLRI is chiefly based on the work of its author, G.T. Barrett— Lennard (1962). While some studies have attempted to replicate Barrett-Lennard's findings (McWhirter, 1973; Wiebe & Pearce, 1973), the most comprehensive reliability and validity data appears in Barrett—Lennard's original monograph. In the discussion that follows, therefore, Barrett—Lennard will be primarily cited with regard to the reliability and validity of the BLRI. Using therapist and client BLRI scores obtained at a university counseling center, Barrett—Lennard (1962) reported split—half reliability coefficients to be above .80 for each BLRI scale. Table 3.4 shows corrected reliability estimates of the BLRI scales taken from data collected after five therapy sessions. It can be seen from the table that the scales display good internal reliability. Satisfactory internal reliabilities were also found in a study by Wiebe and Pearce (1973). These researchers obtained internal consistency reliabilities of .83 for the Level of Regard scale, .64 for the Empathic Understanding 67 scale, .80 for the Congruence scale, .73 for the Uncon- ditionality of Regard scale, .76 for the Willingness To Be Known scale, and .93 for the Total Relationship scale. While these correlations are consistently lower than those obtained by Barrett-Lennard, they nonetheless suggest that the BLRI possesses sufficient internal reliability. Table 3.4 Corrected Split-Half Reliability Coefficients of the Barrett-Lennard Relationship Inventory Scales, from Data Gathered after Five Therapy Sessions Client Data Therapist Data BLRI Scale (N=42) (N=40) Level of Regard .93 .93 Empathic Understanding .86 .96 Congruence .89 .94 Unconditionality of Regard .82 .92 Willingness To Be Known .82 .88 SOURCE: Barrett-Lennard (1962) Since therapist-client relationships are susceptible to fluctuation over time, Barrett-Lennard (1962) used college students to obtain BLRI test-retest reliabilities. Each of 36 college students was asked to rate a close, stable personal relationship using the BLRI. After a four-week time period, students were again asked to rate the same relationship. Test-retest correlations showed 68 the following r's: Level of Regard, .84; Empathic Understanding, .89; Congruence, .86; Unconditionality of Regard, .90; Willingness To Be Known, .78; Total Relationship, .95. The BLRI scales thus seem to show good stability over time. The validity of the BLRI has been supported in studies by Gross and DeRidder (1966) and Clark and Culbert (1965). Gross and DeRidder found all but one of the BLRI scales to be significantly related to client movement in counseling. Only the Level of Regard scale failed to show this relationship. Clark and Culbert, using a T-Group as their sample, found an improvement in self— awareness among group members who had entered mutually therapeutic relationships (measured by the BLRI) with the other group participants. The comprehensive work of Barrett-Lennard (1962) provides the main evidence for the validity of the BLRI. Barrett-Lennard assessed the content validity of the measure by having experienced judges rate the degree of relationship between a scale item and the variable or factor it was supposed to represent. With the exception of four scale items, there was perfect agreement among the judges in classifying items as positive or negative factor indicators. Using therapist and client BLRI data, Barrett— Lennard evaluated the construct validity of the BLRI by examining the test's relationship with other variables 69 that were seen as theoretically relevant. Two such variables were therapist ratings of client change and client self-descriptive data of pre-post-therapy change. These two indices of therapeutic success were combined and rank-ordered to form "more changed" and "less changed" groups. Results showed a highly significant difference between the "more" and "less“ changed groups after five therapy sessions on five of the six client BLRI scales. Only the Willingness To Be Known scale showed no signifi- cant differences between the two groups. Results were in the predicted direction with the "more changed" group showing the highest relationship scores. vThe findings for the therapist BLRI data showed a similar pattern after five therapy sessions, but at more moderate levels of significance. These results thus lend support for the construct validity of the BLRI. Gough—Heilbrun Adjective Check List (ACL) The measure of therapeutic outcome presented some difficulties in the present study. More widely used measures of pre-post-therapy gain such as the Minnesota Multiphasic Personality Inventory (Hathaway & McKinley, 1967) and the Tennessee Self—Concept Scale (Fitts, 1964) were considered either too time-consuming or too difficult to administer. Since clients received no remuneration for participating in the study, an instrument was needed which was quick, easy to administer, and motivating for 70 subjects while also exhibiting good psychometric properties. The Gough—Heilbrun Adjective Check List was such a measure. The ACL has repeatedly shown itself to be an objective, valid measure of personality variables (Masterson, 1975). The test takes only ten minutes to complete, directions are easy to comprehend, and the test fosters minimal test—taking resistance. The ACL consists of 300 adjectives which are commonly used to describe individuals (e.g., confident, trusting). The examinees are asked to check off those adjectives which describe them best (i.e., as they really are, not as they would like to be). Scores are obtained on 24 scales which include four indices of test-taking variables, four scales which measure personal adjustment and personal effectiveness, one scale which assesses counseling readiness, and 15 scales which purport to measure Murray's needs (Murray, 1938). Individual scores are usually converted to standard scores to control for the variable number of adjectives checked. Of the 24 ACL scales administered in the study, four scales were utilized in the statistical analyses to measure therapy gain — Number of Favorable Adjectives Checked, Number of Unfavorable Adjectives Checked, Self— confidence, and Personal Adjustment. Each of these scales will be discussed below: Number of Favorable Adjectives Checked (FAV): This scale reflects the positive nature of an individual's 71 self—description. Individuals who score high on this scale generally show a strong desire to do well, impress others and behave appropriately. People with low scores tend to be individualists who are headstrong and pleasure— seeking and who may experience anxiety and self—doubt. Number of Unfavorable Adjectives Checked (UNF): In contrast to the FAV scale, the UNF scale measures an individual's negative self-description. People Who score high on this scale show an impulsive lack of control over the negative aspects of their personality. They tend to be cynical, rebellious, and arrogant. Low scorers tend to be more complacent, more tactful and more obliging. Self—confidence (SCF): The Self—confidence scale reflects an element of dominance in a person's approach to others. High scorers tend to be persistent, assertive, active, and sociable. Such people are generally seen as self-confident, determined, and ambitious. In contrast, low scorers are seen as ineffective, passive, reserved, and preoccupied. Personal Adjustment (PAJ): The Personal Adjustment scale seems to measure an individual's approach to life rather than the degree of pathology. High scorers tend to be cheerful, optimistic, sociable, trusting, and dependable. Individuals who score low on this scale are seen as dissatisfied, moody, defensive, anxious, and worrying. 72 The above ACL scales were chosen as measures of therapy outcome because they assess characteristics that seem re- lated to therapeutic gain. The first two scales (Number of Favorable Adjectives Checked, Number of Unfavorable Adjectives Checked) seem to probe an individual's self- esteem and self—image. It would seem that clients who show growth as a result of psychotherapy would demonstrate an improved self-image with a higher level of self—esteem. Additionally, it would seem that clients would show greater confidence in their abilities and would tend to be more assertive in their dealings with others (Self-confidence scale). Finally, their attitude might be more positive and they might demonstrate a greater degree of trust of others (Personal Adjustment scale). These four scales thus seem to reflect those qualities which are indicative of therapeutic growth. Reliability and Validity of the ACL: Table 3.5 contains test—retest reliabilities of the four ACL scales used in the study (Gough & Heilbrun, 1965). After a 10-week period, college males showed correlations above .73 (range of .73 - .84) while college females displayed somewhat lower correlations (range of .64 — .79). In general, the ACL scales seem to possess adequate relia- bility over a lO-week period. Support for the construct validity of the ACL was obtained by Scarr (1966) and Scott and Day (1972). Scarr 73 factor analyzed the 24 ACL scales and found three factors of extroversion - introversion, social desirability, and personality traits associated with intelligence. Scarr correlated the ACL scales with other measures of per- sonality (e.g., TAT, Goodenough drawings) and found the ACL to be a valid measure of personality characteristics. Scott and Day likewise factor analyzed the 24 ACL scales and found three similar underlying factors (General Adjustment, Self—Assertiveness, and Ego Control). These factors were then correlated with the Basic Interest scales of the Strong Vocational Interest Blank. The researchers found the ACL factors to be related to the Basic Interest scales in a theoretically consistent and plausible manner. Table 3.5 Test—Retest Reliabilities for the Four Adjective Check List Scales over a Ten-Week Period College Males College Females ACL Scale (N=56) (N=23) Number Favorable Adjectives Checked .76 .67 Number Unfavorable Adjectives Checked .84 .77 Self-Confidence .73 .64 Personal Adjustment .76 .79 SOURCE: Gough and Heilbrun (1965) 74 Gough and Heilbrun (1965) give additional evidence to support the construct validity of the ACL scales by demonstrating the scales' relationship to other well— established measures. For example, the ACL Number of Favorable Adjectives Checked scale correlates —.40 with the Welsh A—scale (anxiety) on the MMPI while the ACL Number of Unfavorable Adjectives Checked scale correlates +.29 with the Welsh scale. The Self—confidence scale shows a high positive relationship with the CPI Dominance (+.57) and Sociability (+.47) scales and a strong negative relationship with the MMPI Social Introversion scale (—.47). Finally, the ACL Personal Adjustment scale correlates negatively with eight of the ten MMPI scales which measure psychopathology. Survey Questionnaire (SQ) The Survey Questionnaire was used as an additional measure of therapeutic outcome in the study to provide client self-ratings of therapy gain. The Survey Question- naire was adapted by the author from the Counseling Service Questionnaire, Form 75, a self—report scale utilized by the Michigan State University Counseling Center to assess the quality of counseling services (Michigan State University Counseling Center, 1975). The Michigan State Questionnaire is itself a modification of the Counseling Services Assessment Blank (Hurst & Weigel, 1968). The current Survey Questionnaire consists of six items 75 which were designed to assess client impressions of their psychotherapy experience. Questions were specifically formulated to measure a client's: 1) satisfaction with therapy, 2) ability to cope with life stresses, and 3) feelings about self. A four-point rating scale was utilized for all six questions (e.g., "not helpful", "somewhat helpful", "helpful", "very helpful"). Reliability and Validity of the Survey Questionnaire: The internal consistency of the Survey Questionnaire was determined in the study by computing Cronbach's coefficient alpha (Cronbach, 1967). Results of the reliability analysis showed an interitem consistency of .87. The Survey Questionnaire thus appears to possess high internal consistency. Validation of the Survey Questionnaire was attempted in the study by means of a correlational analysis of the rating scale with the other measures of therapy success (Adjective Check List, Barrett-Lennard Relationship Inventory). Results of the Pearson Product—Moment correla- tional analysis can be found in Table 3.6. Table 3.6 shows that the Survey Questionnaire correlated significantly with both the four ACL subscales and the BLRI Total Relationship scales. The strongest relationships were found between the Survey Questionnaire and the ACL Personal Adjustment scale (r = .43), and between the Survey Questionnaire and the client BLRI 76 Total Relationship scale (r = .40). These findings lend support for the validity of the Survey Questionnaire as a measure of therapeutic gain. Design A multivariate repeated measures design is employed in the present study. The independent variable is therapist type, with two levels (Judging therapists and Perceiving therapists). Client type, also with two levels (Judging clients and Perceiving clients), serves as the repeated measures factor. Judging and Perceiving subjects are operationally defined as those individuals who score toward either the Judging or Perceiving end of the dichoto- mous MBTI Judging — Perceiving dimension. Since each therapist in the study treated both Judging and Perceiving clients, the assumption of independence across the client groups was violated. A repeated measures design is thus used to control for this dependency across client types (Winer, 1962). Table 3.7 gives a pictorial representation of the repeated measures design used in the study. The employment of a repeated measures design in the study required that sum and difference variables be calculated for each of the fifteen dependent measures. The sum variables were used to test for therapist main effects, while the difference variables were used to test for client repeated measures effects and therapist type x client type (measures) interaction effects. 77 OHHMQGOflumwoo >m>nsm H Om wamom menu Isoflpmaom HMDOB “Snow pmflmmuoaev >HODQO>QH mangOHDMme UHMCQOQIDDOHHmm n HBIqumIQB wHoom mflamnOHpmem HMDOB Afinom pcmflaov >MOHQO>GH QflQmGOHuoamm pqusoqlppmuhwm H HBIqumIHU wamom pcmfipm5fl©< HMGOmHmm .pmflq Mowno m>flpowm©< H hem Hum meom mocwenmcoOImamm .nmnq Momno O>Hnowhe< u mom qua mawom pmxowno mm>HDOOmU< manmno>omcb mo HOQEDZ .Dmflq Momso w>Hpomfl©< u mZD £04 mamom pwxomnu mm>fluowflpm manmuo>mm mo Honfisz .pmflq Momnu w>HDUOm©¢ u > pampnmmwm Howe: map moo€4 muanOmemoo COflpmeHHOU DQOEOE I poopoum Gownmom m.m canoe 78 Table 3.7 Design of the Study Client Type (Measures Effect) Judging Perceiving Thl Th2 Judging - an: — Th e?” 6 P N=6 U) .a h T g 7 Th PI: 8 Perceiving — Th12 =6 N=12 79 Research Hypotheses Research hypotheses were formulated to examine therapist main effects, client repeated measures effects, and therapist—client interaction effects (i.e., effect of personality similarity/dissimilarity). The first two research hypotheses were stated in the null form while the third research hypothesis was given directionality in favor of the facilitative effect of personality similarity on therapy success. The following research hypotheses were tested: Hypothesis I: There will be no significant difference between Judging and Perceiving'therapists on the following measures of therapeutic process and outcome: a. Ratings of the therapy relationship by clients on the Barrett-Lennard Relationship Inventory subscales (BLRI-Client form) . Level of Regard (Cl-BLRI-LR) . Empathic Understanding (Cl-BLRI-EU) . Congruence (Cl-BLRI-Cg) . Unconditionality of Regard (Cl—BLRI-UR) . Total Relationship (Cl-BLRI-Tl) U'lIh-LONH b. Ratings of the therapy relationship by therapists on the Barrett—Lennard Relation— ship Inventory subscales (BLRI-Therapist form) 1. Level of Regard (Th-BLRI—LR) 2. Empathic Understanding (Th-BLRI—EU) 3. Congruence (Th-BLRI-Cg) 4. Unconditionality of Regard (Th-BLRI-UR) 5. Total Relationship (Th-BLRI-Tl) c. Pre-post-treatment gain scores obtained from client self-ratings on the Adjective Check List subscales 80 1. Number of Favorable Adjectives Checked (ACL FAV) 2. Number of Unfavorable Adjectives Checked (ACL UNF) 3. Self-confidence (ACL SCF) 4. Personal Adjustment (ACL PAJ) Ratings by clients on the Survey Question- naire (SQ) Hypothesis II: There will be no significant differ— ence between Judging and Perceiving clients on the following measures of therapeutic process and outcome: a. b. Ratings of the therapy relationship by clients on the Barrett-Lennard Relationship Inventory subscales (BLRI-Client form) Level of Regard (Cl-BLRI—LR) Empathic Understanding (Cl-BLRI—EU) Congruence (Cl-BLRI-Cg) . Unconditionality of Regard (Cl—BLRI—UR) . Total Relationship (Cl-BLRI-Tl) U'IAWNH 0 Ratings of the therapy relationship by therapists on the Barrett-Lennard Relation- ship Inventory subscales (BLRI-Therapist form) . Level of Regard (Th—BLRI-LR) Empathic Understanding (Th-BLRI-EU) Congruence (Th-BLRI-Cg) . Unconditionality of Regard (Th-BLRI—UR) . Total Relationship (Th-BLRI-Tl) U'libwwl‘ I Pre—post-treatment gain scores obtained from client self—ratings on the Adjective Check List subscales 1. Number of Favorable Adjectives Checked (ACL FAV) 2. Number of Unfavorable Adjectives Checked (ACL UNF) 3. Self-confidence (ACL SCF) 4. Personal Adjustment (ACL PAJ) Ratings by clients on the Survey Question— naire (SQ) 81 Hypothesis III: There will be a significant inter- action between therapist and client types based on the Judging - Perceiving dimension of the MBTI. That is, therapists and clients whose types are similar will score significantly higher on the following measures of thera- peutic process and outcome than therapists and clients whose types are dissimilar: a. Ratings of the therapy relationship by clients on the Barrett—Lennard Relationship Inventory subscales (BLRI-Client form) 1. Level of Regard (Cl-BLRI—LR) 2. Empathic Understanding (Cl—BLRI-EU) 3. Congruence (Cl—BLRI—Cg) 4. Unconditionality of Regard (Cl-BLRI—UR) 5. Total Relationship (Cl-BLRI-Tl) Ratings of the therapy relationship by therapists on the Barrett-Lennard Relation- ship Inventory subscales (BLRI-Therapist form) . Level of Regard (Th-BLRI-LR) Empathic Understanding (Th-BLRI—EU) Congruence (Th—BLRI-Cg) . Unconditionality of Regard (Th-BLRI-UR) . Total Relationship (Th-BLRI—Tl) U1I¥>WNH O. Pre-post—treatment gain scores obtained from client self—ratings on the Adjective Check List subscales 1. Number of Favorable Adjectives Checked (ACL FAV) 2. Number of Unfavorable Adjectives Checked (ACL UNF) 3. Self—confidence (ACL SCF) 4. Personal Adjustment (ACL PAJ) Ratings by clients on the Survey Question- naire (SQ) 82 Analysis Multivariate repeated measures analyses of variance were used to test the three research hypotheses. Four separate repeated measures MANOVAS were performed since the number of dependent variables (i.e., fifteen) in the study exceeded the number of SE in cells. All hypotheses were tested at the d = .05 level of significance. Additional analyses were performed to examine the effect of global measures of similarity/dissimilarity on therapeutic process and outcome. Cronbach's D and D2 statistics (Cronbach & Gleaser, 1953) were utilized to form the global indices of similarity based on therapist- client difference scores on the Myers—Briggs Type Indicator (D statistic defined as the sum of the differences between therapist and client dichotomous scores on each MBTI dimension; D2 statistic defined as the square root of the sum of the squared differences between therapist and client continuous scores on each MBTI dimension). Multiple regression analyses were employed to measure the strength of the capacity of the global similarity measures to predict success in therapy. Canonical correlation analyses were also used to determine the amount of common variance between the global similarity indices and the therapy success measures. Finally, bivariate plots were examined to assess the nature of this relationship between global personality similarity and measures of therapy process and outcome. 83 Summary The current investigation was conducted during an ll-month period at three outpatient community mental health centers in the western Michigan area. Twelve experienced psychotherapists and seventy-four of their clients served as subjects in the study. Therapists and clients were administered the Myers-Briggs Type Indicator prior to the start of therapy to determine personality similarity/dissimilarity. Personality similarity/ dissimilarity was operationally defined as therapist- client similarity or dissimilarity on the MBTI Judging - Perceiving dimension. At the conclusion of the third psychotherapy session, clients were administered the two outcome measures, the Gough—Heilbrun Adjective Check List (pre—post-treatment gain scores), and a Survey Questionnaire. Several psychotherapy process variables were also determined at this time on the basis of therapist and client ratings on the Barrett-Lennard Relationship Inventory. Data collection continued until each of six Judging and six Perceiving therapists saw a minimum of two similar and two dissimilar clients. A multivariate repeated measures design was used in the study with therapist type, employing two levels (Judging and Perceiving therapists), serving as the inde- pendent variable. Client type, also with two levels (Judging and Perceiving clients), was the repeated measures factor. A repeated measures design was utilized 84 to control for a dependency across client types. Research hypotheses were presented to examine therapist main effects, client repeated measures effects, and therapist- client interaction effects (i.e., effect of personality similarity/dissimilarity). Repeated measures MANOVAS were used to test all research hypotheses. Additional analyses were also performed to study the effect of MBTI global measures of personality similarity/disSimilarity on therapeutic process and outcome. CHAPTER IV ANALYSIS OF RESULTS The first section of this chapter contains the results of the statistical analyses relevant to the three research hypotheses. The second section of the chapter consists of a presentation of additional analyses in which global per- sonality measures were employed to assess therapist-client similarity/dissimilarity (See Chapter III, Instrumentation section). The latter analyses were performed in order to make the current study consistent with previous similarity research (e.g., Mendelsohn & Geller, 1963; Jones, 1968). One-way repeated measures MANOVAS were used to test the three research hypotheses. The independent variable in the study was therapist type, with two levels (Judging therapists, Perceiving therapists), while client type, also with two levels (Judging clients, Perceiving clients), served as the repeated measures factor. In order to pre- vent the number of dependent variables (i.e., fifteen) in the study from exceeding the number of SE in cells, four repeated measures MANOVAS were performed each using one of the following four sets of dependent variables: 85 86 1. Four client BLRI subscales (Level of Regard, Empathic Understanding, Congruence, Unconditionality of Regard) 2. Four therapist BLRI subscales (Level of Regard, Empathic Understanding, Congruence, Unconditionality of Regard) 3. Therapist and client BLRI Total Rela- tionship scales and the Survey Questionnaire 4. Four ACL subscales (Number of Favorable Adjectives Checked, Number of Unfavorable Adjec- tives Checked, Self—confidence, Personal Adjustment) The use of a repeated measures design in the study required that sum and difference variables be calculated for each of the fifteen dependent measures. The means for Judging and Perceiving therapists on these variables are reported in Appendix P. In the hypotheses that follow, sum variables were used to test for therapist main effects (Hypothesis I), and difference variables were used to test for client repeated measures effects (Hypothesis II) and therapist type x client type (measures) interaction effects (Hypothesis III). Statistical significance was set at the a = .05 level for all analyses. 87 Hypothesis I: Therapist EffeCt The first hypothesis was formulated as follows: There will be no significant difference between Judging and Perceiving therapists on the following measures of thera- peutic process and outcome: a. Ratings of the therapy relationship by clients on the Barrett-Lennard Relationship Inventory subscales (BLRI-Client form) . Level of Regard (Cl-BLRI-LR) . Empathic Understanding (Cl-BLRI—EU) . Congruence (Cl-BLRI-Cg) . Unconditionality of Regard (Cl-BLRI-UR) . Total Relationship (Cl-BLRI-Tl) UlanNH b. Ratings of the therapy relationship by ’therapists on the Barrett-Lennard Relation- ship Inventory subscales (BLRI-Therapist form) 1. Level of Regard (Th-BLRI-LR) 2. Empathic Understanding (Th-BLRI-EU) 3. Congruence (Th-BLRI—Cg) 4. Unconditionality of Regard (Th-BLRI—UR) 5. Total Relationship (Th-BLRI-Tl) c. Pre-post-treatment gain scores obtained from client self—ratings on the Adjective Check List subscales 1. Number of Favorable Adjectives Checked (ACL FAV) 2. Number of Unfavorable Adjectives Checked (ACL UNF) 3. Self—confidence (ACL SCF) 4. Personal Adjustment (ACL PAJ) d. Ratings by clients on the Survey Question- naire (SQ) MANOVA results failed to show a significant difference between Judging and Perceiving therapists on any of the four sets of dependent variables. Significance was not found for the four client BLRI subscales [F (4, 7) = 88 .60, p < .67]; the four therapist BLRI subscales [F (4, 7) .90, p < .SI]; the therapist and client BLRI Total Rela— tionship scales and Survey Questionnaire [F (3, 8) = 1.56, p < .27]; and, finally, the four ACL subscales [T (4, 7) = .74, p < .59]. Hypothesis I was thus not rejected. MANOVA results for the therapist effect on the four sets of dependent variables are reported in Tables 4.1 - 4.4. Table 4.1 Multivariate Analysis of Variance with a Repeated Measures Factor (Clients) for the Barrett-Lennard Subscales (Client Form) - Level of Regard, Empathic Understanding, Congruence, Unconditionality of Regard Source of Variation df F Value P less than Therapist Effect 4, 7 .6004 .6746 Client Effect (Measures) 4, 7 2.3230 .1558 Therapist x Client (Measures) Interaction 4, 7 .6498 .6450 Table 4.2 Multivariate Analysis of Variance with a Repeated Measures Factor (Clients) for the Barrett-Lennard Subscales (Therapist Form) - Level of Regard, Empathic Understanding, Congruence, Unconditionality of Regard Source of Variation df P Value P less than Therapist Effect 4, 7 .9037 .5105 Client Effect (Measures) 4, 7 .2286 .9139 Therapist x Client (Measures) Interaction 4, 7 .5966 .6769 89 Table 4.3 Multivariate Analysis of Variance with a Repeated Measures Factor (Clients) for the Barrett-Lennard Total Relationship Scale (Client Form), Barrett-Lennard Total Rela- tionship Scale (Therapist Form), and Survey Questionnaire Source of Variation df F Value P less than Therapist Effect 3, 8 1.5577 .2736 Client Effect (Measures) 3, 8 .5493 .6626 Therapist x Client (Measures) Interaction 3, 8 .1596 .9206 Table 4.4 Multivariate Analysis of Variance with a Repeated Measures Factor (Clients) for the Adjective Check List Subscales - Number of Favorable Adjectives Checked, Number of Unfavorable Adjectives Checked, Self-confidence, Personal Adjustment Source of Variation df F Value P less than Therapist Effect 4, 7 .7409 .5933 Client Effect (Measures) 4, 7 1.1534 .4064 Therapist x Client (Measures) Interaction 4, 7 .7719 .5766 Hypothesis II: Client Repeated Measures Effect The second hypothesis was formulated as follows: There will be no significant difference between Judging and Perceiving clients on the following measures of therapeutic process and outcome: a. Ratings of the therapy relationship by clients on the Barrett-Lennard 90 Relationship Inventory subscales (BLRI- Client form) Level of Regard (Cl-BLRI-LR) Empathic Understanding (Cl-BLRI-EU) Congruence (Cl-BLRI-Cg) Unconditionality of Regard (Cl-BLRI—UR) . Total Relationship (Cl—BLRI—Tl) U1~l>LONI-‘ 0'. b. Ratings of the therapy relationship by therapists on the Barrett-Lennard Relation- ship Inventory subscales (BLRI-Therapist form) 1. Level of Regard (Th-BLRI-LR) 2. Empathic Understanding (Th—BLRI-EU) 3. Congruence (Th-BLRI-Cg) 4. Unconditionality of Regard (Th-BLRI—UR) 5. Total Relationship (Th—BLRI-Tl) c. Pre—post—treatment gain scores obtained from client self—ratings on the Adjective Check List subscales 1. Number of Favorable Adjectives Checked (ACL FAV) 2. Number of Unfavorable Adjectives Checked (ACL UNF) 3. Self-confidence (ACL SCF) 4. Personal Adjustment (ACL PAJ) d. Ratings by clients on the Survey Questionnaire (SQ) The MANOVAS also failed to show a significant differ— ence between Judging and Perceiving clients on any of the four sets of dependent variables. Significant results were not observed on the four client BLRI subscales [F (4, 7) = 2.32, p < .16]; the four therapist BLRI subscales [F (4, 7) = .23, p < .9I]; the therapist and client BLRI Total Relationship scales and Survey Questionnaire [F (3, 8) = .55, p < .66]; and, the four ACL subscales [F (4, 7) = 1.15, p < .41]. Hypothesis II was, therefore, 91 also not rejected. Tables 4.1 - 4.4 contain the MANOVA results for the client repeated measures effect on the four sets of dependent variables. Hypothesis III: Therapist — Client Interaction Effect The third hypothesis was formulated as follows: There will be a significant interaction between therapist and client types based on the Judging - Perceiving dimen- sion of the MBTI. That is, therapists and clients whose types are similar will score significantly higher on the following measures of therapeutic process and outcome than therapists and clients whose types are dissimilar: a. Ratings of the therapy relationship by clients on the Barrett-Lennard Relation— ship Inventory subscales (BLRI-Client form) 1. Level of Regard (Cl-BLRI—LR) 2. Empathic Understanding (Cl-BLRI-EU) 3. Congruence (Cl-BLRI-Cg) 4. Unconditionality of Regard (Cl—BLRI—UR) 5. Total Relationship (Cl-BLRI-Tl) b. Ratings of the Therapy relationship by therapists on the Barrett-Lennard Relation- ship Inventory subscales (BLRI-Therapist form) 1. Level of Regard (Th-BLRI-LR) 2. Empathic Understanding (Th-BLRI-EU) 3. Congruence (Th-BLRI-Cg) 4. Unconditionality of Regard (Th-BLRI-UR) 5. Total Relationship (Th-BLRI-Tl) c. Pre-post—treatment gain scores obtained from client self-ratings on the Adjective Check List subscales 1. Number of Favorable Adjectives Checked (ACL FAV) 2. Number of Unfavorable Adjectives Checked (ACL UNF) 92 3. Self-confidence (ACL SCF) 4. Personal Adjustment (ACL PAJ) d. Ratings by clients on the Survey Questionnaire (SQ) MANOVA results for Hypothesis III are also found in Tables 4.1 - 4.4. As can be seen from the tables, no significant therapist type x client type interaction was observed on any of the four sets of dependent variables. That is, a significant therapist type x client type inter- action was not found for the four client BLRI subscales [F (4, 7) = .65, p < .65]; the four therapist BLRI sub- scales [F (4, 7) = .60, p < .681; the therapist and client BLRI Total Relationship scales and Survey Questionnaire [F (3, 8) = .16, p < .92]; and, finally, the four ACL subscales [F (4, 7) = .77, p < .58]. Similar and dissimilar therapeutic dyads based on the Judging - Perceiving dimen- sion of the MBTI showed no significant difference on the measures of therapy process and outcome. No support was thus obtained for the main hypothesis of this study (Hypothesis III). Global Measures of Personality Similarity While the above analyses employed the dichotomous MBTI Judging-Perceiving dimension to assess personality similarity/dissimilarity, MBTI global similarity indices were also used to determine the relationship between therapist-client personality similarity and therapy success. Global similarity was determined through the use 93 of Cronbach's D and D2 techniques (Cronbach & Gleaser, 1953) on therapist-client difference scores for the four MBTI dimensions, Extraversion-Introversion (E-I), Sensing- Intuition (S-N), Thinking-Feeling (T—F), and Judging- Perceiving (J-P). For these analyses, the D technique utilized the sum of the differences between therapist and client dichotomous scores on each MBTI dimension while the D2 technique employed the square root of the sum of the squared differences between therapist and client continuous scores on each MBTI dimension. Therapy success was again measured by the BLRI therapist and client Total Relation- ship scales, the Survey Questionnaire, and the four ACL subscales. Multiple regression analyses were performed to obtain estimates of the degree of relationship between the independent variables (D or D2 global similarity indices) and dependent variables (measures of therapy success) and to measure the strength of the capacity of these indepen- dent variables to predict the dependent measures. When the D global index was used as the independent variable, the multiple regression analysis failed to show a significant relationship between the D similarity index and the dependent variables measuring therapy success [F (7, 66) = 1.22, p <_.3I]. Statistics for the regression analysis are reported in Table 4.5. 94 Table 4.5 Multiple Regression Analysis and Canonical Correlation Analysis Using Therapist-Client MBTI D-Scores to Assess Global Personality Similarity F Value for Multivariate Test = 1.2166 df = 7, 66 P less than .3064 2 Multiple F P less Step P less Variable R R Value than Down F than Cl-BLRI-Tl .0014 .0376 .1018 .7507 .1018 .7507 Th-BLRI-Tl .0086 .0927 .6246 .4320 .5304 .4689 so .0137 .1171 1.0009 .3205 .6322 .4293 ACL FAV .0001 .0120 .0103 .9194 .1504 .6994 ACL UNF .0089 .0945 .6484 .4234 .6780 .4132 ACL SCF .0041 .0639 .2952 .5886 .3058 .5822 ACL PAJ .0352 .1877 2.6300 .1093 5.9894 .0171 Canonical Correlation = .3381 Test of Significance of Canonical Correlation: P < .3058 Total Percentage of Variation in Dependent Variables Accounted For = 1.6327 It can be seen from Table 4.5 that little total variance was accounted for through the use of the D global similarity index. When R2 is treated as an indicator of the percentage of the variance accounted for by a variable, it can be seen that the BLRI client Total Relationship scale accounted for only . % of the total variance of therapy success. Likewise, the BLRI therapist Total Rela— tionship scale only accounted for . % of the total variance. Finally, the remaining dependent variables, the Survey 95 Questionnaire and the ACL subscales, also accounted for little of the total variance of therapy success (Table 4.5). 2 The regression analysis for the D global similarity index showed results similar to those found for the D global index. No significant relationship was found between the 2 D similarity measure and the variables of therapy success [F (7, 66) = 1.33, p < .23]. Results of this regression analysis appear in Table 4.6. Table 4.6 Multiple Regression Analysis and Canonical Correlation Analysis Using Therapist-Client MBTI Dz-Scores to Assess Global Personality Similarity F Value for Multivariate Test = 1.3290 df = 7, 66 P less than .2509 Multiple F P less Step P less Variable R2 R Value than ‘(Down F than Cl-BLRI-Tl .0012 .0339 .0829 .7742 .0829 .7742 Th-BLRI-Tl .0028 .0527 .2008 .6555 .1485 .7011 SQ .0105 .1023 .7521 .3856 .5581 .4576 ACL FAV .0039 .0623 .2802 .5982 .0905 .7645 ACL UNF .0135 .1164 .9882 .3236 .6965 .4069 ACL SCF .0072 .0848 .5217 .4725 .1293 .7204 ACL PAJ .0664 .2577 5.1197 .0267 7.4721 .0081 Canonical Correlation = .3515 Test of Significance of Canonical Correlation: P < .2504 Total Percentage of Variation in Dependent Variables Accounted For = 1.7648 96 Examination of R2 indicates that little total variance was accounted for by using the D2 index to predict therapy gain. The BLRI client Total Relationship scale again only accounted for .1% of the total variance of therapy success. Similarly, the BLRI therapist Total Relationship scale accounted for only . % of the total variance. Furthermore, the Survey Questionnaire and ACL subscales also accounted for little of the total variance of therapy success (Table 4.6). The two multiple regression analyses thus failed to show a significant relationship between either of the two global similarity indices, i.e., D or D2, and the measures of therapy process and outcome. An additional correlational analysis, however, further explored the nature of the relationship between these variables. Canonical correla- tions were employed in order to maximize the amount of common variance between the independent variables (D or D2 similarity indices) and dependent variables (therapist, client BLRI Total Relationship scales, the Survey Questionnaire, the four ACL subscales). The canonical correlation between the D similarity index and the measures of therapy success was found to be .34, accounting for only 1.6% of the variation in dependent variables. The test of significance of this canonical correlation revealed a nonsignificant p value of less than .31 (X2 = 8.31, d.f. = 7). The D2 similarity index and therapy success measures showed a canonical correlation of 97 .35, again accounting for only 1.8% of the therapy success variation. This correlation was also not significant, with a p value of less than .25 (x2 = 9.03, d.f. = 7). Results of the canonical correlation analyses for the D and D2 similarity indices are presented in Tables 4.5 and 4.6 respectively. Since multiple regression and canonical correlation analyses failed to show a significant relationship between the global similarity indices and the measures of therapy success, the possibility was explored that one or more of the individual MBTI dimensions would show a relationship between similarity and therapy gain. A step—wise regression analysis was thus employed to analyze the contribution of each of the four MBTI dimensions (E—I, S—N, T—F, J-P) in predicting therapy success. The covariates in the regres- sion analysis were each of the four MBTI dimensions, obtained from therapist-client continuous difference scores on each dimension. The dependent variables were again the measures of therapy process and outcome, i.e., therapist and client BLRI Total Relationship scales, the Survey Questionnaire, and the four ACL subscales. Results of the step—wise regression analysis for the individual MBTI dimensions are found in abbreviated form in Table 4.7 and in complete form in Appendix Q. Examination of the step down F statistics indicated that the first covariate, E-I Difference, was not significantly related to the measures of therapy success [F (7, 66) = .79, 98 p < .60]. The addition of the covariate, S—N Difference, to the regression equation also failed to produce a signifi- cant relationship between the independent and dependent variables [F (7, 65) = .53, p < .81]. The addition of the third covariate, T-F Difference, again added little to the prediction of therapy success [F (7, 64) = 1.18, p < .33]. Nonetheless, the addition of the fourth and final covariate, J—P Difference, to the regression equation significantly added to the predictive capacity of the regression equation [F (7, 63) = 2.41, p < .03]. Table 4.7 Step-wise Regression to Analyze the Contribution of Each of the Four MBTI Dimensions in Predicting Therapy Success % Additional Variance Step P less Covariate Accounted for Down F df than Covariate E-I Difference 5.6709 .7874 7,66 .6004 Covariate S—N Difference 8.2160 .5313 7,65 .8077 Covariate T—F Difference 7.2097 1.1823 7,64 .3256 Covariate J—P Difference 8.6633 2.4141 7,63 .0297 Since the Judging-Perceiving dimension was the only variable which appeared to be related to the dependent measures, an additional multiple regression analysis was performed using the J-P Difference dimension (continuous 99 J-P Scores) as the only independent variable. The depen- dent variables which correlated most highly with the J—P Difference dimension in earlier regression analyses were chosen as the dependent measures in the present regression analysis (i.e., *therapist BLRI Unconditionality of Regard scale; therapist BLRI Total Relationship scale; therapist BLRI Level of Regard scale; client BLRI Congruence scale; ACL Self—confidence scale). Table 4.8 Multiple Regression Analysis Using Therapist— Client MBTI Difference Scores on the J—P Dimension to Predict Therapy Success F Value for Multivariate Test = 1.3934 df = 5, 68 P less than .2378 Multiple F P less Step P less R Variable R Value than Down F than Th—BLRI—UR .0247 .1571 1.8228 .1813 1.8228 .1813 Th-BLRI-Tl .0195 .1396 1.4310 .2356 .0121 .9127 Th-BLRI-LR .0221 .1487 1.6287 .2060 1.1192 .2938 Cl-BLRI-Cg .0346 .1860 2.5805 .1126 1.5667 .2150 ACL SCF .0178 .1335 1.3070 .2568 2.3558 .1295 Results of the multiple regression analysis with the J—P Difference dimension as the sole independent variable are presented in Table 4.8. No significant relationship *Additional regression analyses had examined the rela- tionship of all the BLRI subscales with the various independent variables. 100 was found between the J—P Difference variable and the various measures of therapy success [F (5, 68) = 1.39, p < .24]. Further, examination of the step down F statis- tics indicated that none of the measures were significant when tested conditionally. The J-P Difference variable did not appear then to be a good predictor of therapy gain, as measured by the BLRI Total Relationship scales, the Survey Questionnaire, and the four ACL subscales. The above analyses thus suggest little linear associa- tion between the various therapist-client similarity procedures (D, D2, D-individual MBTI dimensions) and the measures of therapy process and outcome. A final statisti- cal technique, bivariate plots, was utilized to determine the nature of the relationship between these variables. As expected, a linear relationship was not observed. Furthermore, while some research studies have found a curvi- linear relationship between personality similarity and therapy success (See Chapter II), a visual inspection of the bivariate plots offered no evidence to support a curvilinear relationship. Summary Four one-way repeated measures MANOVAS were used to test the research hypotheses. Results of the tests of the research hypotheses were as follows: 1. In Hypothesis I it was stated that there would be no significant difference between Judging and Perceiving m‘x" f, - , .3. 101 therapists on the measures of therapy process and outcome. Hypothesis I was not rejected at the d = .05 level. Judging and Perceiving therapists did not significantly differ on any of the four sets of dependent measures. 2. In Hypothesis II it was stated that there would be no significant difference between Judging and Perceiving clients on the measures of therapy process and outcome. Hypothesis II was not rejected at the d = .05 level. The MANOVAS failed to show a significant client repeated measures effect on any of the four sets of dependent variables. 3. In Hypothesis III a significant interaction was predicted between therapist and client types based on the Judging-Perceiving dimension of the MBTI. That is, similar therapy dyads would score significantly higher on the depen- dent measures than dissimilar therapy dyads. MANOVA results revealed no significant therapist type x client type inter— actions at the d = .05 level, as similar and dissimilar dyads showed no significant difference on any of the four sets of dependent measures. No support was obtained for the main hypothesis of this study. Additional analyses were performed in which global similarity measures were used to determine therapist— client personality similarity/dissimilarity. Cronbach's D and D2 global similarity techniques were thus utilized on therapist and client MBTI total difference scores to examine the relationship between personality similarity and therapy 102 success. Multiple regression and canonical correlation analyses failed to find a significant relationship between the D and D2 global similarity indices and the dependent measures. D-scores on the individual MBTI dimensions also failed to show a relationship between personality similarity and therapy success. Finally, bivariate plots were examined to determine the nature of the relationship between the above variables. Neither a linear nor a curvilinear rela- tionship was found to exist between the measures of therapist-client personality similarity (D, D2, D-individual MBTI dimensions) and the measures of therapy process and outcome. CHAPTER V SUMMARY AND DISCUSSION The current investigation was an attempt to study the effect of therapist—client personality similarity on the process and outcome of short—term psychotherapy. Summary of the Study Research in psychotherapy has repeatedly demonstrated that the factors underlying constructive therapeutic change are quite complex. A plethora of research over the past twenty years has explored those variables that are felt to be crucial to both therapeutic process and outcome, e.g., client variables, therapist variables, client-therapist interaction variables, and technique variables. Each of these variables has at one time or another been shown to affect therapy process and/or outcome (Whiteley, 1967). A key factor which seems to supersede the importance of each of these variables, however, is the nature of the relationship between therapist and client. There appears to be a general consensus among psychologists that the therapist-client relationship is of major importance for the success of psychotherapy (Kell & Burow, 1970; Bordin, 1968). One important variable which many researchers 103 104 feel has a marked effect on the nature of this relation— ship is the similarity or dissimilarity of the personality structures of therapists and clients (Carson & Heine, 1962; Mendelsohn & Geller, 1963). It may then be hypo- thesized that the greater the degree of personality similarity between therapist and client, the more positive will be the therapeutic alliance and the more likely it will be that constructive therapeutic change will take place. Conversely, it may be hypothesized that the greater the degree of personality dissimilarity between therapist and client, the weaker will be the therapeutic alliance and the less likely it will be that constructive therapeutic change will take place. The purpose of the present study was to examine thoroughly this similarity/ dissimilarity hypothesis. Relevant literature was reviewed concerning the effect of personality similarity on therapy success. The findings of past research in this area were generally shown to offer contradictory results. In some studies, high similarity was associated with therapeutic success (Mendelsohn & Geller, 1963; Mendelsohn, 1966), while in other studies the relationship was found to be curvilinear, with medium similarity associated with therapeutic success (Carson & Heine, 1962; Mendelsohn & Geller, 1965). In addition, some studies have found either no relationship or a negative relationship between the same variables 105 (Carson & Llewellyn, 1966; Lichtenstein, 1966; Lesser, 1961; Bare, 1967). The employment of a wide diversity of methods, personality variables, client and therapist samples, and process and outcome measures in many of the studies makes it difficult to offer a meaningful comparison of findings. Specific limitations of some of these previous studies - failure to use in vivo research designs, use of inexperienced rather than experienced therapists, lack of distinction between vocational counseling and personal- social counseling, use of global measures of similarity rather than specific personality dimensions and, use of outcome measures of questionable validity - all were corrected in the present investigation. The study was conducted at three outpatient community mental health centers in the western Michigan area. The three clinics serve a similar client population of white, low-middle income blue-collar workers. The client sample was drawn from these clinics during an ll—month period from October, 1975 to August, 1976. Out of one hundred forty clients who volunteered to participate, seventy-four eventually comprised the client sample. Twelve experienced psychotherapists from these clinics served as the therapist sample. Therapist-client personality similarity/ dissimilarity was determined post hoc on the basis of therapist and client scores on the Judging-Perceiving dimension of the Myers-Briggs Type Indicator - Form F, 106 administered to both groups prior to the start of therapy. Therapists treated clients in a routine manner for a minimum of three therapy sessions. Dependent measures were obtained at the conclusion of this third session. Several psychotherapy process variables were evaluated by client and therapist ratings on the Barrett-Lennard Rela- tionship Inventory - Therapist and Client Forms. Thera- peutic outcome was determined by pre-post-treatmentgain scores on the Gough—Heilbrun Adjective Check List and by client post—treatment self—ratings on a Survey Questionnaire. Collection of data continued until each of the twelve sample therapists (six Judging, six Perceiving therapists) saw a minimum of two similar and two dissimilar clients. A multivariate repeated measures design was employed in the study.' The independent variable was therapist type, with two levels (Judging and Perceiving therapists), while client type, also with two levels (Judging and Perceiving clients), served as the repeated measures factor. Since therapists in the study treated both Judging and Perceiving clients, the assumption of inde- pendence across the client groups was violated. A repeated measures design was thus utilized to control for this dependency across client types. Three research hypotheses were tested using four multivariate repeated measures analyses of variance. Statistical significance was set at the a = .05 level. 107 Results of the tests of the research hypotheses follow below: 1. In Hypothesis I it was stated that there would be no significant difference between Judging and Perceiving ‘ therapists on the measures of therapy process and outcome. Hypothesis I was not rejected. Judging and Perceiving therapists did not significantly differ on any of the four sets of dependent measures. 2. In Hypothesis II it was stated that there would be no significant difference between Judging and Perceiving clients on the measures of therapy process and outcome. Hypothesis II was not rejected. MANOVA results showed no difference between Judging and Perceiving clients on any of the four sets of measures of therapy success. 3. In Hypothesis III it was predicted that there would be a significant interaction between therapist and client types based on the Judging—Perceiving dimension of the Myers-Briggs Type Indicator. That is, similar therapeutic dyads would score significantly higher on the dependent measures than dissimilar therapeutic dyads. MANOVA results showed no significant interactions as similar and dissimilar dyads displayed no difference on any of the four sets of dependent measures. No support was obtained for the main hypothesis of this study. Supplementary analyses explored the relationship between global measures of personality similarity/ dissimilarity and measures of therapy success. Cronbach's 108 D and D2 techniques, using therapist-client difference scores on the MBTI, were used to assess global personality similarity. The dependent measures were identical to those used in the original analyses. Multiple regression analyses failed to show significant predictive capacity for the measures of therapy success based on the global similarity procedures. Further, bivariate plots did not suggest either a linear or a curvilinear relationship between these variables. Discussion While the three research hypotheses failed to achieve statistical significance, some interesting findings or trends were nonetheless observed. A discussion of these findings follows below. Hypothesis I: Although tests of Hypothesis I showed no significant difference between Judging and Perceiving therapists, an examination of BLRI mean scores for these therapists revealed some intriguing data. For example, mean ratings by clients of the therapy relation- ship were consistently higher for Perceiving therapists than for Judging therapists. In contrast, however, Judging therapists typically gave higher BLRI mean ratings to their clients than did Perceiving therapists. These findings achieve greater clarity when the variable "therapist sex" is considered (See next section). Table 5.1 contains therapist and client BLRI mean scores 109 for Judging and Perceiving therapists. It should be noted that no apparent trends were observed for any of the other dependent measures in the study. Table 5.1 Client and Therapist Barrett-Lennard Relation- ship Inventory Mean Scores for Judging and PerceiVing~TheraPistsu'Munw.~-~g;.-. .............................................. Judging Therapists Perceiving Therapists Variable ..... (N=6) -. . , . (N=6) Cl-BLRI-LR 44.55 50.37 Cl-BLRI-EU 25.46 31.49 Cl-BLRI—Cg 41.36 49.40 Cl-BLRI-UR 28.57 38.54 Cl-BLRI-Tl 139.94 169.79 Th—BLRI-LR 51.31 45.27 Th-BLRI-EU 25.03 21.75 Th-BLRI-Cg 38.21 36.58 Th-BLRI-UR 16.89 14.81 Th-BLRI-Tl 131.44 118.41 "Therapist Sex"‘as an'Additional'Independent Variable: Since the above data suggested differences between Judging and Perceiving therapists, a supplementary analysis was employed to examine the contribution of the variable "therapist sex" to the relationship between personality similarity and therapeutic success. Previous research has shown that the sex of a therapist is an important variable affecting therapy outcome (McNair, Lorr, & Callahan, 1963). It was felt that the "therapist sex" variable might clarify some of the differences found between the two types of therapists. 110 Two-way repeated measures MANOVAS were performed in the supplementary analysis. Therapist type, with two levels (Judging and Perceiving therapists), and therapist sex served as the independent variables. Client type, also with two levels (Judging and Perceiving clients), was the repeated measures factor. Statistical significance was set at the a = .05 level. Four separate repeated measures MANOVAS were again used since the number of dependent variables exceeded the number of‘Ss in cells. The four repeated measures MANOVAS each employed one of the following four sets of dependent variables to measure therapy process and outcome: 1. BLRI Total Relationship scales - Therapist and Client Forms 2. Survey Questionnaire 3. ACL Number of Favorable Adjectives Checked scale and ACL Number of Unfavorable Adjectives Checked scale 4. ACL Self-confidence scale and ACL Personal Adjustment scale. Appendix R contains results of the MANOVAS for main effects, repeated measures effects, and interaction effects on the four sets of dependent variables. The only signifi- cant finding was an interaction effect for the client type x therapist sex interaction on the combined dependent variable, ACL Number of Favorable Adjectives Checked and ACL Number of Unfavorable Adjectives Checked [F (2, 7) = 5.04, P < .OE]. None of the other effects reached statistical 111 significance on any of the four sets of dependent measures. Since the client type x therapist sex interaction was the only significant finding for any of the effects tested, this appears to be a spurious finding rather than the result of some systematic relationship between the two variables. BLRI mean scores were then reexamined with additional consideration given to the variable "therapist sex". The lower BLRI mean ratings given by clients to Judging thera— pists (Table 5.1) can be explained by the fact that male Judging therapists received the lowest relationship ratings of all the therapist groups (Table 5.2). While female Judging therapists received the highest BLRI ratings from clients, these ratings were not high enough to raise the mean ratings of Judging therapists as a group. Consequently, the Judging therapists received lower overall BLRI mean ratings from clients than did the Perceiving therapists. Furthermore, the higher BLRI mean ratings given by Judging therapists as a group to their clients seem to be a function of the high ratings offered by the female Judging therapists (Table 5.2). These high ratings seem to have inflated the average for the Judging therapist group. It thus appears that the observed differences on the BLRI ratings between Judging and Perceiving therapists are a result of the variability in performance found for male and female Judging therapists. 112 mm.mmH Nh.NHH HBIHmHmISE Hm.HNH mm.wHH om.mH NH.MH hh.Hm mv.m MDIHmHmIQB mw.mm m4.wm mw.m4 om.4m monqumuse NH.om hm.mm ww.mm mm.HN DMIHmAmIQB mm.mv Ho.vv Hm.mm Hm.nw mHIHmHmIQB om.mmH mm.m>H mm.omH NH.mHH HBIqumIHU mm.mm mn.mm mw.mm mm.mm mDIHmHmIHU mm.m4 mm.mv Ho.Hm mm.om mouqumnHo mv.nm mm.mm Hmswm mm.mH DMIqumIHU HH.Hm mm.mv mH.Hm mn.mm mqumHmIHU Amnzv Amnzv ;;:xmuzv. nauzo 0Hannm> mHmHmewQB mumHmmswnH mpmHmmumna mpmHmwnmna mcH>Hmonmm maH>Hmoumm mgHmUSh mchosb onEmm mHmz mHmEmm mHmz ............................................. ............................ meHmemSB maa>amouwm onEwm Ucm ~mzflfinmoumm 0H0: .mchwdo meEwm .msHmUDb 0H0: How mmHoom cams anoucm>sH mHsmGOHHMHmm Unmasmqlupmunmm pmHmmquB can usmHHo m.m wHQme 113 The relatively meager performance of the male Judging therapists on the BLRI ratings by clients suggests that these therapists had some difficulty establishing positive therapeutic relationships with their clients. They apparently were unable to display the empathic understanding, unconditionality of regard, congruence, and other relationship variables which were assessed by the BLRI measure. Perhaps the explanation for this lies in the nature of the Judging personality itself. The Judging individual's need for control and order and a tendency to form quick judgments all may be antithetical to the patience, tolerance for ambiguity, and open- mindedness which is necessary for constructive interpersonal relationships. While female Judging therapists did not experience this same difficulty forming therapeutic relationships, this sex difference may be due to the influence of cultural sex role stereotypes which encourage the interpersonal responsiveness of females. The above findings thus have an obvious relevance for the area of therapist selection (i.e., which therapist types make the best therapists) and training (i.e., how can male Judging therapists be trained to facilitate interpersonal responsiveness). The reader should be cautioned, however, that this discussion has been based on nonsignificant statistical trends in the data which were observed on only the BLRI process measure. No such trends were evident on the other dependent measures. 114 Hypothesis II: Hypothesis II was also not rejected as Judging and Perceiving clients failed to show signifi- cant differences in any of the four repeated measures MANOVAS. A significant client type (measures) effect was, however, observed on the client BLRI Unconditionality of Regard scale when univariate repeated measures ANOVAS were employed [1 (l, 10) = 8.43, p < .05]. The independent variable in the univariate analyses was therapist type, with two levels (Judging and Perceiving therapists), while client type, also with two levels (Judging and Perceiving clients), again served as the repeated measures factor. The fifteen dependent measures used in earlier analyses were again employed as measures of therapy success. Results only showed significant findings for the client BLRI Unconditionality of Regard scale. While the client (measures) effect was significant on this scale at the .05 level, no significant results were found for either the therapist type effect or the therapist type x client type interaction. Results of the univariate repeated measures ANOVA for the client BLRI Unconditionality of Regard scale can be found in Appendix S. The significant client type effect on the client BLRI Unconditionality of Regard scale suggests that Judging clients feel a greater unconditionality or constancy of regard in their therapy relationships than do Perceiving clients, regardless of therapist type. It could be hypothesized then that Judging clients .._,_____ v ‘r . _ _, = 115 experience a greater feeling of comfort and security in psychotherapy than Perceiving clients, at least in the early stages of psychotherapy. This may be related to the characteristic tendency of Judging types to reach conclusions rather quickly (Myers, 1962). Such clients may make a quick determination of the value of psychotherapy and the suitability of their particular therapist and, having reached these conclusions, they may be able to experience a subjective feeling of acceptance in their therapy relationships. Perceiving clients, on the other hand, may delay judgment (Myers, 1962) of both the merits of psychotherapy and the suitability of their therapist. This greater cautiousness and reserve may prohibit such clients from experiencing constancy of regard in their therapy relationships during the initial stages of therapy (i.e., three therapy sessions). Whether or not Perceiving clients would experience the same unconditionality of regard at a later point in therapy remains to be determined. Hypothesis III: The failure to find support for Hypothesis III indicated that similar and dissimilar therapy dyads (assessed by the MBTI Judging-Perceiving dimension) did not significantly differ on the various measures of therapy success. This finding casts doubt on the main premise of the study, i.e., therapist-client personality similarity facilitates therapeutic process and outcome. In addition, the failure to find significant 116 results when MBTI global shmilarity indices were employed to assess personality similarity/dissimilarity further weakens the similarity hypothesis.“ Nonetheless, certain methodological issues releVant to the present study must be addressed before a definitive statement may be made. A discussion will therefore follow concerning sampling techniques used in the study, various procedural issues, and choice of outcome measures. The client and therapist samples used in the study appear to be a representative sample of the population of clients and therapists at three Michigan community mental health centers. Demographically and otherwise, these samples mirror the larger populations from which they were drawn. A question may be raised, however, as to the comparability of the samples across the three clinics. Since subjects from each clinic were pooled into either the client or therapist samples, homogeneity across the three clinics was imperative. Multivariate analyses of variance were therefore performed in order to compare the MBTI continuous scores of sample clients from the three clinics and, likewise, sample therapists from the same clinics. Statistical significance was set at the a = .05 level. Results of the MANOVAS are reported in Tables 5.3 and 5.4. Multivariate F—values showed no significant differences on the MBTI among clients from the three clinics [F (8, 136) = 1.74, p < .10] and no significant differences on the MBTI for therapists from the same clinics [F (8, 12) = 117 1.31, p < .32]. In addition, the univariate F - tests showed no significant differences on the MBTI Judging- Perceiving dimension across the three clinics for sample clients [F (2, 71) = .30 p < .75:lor for sample therapists [F (2, 9) = 1.19, p < .35]. The reSults of these analyses establish the comparability of the samples across the three clinics and thus seem to justify the pooling of clients and therapists from these clinics into one client sample and one therapist sample. Table 5.3 Multivariate and Univariate Analysis of Variance for Client MBTI Continuous Scores for the Three Clinics F Value for Multivariate Test = 1.7367 df = 8, 136 .P less than .0953 ,,,,,,,,, MBTI Hypothesis Mean Univariate F P less Step P less Dimension. 'Square.y.y ..d£f=727‘7l'.. than 'Down F than E - I 2008.7867 2.9847 .0570 2.9847 .0570 S - N 2870.2900 3.4182 .0383 2.7634 .0700 T — F 241.7110 .6596 .5203 1.2733 .2864 J - P 245.0302 .2953 .7453 .0535 .9480 NOTE: Clinic #1 (N=42) Clinic #2 (N=21) Clinic #3 (N=ll) 118 Table 5.4 Multivariate and Univariate Analysis of Variance for Therapist MBTI Continuous Scores for the Three Clinics F Value for Multivariate Test = 1.3139 df = 8, 12 ~Pilessgthan~y3232. MBTI Hypothesis Mean Univariate F P less Step P less Dimension Square. ~ - df =~2, 9 . . than, Down F than B — I 216.0000 .6287 .5552 .6287 .5552 S - N 3.3333 .0056 .9945 .1203 .8883 T - F 234.1667 4.2291 .0508 3.2948 .0981 J - P 941.6667. n . ~ ~l.1876 - - .3486 1.6321 .2717 NOTE: Clinic #1 (N=6) Clinic #2 (N=4) Clinic #3 (N=2) An additional question may be raised as to the compara- bility of the final client sample and the twenty-eight clients who terminated treatment prematurely (i.e., less than three therapy sessions). Perhaps the early termination of the latter group of clients represented the effects of similarity/ dissimilarity. A comparison of Appendices A and C reveals that the two client groups were comparable with regard to age, sex, race, employment, and diagnosis. The only variable which seemed to show a difference between the two groups was education level. That is, 69% of the final client sample were at least high school graduates or higher, while only 39% of the terminated client group attained 119 this level of education. In order to ascertain any MBTI type differences, a multivariate analysis of variance was performed comparing the two client groups on the MBTI Judging-Perceiving similarity measure (continuous J-P scores). MANOVA results showed that the two client groups did not significantly differ on the Judging-Perceiving dimension [F (l, 100) = .18, p < 68]. It would thus appear that the early termination of the twenty-eight clients was not due to the effects of similarity/dissimilarity as measured by the present similarity measure. A procedural issue which merits discussion is the method of assigning clients to therapists employed in the study. It was stated in Chapter II that the chief consideration for client assignment was therapist-client schedule compatibility. No attempt was made to assign clients selectively to particular therapists. It may be argued, however, that true randomization did not occur since a given therapist may have been either "over-booked" or "under-booked," thus predisposing him to a certain bias in client assignment. Nonetheless, this did not appear to be the case. Therapists generally had full case-loads with no apparent discrepancies in either new client intakes or in schedule availability. It would thus appear that essentially a random procedure was used in client assignment. An additional procedural issue concerns the pairing of therapists and clients into "similar" or "dissimilar" therapy dyads based on MBTI Judging-Perceiving scores. 120 Such pairings imply equivalent strength for the therapist and client on either the Judging or Perceiving personality dimension. Nonetheless, this equivalence may be questioned in light of the dichotomous nature of the MBTI scoring system. Thus, two individuals or groups may show the same MBTI "preference" while displaying a large discrepancy between their actual scaled scores. It may be questioned then whether the mean MBTI J—P scores for Judging therapists were significantly different from the mean J-P scores for Judging clients. Likewise, it may be questioned whether the mean MBTI J-P scores for Perceiving therapists were significantly different from the mean J-P scores for Perceiving clients. To answer these questions, Student's t-tests were per— formed comparing the MBTI mean J-P continuous scores of Judging therapists with Judging clients and, in addition, the mean J—P continuous scores of Perceiving therapists with Perceiving clients. When tested at the d = .05 level of significance, results showed no significant difference between Judging therapists and Judging clients [:t (36) = 1.01, ng] and no significant difference between Perceiving therapists and Perceiving clients [1 (46) = .75, né] on the MBTI Judging-Perceiving dimension. It would appear that the different samples were comparable on the paired dimension, thus justifying the current pairing procedure. The selection of three therapy sessions as the criterion for administration of the dependent measures poses some 121 problems for the study. It may be argued that the therapy process does not begin for several sessions and that three sessions represent a premature point in the therapy process. Dependent measures obtained at such a point in therapy would then be considered incomplete and, perhaps, misleading. Nonetheless, the three session criterion was chosen for practical considerations, i.e., to minimize the effects of client mortality. The selection of a longer criterion would have severely reduced the final client sample in light of the time constraints of the study. Furthermore, three sessions may, in fact, represent a sufficient length of time for therapists and clients to have established a therapeutic relationship. Relationship or "process" measures (BLRI scales) would then be valid at such a point in therapy. While a limitation of this study, the three session criterion may have had a negligible effect on several of the dependent measures. The final methodological issue concerns the choice of suitable therapy outcome measures. While the Adjective Check List and Survey Questionnaire were employed as measures of therapy outcome, questions may be raised as to the validity of these measures in assessing therapeutic gain. Further, the sensitivity of these instruments in detecting therapy changes after only three treatment sessions can also be questioned. Information concerning the validity of the Adjective Check List and Survey Questionnaire appears in Chapter III. 122 Concurrent validation of the measures was assessed through Pearson Product-Moment correlational analyses of the entire matrix of dependent measures (Table 3.6). Results showed significant correlations between each of the four ACL subscales used in the study and the Survey Questionnaire. The ACL FAV scale correlated .31 with the Survey Questionnaire (p = .004) while the ACL UNF scale correlated -.27 with the same measure (p = .011). In addition, the ACL SCF and ACL PAJ scales correlated .23 (p = .024) and .43 (p = .001) respectively, with the Survey Questionnaire. These findings suggest that the two indices of therapy outcome are measuring some common dimensions of personality change. Table 3.6 also reveals significant correlations for all but one of the outcome measures when contrasted with the client process measure (BLRI — Total Relationship scale-Client Form). That is, significant correlations were found between client total ratings of the therapy relation- ship and the ACL FAV scale (r = .30, p = .005); the ACL UNF scale (r .32, p = .003); the ACL PAJ scale (r = .25, p = .016); the Survey Questionnaire (r = .40, p = .001). Only the ACL SCF scale failed to show a significant correlation with the client process measure (r = .12, p = .148). Finally, the Survey Questionnaire was found to be significantly correlated (r = .30, p = .004) with the therapist process measure (BLRI-Total Relationship scale—Therapist Form). These results thus suggest a 123 concurrent validation of the various indices of therapy success. The sensitivity of these outcome measures in detecting therapeutic change after only three therapy sessions is a particularly troublesome area. Therapeutic change generally implies a cognitive-affective reorganization of existing behavioral and attitudinal patterns. It appears that such a reorganization can take place at any point during the course of psychotherapy. While the extent or strength of this reorganization varies considerably from one client to another, its presence even after only three sessions seems likely. It may be questioned, however, whether self—report rating scales, such as the Adjective Check List and Survey Questionnaire, are sensitive enough to reflect these changes. For instance, test-retest reliabilities of the four ACL subscales after a ten-week period generally fall in the .70 to low .80 range (Table 3.5). These reasonably high coefficients seem high enough to measure stable dimensions of personality, yet not too high to be insensitive to detect changes in personality dimensions. Additionally, research with the ACL has demonstrated progressive changes in self—concept after short periods of time as a result of treatment procedures (Williams, 1966). These findings suggest that the outcome measures used in the study, particularly the ACL subscales, may have been sensitive enough to detect personality changes that resulted from the three session treatment intervention. 124 Nonetheless, the lack of definitive data in this regard makes it difficult to reject the similarity hypothesis without additional supportive evidence. In conclusion, several methodological issues can be raised relevant to the failure to accept the similarity hypothesis (Hypothesis III). While the statistical results did not support this hypothesis, the substantive points noted above suggest the necessity for an additional look at the relationship between similarity and thera— peutic success. It would seem premature at this time to rule out the similarity hypothesis on the basis of the current findings. Moreover, the possibility should also be explored that a combination of similarity and dissimi- larity may facilitate therapeutic success. The encouraging results found in therapist-client compatibility studies, (e.g., Hetrick, 1970; Seidman, 1971) suggest that some similarity and some dissimilarity on selected personality dimensions may positively influence therapeutic process and outcome. Limitations and Suggestions for Future Research Perhaps the most crucial experimental weakness of the study was the failure to employ true random assignment of clients to therapists. While the assignment procedure used in the study can be justified on several grounds (See discussion in the preceding section), the absence of true randomization posed a threat to the internal validity 125 of the study (Campbell & Stanley, 1963). For this reason, future researchers in the area of therapist-client personality similarity should attempt to use experimental designs which employ true random assignment of clients to therapists. Such designs would provide a more internally valid test of the similarity hypothesis. Moreover, the employment of randomly assigned matched similarity/dissimilarity therapy dyads would appear to be a substantial improvement over the current practice of measuring already—existing therapy dyads. Such an approach would enable researchers to use a true experi- mental design rather than a quasi-experimental design (Borg & Gall, 1971) as was used in the present research. An additional limitation in the study was the employ- ment of a relatively small number of clients in similar and dissimilar therapy dyads. While a total of seventy- four clients was used in the research, an examination of the distribution of similar/dissimilar clients for each of the twelve subject therapists showed some serious discrep- ancies (Appendix J). That is, three of the twelve therapists treated only a minimum number of four clients (two similar and two dissimilar clients). Furthermore, seven of the twelve therapists treated only two dissimilar clients each. The similarity hypothesis may thus have been tested on too small a client sample thus lessening the chances of finding a significant difference between similar and dissimilar dyads. Future research in the area should then attempt to employ 126 larger numbers of clients in similar/dissimilar dyads. A further limitation in the study was the failure to employ a more diverse group of therapists for the therapist sample. It has been stated previously that the therapist sample was chosen from a population of primarily MBTI "NF" therapist types. The failure to obtain significant results when MBTI global similarity measures were utilized may have been a function of this lack of variation in the therapist group. It is for this reason that future therapist—client similarity research should include a more balanced distribution of therapist types. The use of a three session treatment criterion appeared to be another limitation in the study. While a short—term treatment approach was used in this investigation, most research in the area has favored a criterion of at least five or six therapy sessions (See Chapter II). The employment of a three—session criterion may have been too brief an exposure for a valid assessment of therapeutic outcome. Future investigators should then attempt to maximize the potentiality for finding a true similarity effect by expanding the treatment criterion. In addition, it would seem advisable to make several assessments of therapy process and outcome at fixed intervals along the course of treatment. This approach might then determine the differential effect of similarity at various stages of psychotherapy. 127 As stated previously, the choice of valid measures of therapeutic outcome was a particularly difficult problem in the current research. While the Adjective Check List and Survey Questionnaire were selected as measures of therapy gain largely as a result of their economical features (quick, easy to administer, easy to comprehend), more valid and psychometrically sound instruments are available to measure therapy outcome (e.g., Minnesota Multiphasic Personality Inventory, Tennessee Self—Concept Scale). Although the Adjective Check List seems to be a useful measure of personality characteristics, the utility of the instrument in detecting subtle personality changes may be legitimately questioned. Likewise, the Survey Questionnaire can be questioned regarding its ability to discriminate fine differences among clients' subjective impressions of psychotherapy. For these reasons, future researchers should attempt to employ more sensitive outcome measures than were used in the present study. In fact, the use of multiple outcome measures of shown validity may be the most effective approach to the problem of therapy outcome assessment. An equally difficult problem concerned the assessment of therapist-client personality similarity/dissimilarity. While the current study favored the use of a discrete personality dimension to determine personality similarity (MBTI Judging—Perceiving dimension), global similarity 2 measures were also employed (Cronbach's D and D indices). 128 The question arises, however, as to which particular personality measure(s) should be used to assess similarity, regardless of whether a discrete or global procedure is used. The use of the Myers-Briggs Type Indicator in the research was favored on the basis of its prior use in similarity studies (Mendelsohn & Geller, 1963; Jones, 1968). Nonetheless, perhaps a different personality measure (e.g., Minnesota Multiphasic Personality Inventory, California Psychological Inventory) would display differences which were not readily apparent with the use of the MBTI. Additional research is thus needed to explore the simi- larity hypothesis with differing measures of personality characteristics. Perhaps a factor analytic approach could be used to select relevant personality variables across a broad range of personality inventories. This method would enable future researchers to avoid the use of a trial—and-error method in the selection of personality variables and would thus afford a more sophisticated approach to the assessment of personality similarity. In summary, several weaknesses and limitations of the study were cited with suggestions given for future researchers. In light of the above discussion, additional similarity research seems justified before the similarity hypothesis is laid to rest. APPENDICES APPENDIX A SUMMARY DATA FOR CLIENTS USED IN THE STUDY immmHv HHuzm0444 immmac soHficom cam umosmmcflfiaom scum uwuamca.* ucwHHo nomm How mHanHm>m you dump HH<« -|< Hr-iH'm Ln HQGHOmHQ Hmucwz oz HmcHOmHQ Hopcmz .mmHGGD .mot ..wmm-.nom ..oNHrow me3omsom unwasum meonEwGD muwxuoz UwHHmeco mumeOB pwHHmeIHEmm muwxuoz onHme H nwmmu ucwumH ..0NH£om H EmHuHumw>mcmna u mGMHoHccome H .mwo msuo .owmmCD “mnoxuoz memm a .HwHO H .mHo .ooucm .mwnm “.msm wHuuHH mo mu0c3o H .mHa .OHtumO .mwnm N .mHo .mumm Hwnuo u .moum “OCH: “.msm .Em N .mumm mumsuwomcH wo mnwc3o imcuwocoo Mlnmumspmno mmmHHou N .mem .mm<|.mmmm .mH Ho .mumm .cHfipm H H .mnom HmoHkummm u wmeHOU 080m N n .muwm cHocmumm u mHmconmwmoum mequ 4mm OHM .wmez HmuHHmz “.msm prHmI.cwE u oumnpmuw .m.m II. mummy 4H" haswa .uommm .Hca mo .mmoum “.mumz .msm H u.uos¢ .xmz mmusanmmcmm 4mm N u .Hocm .uomwm .Hwa u .m.m meow II H n mHmouswz .:osoom>m u meconmmwonm m n xomHm Aimmuzv 0H": m u mHmoudwz w>Hmwmewo Hens: lmcuwocoo .oq n Hoonom we mummy OH" mHmousmz kHOHXC¢ wo .mmoum “.00xm cmHm pcmHm cuss mmoq mm” .0500 m.mNux nvnm «iamHmocmMHo h..1.3HAMBEHOHAEE ««:0Humosvm womm 0mm xwm Ammnzv wooem was 2H ammo Hoz mezmHuo mom aeao HmHEZDm m xHQmem¢ 130 APPENDIX C SUMMARY DATA FOR CLIENTS WHO TERMINATED EARLY IN TREATMENT (LESS THAN THREE SESSIONS) immmHv :oHHcmm mam smonSQHHHom eonm wmuamw<¥ AmmmHv HHISmQ¥* mN NIH meBmmsom H n usm©SDm m n oomonEOCD o n mnmxxoz wwHHmecD N n muwxnoz vaHmeIHEmm o n mumxuoz thHHMm mm m u mGMHochomB H n HQUHOmHQ “mnoxuoz mmHmm w .uoHo Hmucmz .mmeco “.msm mHupHH mo mumsao H momma ucmumH ..oNHnom mN H n .mHo .oowam .mmnm H u .moum HOGHZ «.mdm .Em II N n .mHo .mumm Hmauo mo mnmczo umcumocoo H umpmspwuw mmmHHou H u .mumm mumsvmomsH .mH mo .mnmm .GHE©¢ H n .mumm .mmml.mmmm o n wmoHHoo mEom H n .mumm UHoamumm H n mHmsonmmmon HmmmmH mN v u .UMHmz HmuHHmz “.msm wmmHm|.©mz OHM ou83©muo .m.m II m u HHSGH .uommm .ncé mo .mmoum “.mmmz .msm H u.uoE< .xmz wN H n .Hooa .uommm .Hca mHn .m.m meow mummw II H n mHmousoz .ommm o n mHmconmmmomm H u MOMHm nvanummsmm w u: N u mHmousmz .mmo Mona: “mammocoo .mH N u Hoosom mo mummw n u mHmouswz .x:¢ mo .mmoum “.00xm zmHm usmHm cane mme wNu .osmu m.mNum NNnm h_.....meoE.vsHo *uamfimonEm «COHHMODUM momm 0mm xwm Ammuzv Amonmmmm mmmmfi zmmB mmmHv BzmzfiHmou0m OH H N N N N mcH>Hooumm No N 4 m H m mchcsn No 4 N N N N mcHNUSH so HH m N N m mcH>HmoN0N so 6 m m m m mcH>Hmoumm mo m N m m N maH>Hmou0m Ho 4 N N N N qu>Hmou0m mo 6 N 4 N 4 maHmcsb No OH H m H m maHmwsb Ho comm wuquHU mHGmHHU mHsmHHO comm comm mama Hmnfisz Hmnssz HmHHsHmmHn HMHHaHm muawHHo muamHHo Hem: umHmmumse HmHmmumse Hopoe Hmnfisz Hmnfisz mGH>Hmoumm mchUSb HwQEDZ Hmnfisz mEmHmHMMmB MHmzflm H>Hm39 mmB m0 mofim mom ZOHmZMSHQ UZH>HM0mmmIUZH®QDb HBmS HEB ZO Ommfim MBZmHHU deHEHmmHQ 92¢ MHHHEHm m0 mmmSDZ b NHQmem4 139 APPENDIX K MBTI TYPE TABLE FOR THE THERAPIST SAMPLE APPENDIX K MBTI TYPE TABLE FOR THE THERAPIST SAMPLE (N=12) N % ISTJ ISFJ INFJ ”INTJ E 9 75 N = 0 N = 0 N = l N = 0 I 3 25 % = O % = 0 % = 8 % = O S 2 17 X N 10 83 T 1 8 F 11 92 ISTP ISFP 'INFP 'INTP J 6 50 N=0 N=0 N=2 N=0 P 650 % = 0 % = 0 % = 17 % = 0 IJ 1 8 x IP 2 17 0 EP 4 33 EJ 5 42 ESTP ESFP ENFP ‘ENTP ‘ ST 0 0 N = 0 N = 0 N = 4 N = 0 SF 2 17 % = 0 % = 0 % = 33 % = 0 NF 9 75 XX NT 1 8 00 SJ 2 17 SP 0 0 ESTJ ESFJ ENFJ ‘ENTJ NP 6 50 N = 0 N = 2 N = 2 N = 1 NJ 4 33 % = 0 % = 17 % = 17 % = 8 TJ l 8 x x x TP 0 0 O O FP 6 50 FJ 5 42 Note: X = Males (N=7) O = Females (N=5) 140 APPENDIX L MBTI TYPE TABLE FOR THE CLIENT SAMPLE APPENDIX L MBTI TYPE TABLE FOR THE CLIENT SAMPLE (N=74) N % ISTJ 3 ISFJ INFJ INTJ - E 30 41 N = 5 N = 11 N = 4 N = 1 I 44 59 % = 7 % = 15 % = 5 % = 1 s 38 51 xx xx x X N 36 49 000 00000 000 0000 T 18 24 L F 56 76 ISTP f ISFP INFP INTP ' J 32 43 N = 3 N = 7 N = 11 N = 2 P 42 57 % = 4 % = 9 % = 15 % = 3 IJ 21 28 xx X xx xx IP 23 31 0 000000 00000 0000 EP 19 26 EJ 11 15 ESTP ESFP ENFP ENTP ST 11 15 N = 1 N = 5 N = 10 N = 3 SF 27 36 % = 1 % = 7 % = 14 % = 4 NF 29 39 x xx x NT 7 9 000 00000 000 0000 SJ 22 30 i SP 16 22 ESTJ ESFJ ENFJ ENTJ NP 26 35 N = 2 N = 4 N = 4 N = 1 NJ 10 14 % = 3 % = 5 % = 5 % = 1 TJ 9 12 x xx TP 9 12 o 00 0000 o FP 33 45 FJ 23 31 Note: X = Males (N=20) O = Females (N=54) 141 APPENDIX M MBTI TYPE TABLE FOR THE ORIGINAL POPULATION OF THERAPISTS APPENDIX M MBTI TYPE TABLE FOR THE ORIGINAL POPULATION OF THERAPISTS (N=20) ISTJ ISFJ INFJ INTJ N=0 N=0 N=1 N=1 %=0 %=0 %=5 %=5 X X ISTP ISFP INFP INTP N = 0 N = 0 N = 4 N = 0 %=0 %=o %=20 %=0 X 000 ESTP ESFP ENFP ENTP N = 1 N = 0 N = 5 N = 1 %=5 %=0 %=25 %=5 X XXX X 00 ESTJ ESFJ ENFJ ENTJ N = 1 N = 2 N = 3 N = l % = 5 % = 10 % = 15 % = 5 O X X X o {00 Note: X = Males (N=11) O = Females (N=9) 142 EP EJ ST SF NF NT SJ SP NP NJ TJ TP FP FJ ONKO o\0 70 30 20 80 25 75 45 55 10 20 35 35 10 10 65 15 50 3O 15 10 45 3O APPENDIX N PHI COEFFICIENT INTERCORRELATIONS BETWEEN MBTI TYPE CATEGORIES Hw>mH Ho. Hm HQHOHngmHmNN H0>0H mo. um pamonquHm4 .BOHmQ Hammad wmeEmm How mmOQH paw chomep map m>onm Hmwmmm mmHmE How mGOHHMHmHHOU AmmmHv mmom paw HOHOHHum «MUmDOm mo.l *«Hm. MH. mlb mo. No.l mo. mIB 44mm. no. mo.l Zlm va. vo.l mo. Him mummwspm DHH *NHH. *«HN. mo. mlb NMH. no. mo. mIB NNwH. 00. mo. Zlm we. Ho.l No.l Him mpqmcsum Hoosom smHm mlb mIB Zlm Him mHmom mmHmomeHU mmNB HBmZ mezfimm mZOHBHHmmmoumMBZH EZHHUHmmmOU Hmm Z NHQZHmmfl 143 APPENDIX 0 SURVEY QUESTIONNAIRE APPENDIX O SURVEY QUESTIONNAIRE Code: Date: ' Directions: Please circle the number on the four-point Survey‘Questionnaire scales provided, which best describes your feelings or opinions. How helpful has therapy been for you? ‘l' ' 2 " ' ' 3 "“ ‘4 Not Somewhat Helpful Very helpful helpful helpful How satisfied are you with the progress you have made in therapy? 1 2 ‘3 ' 4 Not Somewhat Satisfied Very satisfied satisfied satisfied As a result of therapy, to what extent have your problems disappeared or have become less of a concern to you? l 2 3 4 Not at Somewhat Much Very all much As a result of therapy, how much improvement do you see in your ability to manage your current problems? 7.17,, .._ .2. . .7...7.73. . ....7 4 None at Some Much Very all much 144 145 APPENDIX 0 (continued) How do you feel you will be able to manage future problems that arise? l . 2 7 .. . .3.. 4 Not Somewhat Well Very very well well well As a result of therapy, to what extent do you feel better about yourself? 1 ‘ ‘2" ‘ 3' "4' Not Somewhat Much Very at all much APPENDIX P CELL MEANS FOR SUM AND DIFFERENCE VARIABLES APPENDIX P CELL MEANS FOR SUM AND DIFFERENCE VARIABLES Table P.l Cell Means for Judging and Perceiving Therapists on the Sum Scores for the Dependent Measures Judging Perceiving Dependent Therapists Therapists Measure (N=6) (N=6) Cl-BLRI-LR 44.55 50.37 Cl-BLRI-EU 25.46 31.49 Cl—BLRI-Cg 41.36 49.40 Cl—BLRI-UR 28.57 38.54 Cl-BLRI-Tl 139.94 169.79 Th-BLRI-LR 51.31 45.27 Th—BLRI—EU 25.03 21.75 Th—BLRI-Cg 38.21 36.58 Th—BLRI-UR 16.89 14.81 Th—BLRI-Tl 131.44 118.41 SQ 22.14 23.32 ACL FAV 4.52 6.69 ACL UNF - 8.83 - 2.55 ACL SCF - .30 .53 ACL PAJ 4.97 3.99 146 APPENDIX Q STEP-WISE REGRESSION TO ANALYZE THE CONTRIBUTION OF EACH OF THE FOUR MBTI DIMENSIONS IN PREDICTING THERAPY SUCCESS ammo.N nmam.H «mmH. HmmN. mmmH.H bmm Hum hmmH.N mmvH. mmNH.N NNNH. . mmHN.H mom HUN mmmN.H nomm. mooo.H Nvmm. wam. mZD HUN Hhoo. Homm. ommo. Hmvm. Hmoo. >¢m HON mmmN. mmvm. mnoN. mNmm. ONmH. 0m mvoo. Hmmm. HNoo. ommm. Hmoo. HBIHmHmIQB vNoo. thm. mHoo. Humm. hHoo. HBIHmHmIHO Hom pmgssoood swap m c300 swap m QHMHHM>HsD mHQMHHw> OUCMHHM> mme m mwum mme m Hw:OHqu©¢ w voom. swap mmmH m mm.n u mp Hbmh. u ummB mHMHHm>HpHsz How 05Hm> m COHHMSUM GOHmmmHmwm or“ 0p mocmummmHQ Hum oHMHHm>ou mchpm H.O mHnma mmMUUDm wmflmmmfi UZHBUHDmmm ZH mZOHmZMSHQ HBmS Mbom HEB m0 modm m0 ZOHBDmHMBZOU HEB MNNH¢Z¢ OB ZOHmmmmwmm mmHBImMBm O XHQmem< 148 149 OHmH.H NNNm. omvo. Nvmm. mmmm. bfim Hum mmmm.H Nvmv. hth. HHmN. Nmmv.H mom Hod CNNw. mem. ommH. mNHH. momm. m2: HUN Sham. vmmm. mNNN. mmNH. wam. >¢m HUN mHmm.m Homo. Nomm.N NHNH. OHmH.N om mmoo. oemm. mmoo. mmmm. mooo. HBIHmHmlse Hooo. ommm. Hooo. ommm. Hooo. HBIHmHmIHO Mom pmuasooom swap m 2300 many m mHMHHm>HcD mHQwHHm> mosMHHm> mmoH m mopm mmmH m HMGOHquvd N show. can» mme m mm.n u MU MHmm. n umoe mHMHHm>HHHsz How ous> m ApmscHHaoov GOHpmsum GOHmmmMmmm map Op mozmummmHo 21m mHMHHm>OO mcHCUH N.o mHQme O XHQmem¢ 150 HNOH. mHHm. momN. SHmn. Hume. bdm HON Home. Hamm. NHHN. mNom. Nmoo. mom Hoe homm.m omHo. oan.m mnvo. mono.w m2: H04 mmHN.H HNNm. Hmmm. mmmm. mHNm. >¢m HON mNNN. mmwm. momm. mmmm. HmmH. Om Neoo. HNnm. NHoo. momm. omoo. HBIHmHmIQB mmNH. mmNh. omNH. moNn. OMNH. HBIHMHmIHO Hom Umpssooofl gmsH m c309 gmsw m GHMHHMPHQD mewHHm> mocMHHm> mmmH m mmpm mmmH m HmcoHuHU©¢ w mmNm. swap mme m 46.5 n we NNNH.H u umma mumHum>HuHsz HON 05H6> m cOHpmsvm HOHmmonmm map 0p mocmHmeHo mIB QHMHHM>OO quU©< m.0 mHQMB AUOSCHHQOOV O NHQmemfi 151 hHhH.H mHoo. Hmmm.OH mHmm. wmvm. hfim H04 mmvo.N mmNm. NNmm. HMNN. HNHm.H mom HON OHmN.N NHON. mmvm.H onH. NHHN.H MZD HON vaH. mwwh. HnoH. Hmmn. _ MHmo. >¢m H04 vaN. vHom. ommo.H ammo. mowH. Om mono.N ova. mmHH.H momN. vmmv.H HEIqumISB «Hmh. onww. HHHm. onwv. HHHm. HBIHMHmIHU Mom Umpssoood QMSH m 2309 cmnH m mHMHHw>HQD oHQMHHm> mocMHHm> mmmH m mmpm mmmH m quoHuH66< N HNNo. carp mmmH m mm.s n ma HHHH.N n umme 0u6HHm>HuHss mom msHm> m cOHHmsum nOHmmmHmwm map ou mocmnmmmHo mun GHMHHM>OO mGHUUH v.0 mHQMB Apmschsoov O NHszmmH APPENDIX R SUPPLEMENTARY MANOVAS Nomm. Nmao. .N COHuomuchH Ammnsmmmzv mama meHHU.x xmm umHamumna x mama umHamumga wwmm. ammo. .N COHpomHmqu AmmHDmmmzv mama pcmHHU x Nmm pmHamqua Namm. mme. .N sOHuomamucH AmmHSmmmzv mama pcmHHU x mama umHamumaa Hamm. ONmm. 4N AmmHSmmmEV pommmm mama quHHO mwNN. mmmm.H .N nOHuomumpaH xmm umHaMHmna N mama HmHamumna mamm. mmmH.H .N Hommmm xmm PmHamumna HmoH. wmmm.N .N pommmm mama umHamnmsa swap mDHw> a Hp soHHMHHm> mo monsom mmmH a AmEHom umHaMHmna can HcmHHOV mmHmom aHSmGOHH IMHmm Hmuoa UnwsamHlupmHHmm map How mHQmHHm> HampgmamUnH HchHuHUUN mm mm =xmm umHamum£a= QHHB mUGMHHm> mo mHmmHmc< mmHSmmmS Umummamm mHMHHm>HpHsz H.m mema m¢>0242 mmfiaZMSMHaaDm m XHQZmaafi 152 154 mHHH. NOHo.H N .N qOHpounmuqH Ammnsmmmzv mama meHHO x xmm umHamumsa x mama HmHamumaa Nave. mmmo.m a .N coHpomuchH AmmHSmmmEV mama psmHHO x xmm HmHamnmna mmmw. mmHH. m .N HOHuomsmch Ammusmmmzv mama HcmHHU N mama umHammmna mmHm. Name. a .N AmmHSmmmzv pommmm mama psmHHO mHmm. moHH. a .N coHpomnmqu Nmm pmHamumaa N mama pmHamnmsa mama. @000. a .N Hommmm xmm pmHamnmna mmvw. moom. m .N Hommmm mama HmHamnmna swap msHm> a mp COHHMHH6> mo monsom mmmH a mHQmuo>8mcD Ho Insm HmHH Homao Umxomsu mm>HHomnpm Hmnasz .Umxomnu mm>Hpomfl©¢ mHQmuo>ma mo Hmnasz I mmHmom m>Hpomm64 map How mHQMHHm> HamwsmamcsH HchHuHUpm cm mm =xmm umHaMHmQa: nuHB mosmHHm> mo mHmmHmcm mmusmmmz Umpmmamm mHMHHm>HHHSE m.m mHQma Apmchpcoov m NHozmaam 155 mama. mmme. N .N nOHHomumuaH Ammusmmmzv mama pgmHHo N xmm pmHamumna x mama pmHamumaa mema. NHmN. N .N eoHuomumqu immusmmmsv mama pamHHo x xmm HmHamumaH momm. MHmH.H a .N GOHHomumuaH AmmHSwmmzv mama usmHHU x mama pmHaMHmsa Nmmm. mmmm. a .N Ammnsmmmzv Hommmm mama HgmHHU mmNm. meH. a .N soHpomumpsH xmm pmHaMHmaa x mama #mHamumsa ommm. mHoo. a .N pommmm xmm umHamumna mmmm. mmmH. a .N Hommmm mama pmHamumsa swap msHm> a mm COHHMHHM> mo mousom mmmH a usmfipmsapm HMGOmHma .mosmpHHGOOImHmm I mmHmom Insm HmHH Homno 0>HpomH6N was How 0HanHm> unmcqmmmecH HmaOHuHawa mm mm =Nmm pmHamnmza= apHB mUGMHHm> mo mHmmHm:< mmusmmmz pmpmmamm mHMHHm>HuHsz H.m mHQma HemscHucoov m xHozmmma APPENDIX S UNIVARIATE ANALYSIS OF VARIANCE WITH A REPEATED MEASURES FACTOR (CLIENTS) FOR THE BARRETT-LENNARD RELATIONSHIP INVENTORY SUBSCALE (CLIENT FORM) - UNCONDITIONALITY OF REGARD Hm>mH mo. map um HamoHNHcmHm4 om.¢ Ammo. u a HNmn.mH .OH .__..mwaJQHULGHAHHs wuomhoam_x.fiumnsmmmzv pamHHo mam.H mmHm.HN H sOHuomschH Hmasouwv umHamumna x mensmmmzv psmHHO «Hmv.w mmov.NHH H Ammusmmmzv musmHHO NH . muomHnsm angHz NHGm.NmH OH masouo :HapHa mpUmHnsm moH.N NNmm.mmN H Amazonwv mumHamumsa .mm_ mummHQSm cmmBumm dem> a m2,. ..Ho. .aoHHMHHm> mo mousom QMflwmm m0 NBHHHZOHBHQZOUZD Ifizmom BZMHHUV mqmummbm mmOBZH>ZH aHmmZOHaflHmm QmHZZMHIBBmmm4m HEB mom AmBZMHHUV moavflm mmmDm a0 mHmMH¢Z¢ mafiHm¢>HZD m NHmzmaa4 156 BIBLIOGRAPHY BIBLIOGRAPHY American Psychiatric Association: Diagnostic and statistical manual of mental disorders, 2nd ed., Washington, D.C., 1968. Axelrod, J. An evaluation of the effect on progress in therapy of similarities and differences between the personalities of patients and their therapists. Dissertation Abstracts, 1952, 12 (3), 329. Bare, C.E. Relationship of counselor personality and counselor-client personality similarity to selected counseling success criteria. Journal of Counseling Psychology, 1967, 14, 419-425. Barrett-Lennard, G.T. Dimensions of therapist response as causal factors in therapeutic change. Psychological Monographs, 1962, 16, Whole No. 43. Bordin, E.S. Psychological counseling. New York: Appleton-Century-Crofts, 1968. Borg, W.R. & Gall, M.D. Educational research: An introduction. New York: McKay, 1971. Bradway, K. Jung's psychological types: Classification by test versus classification by self. Journal of Analytical Psychology, 1964, 9, 129-135. Braun, J.A. The empathic ability of psychotherapists as related to therapist perceptual flexibility and professional experience, patient insight, and therapist-patient similarity. Dissertation Abstracts International, 1971, 32 (4—B), 2391-2392. Byrne, D. & Clore, G.L. Effectance arousal and attraction. Journal of Personality & Social Psychology, 1967, 6, 1—18. Byrne, D. & Griffitt, W. Interpersonal attraction. In P.H. Mussen & M.R. Rosenzweig (Eds.), Annual review of psychology, Vol. 24, Palo Alto: Annual Reviews, Inc., 1973. 157 158 Byrne, D., Griffitt, W., & Stefaniak, D. Attraction and similarity of personality characteristics. Journal of Personality & Social Psychology, 1967, 5, 82-90. Campbell, D.T. & Stanley, J.C. Experimental and quasi- experimental designs for research. Chicago: Rand McNally, 1963. Carkhuff, R.R. & Pierce, R. Differential effects of therapist race and social class upon patient depth of self-exploration in the initial clinical interview. Journal of Consultinngsychology, 1967, 31, 632-634. Carlyn, M. The relationship between Myers—Briggs personality characteristics and teaching preferences of prospective teachers. Unpublished doctoral dissertation, Michigan State University, 1976. Carson, R.C. & Heine, R.W. Similarity and success in therapeutic dyads. Journal of Consulting Psychology, 1962, 26, 38-43. Carson, R.C. & Llewellyn, C.E., Jr. Similarity in therapeutic dyads: A reevaluation. Journal of Consulting Psychology, 1966, 30, 458. Clark, J.V. & Culbert, S.A. Mutually therapeutic perception and self-awareness in a T group. Journal of Applied Behavioral Science, 1965, I, 180-194. Cronbach, L.J. Processes affecting scores on "under- standing of others" and assumed similarity. Psychological Bulletin, 1955, 53, 177-193. Cronbach, L.J. Coefficient alpha and the internal structure of tests. In W.A. Mehrens & R.L. Ebel (Eds.), Principles of educational and psychological measurement. Chicago: Rand McNally, 1967. Cronbach, L.J. & Gleaser, G.C. Assessing similarity between profiles. Psychological Bulletin, 1953, 50, 456-473. Cutler, R.L. Countertransference effects in psychotherapy. Journal of Consulting Psychology, 1958, 23, 349-356. Farnsworth, K.E. Application of scaling techniques to the evaluation of counseling outcomes. Psychological Bulletin, 1966, 66, 81—93. Fitts, W.H. Tennessee Self Concept Scale: Manual. Nashville: Counselor Recordings and Tests, 1964. 159 Fosshage, J.L. Effect of counselor—client interest similarity on counseling relationship and outcome. Dissertation'Abstracts, 1969,'22_(9-B), 3481-3482. Fromm-Reichmann, F. ‘PrincipleS'of“intensiVe'psychotherapy. Chicago: University of Chicago Press, 1950. Gassner, S.M. Relationship between patient-therapist compatibility and treatment effectiveness.‘ Journal 'of consulting &'C1inical'Psych010gy, l970,’§4, 408—414. Gerler, W. Outcome of psychotherapy aS-a function of client-counselor similarity. 'Dissertation Abstracts, 1958, 1§_(5), 1864-1865. Gough, H.G. &‘Heilbrun, A.B., Jr. 'The Adjective Check LiSt: ManUal. Palo Alto: Consulting Psychologists Press, 1965. Gray, H. & Wheelwright, J.B. Jung's psychological types, their frequency of occurrence. ‘Journa '0 ‘General Gross, W.F. & DeRidder, L.M. Significant movement in comparatively short—term counseling. ‘JOUrnal of COunseling Psychology, l966,‘13, 98—99. Halpern, H.H. Empathy, similarity, and self-satisfaction. Journal of Consulting Psych01ogy, 1955,'12, 449-452. Hathaway, S.R. & McKinley, J.C. 'Minnesota'Multiphasic Personality‘InVentory: Manual'for‘administration and scoring. New York: Psychological Corporation, 1967. Hebert, D.J. The counseling relationship as a function of client—counselor personality need and sex similarity. DiSsertation Abstracts, 1968, 28 (ll-A), 4479-4480. '— Heider, F. The psycholOgyof interpersonal relations. New York: Wiley, 1958. Hetrick, S.H. An analysis of "client" - "therapist" compatibility and its effect on transmissive and receptive communication in a psychotherapy analogue. Dissertation AbStracts International, 1970, 31 (3-B), 1538-1539. Hollingshead, A.B. & Redlich, F.C.‘ Social class and mental illneSS. New York: Wiley, 1958. 160 Holzman, M.S. The significance of the value systems of patient and therapist for the outcome of psychotherapy. Dissertation Abstracts, 1962, 22_(ll), 4073. Hurst, J.C. & Weigel, R.G. Counseling Services Assessment Blank. Fort Collins: Rocky Mountain Behavioral Soience Institute, 1968. Izard, C.E. Personality similarity and friendship. Journal of Abnormal & Social Psychology, 1960, 61, 47-51. (a) Izard, C.E. Personality similarity, positive affect, and interpersonal attraction. Journal of Abnormal & Social PSychology, 1960, 61, 484-485. (b) Jones, H.T. The relationship of counselor-client personality similarity to counseling process and outcome. Unpublished doctoral dissertation, University of Missouri, 1968. Jung, C.G. Psychological types (H.G. Baynes, trans.). New York: Harcourt, Brace, 1923. Kell, B.L. & Burow, J.M. Developmental cougseling and therapy. Boston: Houghton Mifflin, 1970. Kell, B.L. & Mueller, W.J. Impact and change: A study of counseling relationships. New York: Appleton- Century-Crofts, 1966. Kessel, P. Control of verbal behavior as a function of social reinforcement, the subject's conception of the interviewer's values relative to his own, and need for social approval: A psychotherapy analogue study. Dissertation Abstracts, 1967, 28 (3-B), 1198-1199. "" Kiesler, D.J. Some myths of psychotherapy research and the search for a paradigm. Psychological Bulletin, 1966, 66, 110-136. Lesser, W.M. The relationship between counseling progress and empathic understanding. Journal of Counseling Psychology, 1961, 6, 330-336. Levy, N., Murphy, C., Jr., & Carlson, R. Personality types among Negro college students. Education§1 and Psychological Measurement, 1972, 61, 641-653. Lichtenstein, E. Personality similarity and therapeutic success: A failure to replicate. Journal of Consulting Psychology, 1966, 66, 282. 161 Luborsky, L. Psychotherapy. In P.R. Farnsworth & Q. McNemar (Eds.), Annual review of psychology, Vol. 10, Palo Alto: Annual Reviews, Inc., 1959. Masterson, S. The adjective checklist technique: A review and critique. In P. McReynolds (Ed.), Advances in psychological assessment, Vol. 3, San Francisco: Jossey-Bass, 1975. McNair, D.M., Lorr, M., & Callahan, D.M. Patient and therapist influences on quitting psychotherapy. Journal of Consulting Psychology, 1963, 22, 10-17. McNally, H.A. An investigation of selected counselor and client characteristics as possible predictors of counseling effectiveness. Dissertation Abstracts International, 1973, 26_(12-A), 6672-6673. McWhirter, J.J. Two measures of the facilitative conditions: A correlational study. _Journal of Counseling Psychology, 1973, 26, 317-320. Meltzoff, J. & Kornreich, M. Research in psychotherapy. New York: Atherton Press, 1970. Mendelsohn, G.A. Review of the Myers-Briggs Type Indicator. In O.K. Buros (Ed.), The sixth mental measurements yearbook. Highland Park, N.J.: Gryphon Press, 1965. Mendelsohn, G.A. Effects of client personality and client- counselor similarity on the duration of counseling: A replication and extension. Journal of Counseling Psychology, 1966, 12, 228-234. Mendelsohn, G.A. & Geller, M.H. Effects of counselor- client similarity on the outcome of counseling. Journal of Counseling Psychology, 1963, 16, 71—77. Mendelsohn, G.A. & Geller, M.H. Structure of client attitudes toward counseling and their relation to client-counselor similarity. Journal of Consulting Psychology, 1965, 26, 63-72. Mendelsohn, G.A. & Geller, M.H. Similarity, missed sessions, and early termination. Journal of Counseling Psychology, 1967, 16, 210-215. Mendelsohn, G.A. & Rankin, N.O. Client-counselor compati- bility and the outcome of counseling. Journal of Abnormal Psychology, 1969, 26, 157-163. Michigan State University. Counseling Service Questionnaire, Form 75. East Lansing: Michigan State University Counseling Center, 1975. 162 Mihalick, R.E. Values and psychotherapy. Dissertation Abstracts International, 1970, 26 (9-B), 4377. Moses, R.G. Perceived and tested value similarities between client and counselor and their relationship to counseling outcome criteria. Dissertation Abstracts International, 1970, 66 (9-B), 4378. Murray, H.A., §E_§1. Explorations in personality. New York: Oxford University Press, 1938. Myers, I.B. The Myers-Briggs Type Indicator: Manual. Princeton: Educational Testing Service, 1962. Myers, R.A. Research in counseling psychology - 1964. Journal of Counseling Psychology, 1966, 12, 371-379. Newcomb, T.M. An approach to the study of communicative acts. Psychological Review, 1953, 66, 393-404. Newcomb, T.M. The acquaintance process. New York: Holt, Rinehart and Winston, 1961. Ourth, L.L. The relationship of similarity in therapist- client pairs to clients' stay and improvement in psychotherapy. DisSertation AbStracts, 1964, 26 (9), 3839-3840. Patterson, C.H. Theories of counseling and psychotherapy. New York: Harper & Row, 1966. Paul, G.L. Strategy of outcome research in psychotherapy. In D.E. Hendrickson & F.H. Krause (Eds.), Counseling and psychotherapy: Training and supervision. Columbus: Charles E. Merrill, 1972. Powell, T.L. An investigation of the effectiveness of therapist-patient dyads in an initial psychotherapy interview under conditions of A-B complementarity and similarity. Dissertation Abstracts International, 1970, 21 (6-B), 3712-3713. Praul, J.A. The relationship of personality variables to selected outcomes of the counselor training process. Dissertation Abstracts International, 1970, 61 (l-A), 161. Rogers, C.R. Client-centered therapy. Boston: Houghton Mifflin, 1951. Rogers, C.R. On becoming a person. Boston: Houghton Mifflin, 1961. 163 Sapolsky, A. Relationship between patient-doctor compatibility, mutual perception, and outcome of treatment. Journal of Abnormal Psychology, 1965, 26, 70-76. Scarr, S. The Adjective Check List as a personality assessment technique with children: Validity of the scales. Journal of Consulting Psychology, 1966, ‘66, 122-128. Schopler, J.H. The relation of patient-therapist personality similarity to the outcome of psychotherapy. Dissertation Abstracts, 1959, 16 (10), 2659. Scott, W.E., Jr. & Day, G.J. Personality dimensions and vocational interests among graduate business students. Journal of Counseling Psychology, 1972, 16, 30-36. Seidman, E. A and B subject-therapistS' responses to videotaped schizoid and intropunitive-neurotic prototypes. Journa1 of Consulting & Clinical Psychology, 1971, 62, 201-208. Snett, P.I. The effect of therapist-patient conflict similarity upon therapists' prognostic evaluations and other clinical judgments of their patients. Dissertation Abstracts International, 1973, 66 (ll-B), 5526. Spencer, C.D. Two types of role playing: Threats to internal and external validity. American Psychologist, 1978, 66, 265-268. Spivak, M. Factors influencing the formation of a patient- percept by psychiatrists following the initial interview. Dissertation Abstracts, 1962, 26 (l), 341. Stricker, L.J. & Ross, J. Intercorrelations and relia- bility of the Myers-Briggs Type Indicator scales. Psychological Reports, 1963, 12, 287-293. Stricker, L.J. & Ross, J. An assessment of some structural properties of the Jungian personality typology. Jougnal of Abnormal & Social Psychology, 1964, _6, 62-71. (a) Stricker, L.J. & Ross, J. Some correlates of a Jungian personality inventory. Psychological Reports, 1964, 16, 623-643. (b) Sullivan, H.S. The interpersonal theory of psychiatry. New York: W.W. Norton & Co., 1953. 164 Tessler, R.C. Clients' reactions to initial interviews: A study of relationship-formation."Dissertatiqg Abstracts International, 1973, 66,(7—A), 3777-3778. Thibaut, J.W. & Kelley, H.H. 'Th ’social'psyChology'of 'groups. New York: John Wiley & Sons, 1959. Tosi, D.J. The counseling relationship as perceived by the client following the initial encounter as a function of dogmatism within the counselor-client dyad. ‘Dissertation Abstracts, 1969, 29 (lZ-B), 4856-4857. "'" Tuma, A.H. & Gustad, J.W. The effects of client and counselor personality characteristics on client learning in counseling. 'Journal‘Of counseling Psychology, 1957, 6, 136-141. Vogel, J.L. Authoritarianism in the therapeutic rela- tionship.'Journalofconsulting'PsyphologY: 1961, 26, 102-108. . Welkowitz, J., Cohen, J. & Ortmeyer, D. Value system similarity: Investigation of patient—therapist dyads. JOurnal'of Consulting‘Psychology, 1967, 61, 48-55. Wellman, F.E. A conceptual framework for the derivation of guidance objectives and outcome criteria: Preliminary statement. In J.M. Whiteley (Ed.), ‘ ResearCh‘in’Counseling}'Evaluation'and refocus. Columbus: Charles E. Merrill, 1967. Whiteley, J.M. ResearCh in counseling: Evaluation and refOCus. Columbus: Charles E. Merrill, 1967. Whittlesey, R.R. The effect of client-counselor similarity on selected measures of counseling outcome: An application of Holland's hexagonal configuration. DissertgtionAbstracts'International, 1972, 62 (9-A), 4973-4974. ' Wiebe, B. & Pearce, W.B. An item-analysis and revision of the Barrett-Lennard Relationship Inventory. Journal of Clinical Psychology, 1973,26, 495-497. Williams, A.F. Social drinking, anxiety, and depression. Journal'of‘PerSOnality'6'Social‘PsycholOgy, 1966, 6, 689—693. Winer, B.J. ‘StatisticalgprincipleS'in'experimental design. New York: McGraw-Hill, 1962. Wogan, M. Effect of therapist-patient personality variables on therapeutic outcome. JOurnal of Consulting & Clinical Psychology, 1970, 66, 356-361. IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII IIIIII