JOURNAL OF PERSONALITY ASSESSMENT, 83(3), 345–356 Copyright © 2004, Lawrence Erlbaum Associates, Inc.
Quality of Object Relations as a Moderator of the Relationship Between Pattern of Alliance and Outcome in Short-Term Individual Psychotherapy QUALITY PIPER, OGRODNICZUK, OF OBJECT RELATIONS JOYCE
William E. Piper and John S. Ogrodniczuk Department of Psychiatry University of British Columbia
Anthony S. Joyce Department of Psychiatry University of Alberta
In this study, we investigated the personality variable quality of object relations (QOR) as a moderator of the relationship between the pattern of the therapeutic alliance and treatment outcome in two forms (interpretive, supportive) of short-term individual psychotherapy. In a sample of 72 psychiatric outpatients who completed interpretive therapy, QOR emerged as a moderator for the outcome factor general symptomatology and dysfunction. For high-QOR patients, an increasing level of alliance was directly related to benefit, whereas for low-QOR patients, a decreasing level of alliance was directly related to benefit. An explanation for these findings emphasized the importance of patients repeating their typical pattern of maladaptive interpersonal behavior in the therapy sessions in the context of the therapist working with the transference. In a sample of 72 psychiatric outpatients who completed supportive therapy, QOR did not emerge as a moderator. These findings in combination with evidence from previous studies suggest that QOR should be investigated as a moderator variable in future studies of short-term psychotherapy.
There is general agreement in the literature (e.g., Horvath & Symonds, 1991; Martin, Garske, & Davis, 2000; Orlinsky, Ronnestad, & Wollutzki, 2004) that the therapeutic alliance has a moderately strong, direct relationship to favorable outcome for many types of psychotherapy. It has been viewed as an important common factor of different psychotherapies. The relationship between the average level of alliance and outcome appears to be quite robust, that is, relatively unaffected by possible moderator variables such as the time of assessment (e.g., early, late in therapy) or the type of rater (e.g., patient, therapist). Moderator variables indicate under which conditions or for which participants the relationship between two other variables differ. In most studies, the alliance has been represented by a single score that is based on one assessment or an average score that is based on multiple assessments. This is despite the fact that the alliance is usually conceptualized as a dynamic variable that changes over time. Only a small number of studies have described patterns of alliance over time. Mann (1973) and Miller et al. (1983) reported a curvilinear, high-low-high pattern over the course of
therapy. Others (Gunderson, Najavits, Leonhard, Sullivan, & Sabo, 1997; Joyce & Piper, 1990; Paivio & Patterson, 1999) have reported a direct linear pattern. Still others (Hellerstein et al., 1998; Klee, Abeles, & Muller, 1990; Mallinckrodt, 1996; Piper, Boroto, Joyce, McCallum, & Azim, 1995) have reported considerable fluctuations or variability in patterns among patients. Thus, the findings from the different studies concerning the nature of the pattern of alliance over time have been divergent. Even fewer studies have focused on the relationships between patterns of alliance over time and outcomes. Again, the findings have been divergent. However, the divergent relationships between patterns and outcomes have usually occurred within the same study and may have been attributable to moderator variables. Hellerstein et al. (1998) found that the degree of fluctuation of patient-rated alliance was directly related to favorable outcome in confrontive dynamic therapy and inversely related to favorable outcome in supportive therapy. Westerman, Foote, and Winston (1995) found that an increase in patient-rated alliance was directly
346
PIPER, OGRODNICZUK, JOYCE
related to favorable outcome in confrontive dynamic therapy and inversely related to favorable outcome in nonconfrontive dynamic therapy. In these two studies, it appears that the relationship between pattern of alliance and outcome was moderated by form of therapy. Our research team has been interested in the personality variable known as quality of object relations (QOR) as a potential moderator of relationships involving alliance and outcome. QOR is defined as a person’s lifelong pattern of establishing certain kinds of relationships that range along an overall dimension from primitive to mature. At the primitive end of the scale, relationships are characterized by emotionality, instability, and destructiveness. At the mature end, relationships are characterized by love, tenderness, and concern. QOR is measured by a semistructured interview. Its reliability and construct validity have been investigated in a series of clinical studies and summarized in a review article (Piper & Duncan, 1999). In terms of predictive validity, studies have consistently demonstrated a direct relationship between QOR and benefit in time-limited interpretive therapies (de Carufel & Piper, 1988; Piper, Azim, Joyce, McCallum, Nixon, et al., 1991; Piper, Joyce, Azim, & Rosie, 1994). Other investigators, using other measures of object relations, have also reported direct relationships between QOR and benefit in psychotherapy (Cook, Blatt, & Ford, 1995; Ford, Fisher, & Larson, 1997; Honig, Farber, & Geller, 1997). The primary explanation for the consistent findings is that the lifelong pattern of satisfactory give-and-take relationships of high-QOR patients compared to low-QOR patients enable them to better tolerate and work with the demanding, depriving, and anxiety-arousing features of interpretive therapy and with the time-limit and impending termination. In addition to finding evidence for a direct relationship between QOR and outcome, our research team has found evidence of QOR acting as a moderator variable in a series of studies that are summarized in Table 1. In a study that investigated the pattern of alliance over time, Piper et al. (1995) found that an increase in therapist-rated alliance was directly related to favorable outcome for patients with low QOR and was virtually unrelated to favorable outcome for patients with high QOR. Instead, average level of therapist-rated or patient-rated alliance was directly related to favorable improvement for patients with high QOR. Our research team has also studied the degree to which QOR moderates the relationship between the therapist’s use of transference interpretations and treatment outcome. In an initial study (Piper, Azim, Joyce, & McCallum, 1991), high levels of transference interpretations were associated with a weaker alliance and poorer outcome for high-QOR patients but not low-QOR patients. In a later study (Ogrodniczuk, Piper, Joyce, & McCallum, 1999), the reverse pattern was found, that is, high levels of transference interpretations were associated with a weaker alliance and poorer outcome for low-QOR patients but not high-QOR patients. We have reasoned that the seemingly discrepant findings actually com-
plement each other when the different absolute levels of transference interpretations in the two studies are taken into consideration. It is possible that the much higher levels in the initial study had a differential effect on high-QOR patients and a constant effect on low-QOR patients. The reverse may have occurred in the latter study. The relationship between the therapist’s provision of “accurate” transference interpretations and treatment outcome has also differed depending on the level of QOR (Piper, Joyce, McCallum, & Azim, 1993). Accuracy was defined as the extent to which the transference interpretation corresponded to the therapist’s initial psychodynamic formulation about the patient. Accurate transference interpretations had beneficial effects for high-QOR patients and detrimental effects for low-QOR patients. These findings concerning both the extent of use and the accuracy of transference interpretations have received support in independent studies by other investigators (Connolly et al., 1999; Høglend, 1993; Høglend & Piper, 1995). More recently, two additional clinical trials that have provided evidence in support of QOR serving as a moderator variable have been completed. The first (Piper, Joyce, McCallum, & Azim, 1998) compared short-term, interpretive individual therapy and short-term, supportive individual therapy. The second (Piper, McCallum, Joyce, Rosie, & Ogrodniczuk, 2001) compared short-term, interpretive group therapy and short-term, supportive group therapy. In both studies, there was an interaction effect between QOR and form of therapy. QOR was directly related to benefit in interpretive therapy and inversely, although minimally, related to benefit in supportive therapy. Thus, QOR moderated the relationship between each form of therapy and outcome in each of the two studies. This provided further evidence of the importance of QOR as a moderator variable in short-term psychotherapy. Many of the previous studies that have investigated patterns of alliance have had basic methodological limitations. These have included small sample sizes, unsophisticated methods for determining patterns of change, and inattention to patient personality characteristics as moderators. The recently completed comparative trial that involved individual therapies included assessments of the therapeutic alliance after each session. It provided a new opportunity to investigate the potential role of QOR as a moderator of the relationship between pattern of alliance and outcome with two large samples of patients (interpretive therapy, supportive therapy) and a sophisticated method (hierarchical linear modeling [HLM]) of determining patterns of change. This objective is the focus of this article. On the basis of the findings from the previous study (Piper et al., 1995) that investigated pattern of alliance and outcome, we formulated the following two hypotheses. 1. In interpretive therapy, an increase in therapist-rated alliance will be directly related to favorable outcome for low-QOR patients but not high-QOR patients.
347
QUALITY OF OBJECT RELATIONS
TABLE 1 QOR Moderator Effect Studies
Reference
Sample
Treatment
Criterion Variables
Piper, Boroto, Joyce, McCallum, & Azim (1995)
32 high-QOR and 32 low-QOR psychiatric outpatients who completed therapy
Short-term (20 sessions), interpretive individual psychotherapy
Outcome of therapy
Piper, Azim, Joyce, & McCallum (1991)
32 high-QOR and 32 low-QOR psychiatric outpatients who completed therapy
Short-term (20 sessions), interpretive individual psychotherapy
Therapeutic alliance, outcome of therapy
Ogrodniczuk, Piper, Joyce, & McCallum (1999)
10 high-QOR and 30 low-QOR psychiatric outpatients who completed therapy
Short-term (20 sessions), interpretive individual psychotherapy
Therapeutic alliance, outcome of therapy
Piper, Joyce, McCallum, & Azim (1993)
32 high-QOR and 32 low-QOR psychiatric outpatients who completed therapy
Short-term (20 sessions), interpretive individual psychotherapy
Therapeutic alliance, outcome of therapy
Piper, Joyce, McCallum, & Azim (1998)
72 psychiatric outpatients who completed interpretive therapy and 72 psychiatric outpatients who completed supportive therapy
Outcome of therapy
Piper, McCallum, Joyce, Rosie, & Ogrodniczuk (2001)
53 psychiatric outpatients with complicated grief who completed interpretive group therapy and 54 psychiatric outpatients with complicated grief who completed supportive group therapy
Short-term (20 sessions), interpretive individual psychotherapy or short-term (20 sessions), supportive individual psychotherapy Short-term (12 sessions), interpretive group psychotherapy or short-term (12 sessions), supportive group psychotherapy
Note.
Outcome of therapy
Effect sizes are expressed as Pearson correlation coefficients. QOR = quality of object relations.
Main Findings and Effect Sizes For high-QOR patients, average level of therapeutic alliance was directly related to favorable outcome (average r = .47), and increase in alliance was not significantly related to favorable outcome; for low-QOR patients, increase in therapist-rated alliance was directly related to favorable outcome (average r = .48) and was more strongly related to favorable outcome than average level of alliance For high-QOR patients, amount of transference interpretations was inversely related to two types of therapist–rated therapeutic alliance (r = –.36 and r = –.51) and inversely related to two types of favorable outcome (r = –.58 and r = .47); for low-QOR patients, the relationships were nonsignificant. For high-QOR patients, amount of transference interpretations was directly related to therapist-rated alliance (r = .66) and not significantly related to outcome; for low-QOR patients, amount of transference interpretations was inversely related to patient-rated alliance (r = –.41) and inversely related to favourable outcome (r = –.43) For high-QOR patients, accuracy (correspondence) of transference interpretations was directly related to favorable follow-up outcome (r = .39), and was not significantly related to patient-rated or therapist-rated alliance; for low-QOR patients, accuracy of transference interpretations was inversely related to favorable follow-up outcome (r = –.49), and was inversely related to patient-rated alliance (r = –.37) and therapist-rated alliance (r = –.37) In interpretive therapy, QOR was directly related to favorable outcome (average r = .34); in supportive therapy, QOR was virtually unrelated to favorable outcome
In interpretive therapy, QOR was directly related to favorable outcome (r = .37); in supportive therapy, QOR was inversely, but not significantly, related to favorable outcome
348
PIPER, OGRODNICZUK, JOYCE
2. In interpretive therapy, average level of patient-rated or therapist-rated alliance will be directly related to favorable outcome for high-QOR patients.
METHOD Setting and Procedure A detailed description of the methods of the comparative trial are provided in a previous publication (Piper et al., 1998). Patients had been referred from the Walk-In Clinic of the Department of Psychiatry, University of Alberta Hospital Site, Edmonton, Alberta, Canada. After obtaining informed consent, patients were matched in pairs on QOR, psychological mindedness, use of medication, and when possible, age and gender. One patient from each pair was randomly assigned to interpretive therapy and the other to supportive therapy for a given therapist. During a 3-year period, 72 patients completed interpretive therapy and 72 patients completed supportive therapy for a total of 144 patients. Dropouts were defined as patients who attended 13 or fewer of the 20 sessions and who terminated against the advice of their therapist. Dropouts were replaced by a comparable matched patient. There was no significant relationship between QOR and dropping out nor was there a significant interaction between QOR and form of therapy in regard to dropping out. Patients and Therapists Diagnoses were made using the Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev. [DSM–III–R]; American Psychiatric Association, 1987). Axis I diagnoses were identified by the computer-administered Mini-Structured Clinical Interview for DSM–III–R (Mini–SCID; First, Gibbon, Williams, & Spitzer, 1990) and validated by an independent clinical diagnosis assigned jointly by the intake assessor and a staff psychiatrist, both of whom saw the patient on the day of intake. Among the 144 therapy completers, a total of 73% of the patients received an Axis I diagnosis. The most frequent disorders were current major depression (49%) and dysthymia (26%). Axis II diagnoses were determined by the computer-administered SCID Personality Questionnaire (SCID–II PQ) and Auto–SCID II (First, Gibbon, Williams, & Spitzer, 1991). A total of 60% of the patients received an Axis II diagnosis. The most frequent Axis II disorders were avoidant (29%), obsessive–compulsive (24%), and borderline (22%). A total of 46% of the patients received both Axis I and Axis II diagnoses. The average age of the patients was 34.3 years (SD = 9.6, range = 18 to 62). Sixty-one percent were women. Forty-two percent were married or living with a partner, 21% were separated or divorced, and 37% had never been married. About two thirds (67%) of the patients were educated beyond high
school, and 71% were employed. The racial composition was White (94%), East Indian (2%), Native American (2%), Asian (1%), and Semitic (1%). Many (73%) of the patients reported receiving previous psychiatric treatment, but few (8%) had a prior history of psychiatric hospitalization. There were eight therapists (three psychologists, two social workers, two occupational therapists, and one psychiatrist). Seven were White and one East Indian. Five were women. The therapists’ average age was 43.6 years (SD = 6.1, range = 37 to 52), and their average experience practicing individual psychotherapy was 11.8 years (SD = 4.9, range = 3 to 19). The therapists were experienced with both interpretive and supportive forms of psychotherapy. Each therapist treated nine interpretive therapy cases and nine supportive therapy cases. Therapy Each patient received an interpretive or a supportive form of short-term, time-limited, individual dynamically oriented psychotherapy. The patient was scheduled for weekly 50-min sessions at a regular, prearranged time for 20 weeks. In interpretive therapy, the primary objective is to enhance the patient’s insight about repetitive conflicts (intrapsychic and interpersonal) and trauma that serve to underlie and sustain the patient’s problems. The therapist encourages the patient to explore uncomfortable emotions and withholds immediate praise and gratification. The therapist is active, interpretive, and transference focused. In supportive therapy, the primary objective is to improve the patient’s immediate adaptation to his or her life situation. Rational problem solving is encouraged. The therapist attempts to minimize anxiety and regression in the session and provides praise and immediate gratification. The therapist is active, noninterpretive, and other focused (i.e., focused on the patient’s current external relationships). Although the therapists were experienced in providing a variety of interpretive and supportive therapies in the clinic, they participated in a 6-month training period before taking cases in the project. The therapists followed a two-part technical manual that described, illustrated, and compared the technical emphases associated with the two forms of therapy (Piper, Azim, Joyce, & McCallum, 1993). Adherence to the manual was monitored using the Interpretive and Supportive Technique Scale (Ogrodniczuk & Piper, 1999). Adherence monitoring throughout the comparative trial indicated that the therapists conformed to the technical manual and that the interpretive and supportive therapies were well differentiated. Measures
QOR. QOR is defined as a person’s internal enduring tendency to establish certain types of relationships that range along an overall dimension from primitive to mature (Azim, Piper, Segal, Nixon, & Duncan, 1991; Piper & Duncan,
QUALITY OF OBJECT RELATIONS 1999). As the word quality suggests, the kinds of relationships at high levels represent a more favorable and desirable set of circumstances for the person. QOR was assessed with a 1-hr semistructured interview. Lifelong relationships with recent and past significant persons and the immediate relationship with the interviewer were examined. Although the focus was on external relationships, it was assumed that they reflect the internal object representations and conflictual components of the patient’s internal world. The administration and scoring of the interview were guided by a manual (Piper, McCallum, & Joyce, 1996). The scale differentiates five levels of relations. They serve as anchor points. At the mature level, the person enjoys equitable relationships characterized by love, tenderness, and concern for objects of both sexes. There is a capacity to mourn and tolerate unobtainable relationships. At the triangular level, the person is involved in real or fantasized triangular relationships. Competition for one object is inspired by victory over the other object. There is concern for the objects. At the controlling level, the person engages in well-meaning attempts to control and possess objects. Relationships are characterized by ambivalence. At the searching level, the person is driven to find substitutes for a longed for lost object. Substitutes provide a short-lived sense of optimism and self-worth, which is followed by disillusionment and the reexperience of loss. At the primitive level, the person reacts to perceived separation or loss of the object or disapproval or rejection by the object with intense anxiety and affect. There is inordinate dependence on the lost object who provides a sense of identity for the person. To assist the interviewer in assigning scores for each level, there are criteria for each that are organized under four areas. Behavioral manifestations refer to descriptions of an individual’s typical relationship patterns. Affect regulation refers to the type of relationships the person wishes for and engages in to reduce anxiety and increase gratification. Self-esteem regulation refers to the relationships that enhance self-esteem and reduce mortification. Antecedents refer to past events or relationships that predispose a person to a given level. Examples of major criteria for the four areas within each of the five levels are listed in Table 2. As part of the semistructured interview, certain essential questions were raised. These are presented in Table 3. After the interview was completed, the interviewer used instructions and criteria that were presented in the manual to distribute 100 points among the five levels. More weight was given to behavioral manifestations because they are experience near and are usually observable in the interviewer–interviewee interaction. Because of their relatively experience-distant nature and the greater need for inference in evaluating them, affect regulation, self-esteem regulation, and antecedents, in that order, were given less weight. Although one level usually receives the most points, it is common for a second level to also receive considerable points, for example, triangular followed closely by primitive. A bi-
349
modal pattern was consistent with our notion that levels are not discrete categories with rigid boundaries but sets of criteria that blend into each other. It is also consistent with the idea that a person’s usual mode of relating may be at one level, whereas at other times, for example, when under stress, it may be at another level. The overall QOR score, which ranges from 1 to 9, was determined by using a simple algorithm. The points for each level were multiplied by weights assigned to each level: primitive (1), searching (3), controlling (5), triangular (7), and mature (9). The products were summed and divided by 100. Two psychologists and three psychiatrists served as the QOR interviewers. None was a therapist for the patients in the study. To determine rater reliability, we asked all five interviewers to rate 24 cases using audiotapes of the original interviews. The intraclass correlation coefficient, ICC (2,2), for the overall score using two averaged ratings was .68. The QOR scores used in the analyses for all 144 patients were the average of two ratings.
Therapeutic alliance. The therapeutic alliance was defined as the working relationship between the patient and therapist. After each therapy session, the patient and the therapist each made brief ratings of the alliance by means of six, 7-point Likert-type items that ranged from very little to very much. The items focused on whether the patient had talked about private material, felt understood by the therapist, understood and worked with what the therapist had said, felt that the session enhanced understanding, whether the therapist was helpful, and whether the patient and therapist worked well together. Principal components analysis of the six, patient-rated alliance items resulted in one factor. Internal consistency of the items, indicated by Cronbach’s alpha, was high (0.97). Principal components analysis of the six, therapist-rated alliance items also resulted in one factor and its Cronbach’s alpha was also high (0.96). Thus, two scores (patient, therapist) served as the alliance variables. The average level of the alliance was calculated by aggregating the alliance scores across the 20 sessions. The correlation between the patient-rated average level of alliance variable and the therapist-rated average level of alliance variable was significant but not high, r(69) = .36, p = .002. To determine the pattern of change in alliance ratings (patient, therapist), we employed HLM procedures using the HLM 5 statistical software (Raudenbush, Bryk, Cheong, & Congdon, 2001). It was determined that change in the alliance ratings for both patient and therapist were best represented by a linear pattern. The HLM procedure provided a slope estimate for each patient and therapist, which indicated the rate of change in the alliance across sessions and the direction of change. Preliminary analyses revealed a significant amount of variation among patients’ and therapists’ slope estimates. The reliability (indicating the proportion on nonerror variance) of the patients’ slope estimates was .67 and for the therapists’ slope estimates, it was .72.
TABLE 2 Examples of Major Criteria Within the Five Levels and Four Areas of the Quality of Object Relations Scale Levels Areas Behavioral manifestations
Affect regulation
Self-esteem regulation
Antecedents
Primitive 1. Persistent destructive relationships 2. Inordinate attachment to objects 3. Loss, separation, rejection, and disapproval lead to extreme reactions 4. Emotionally laden relationships 5. Splitting 6. Impaired trust 7. Objects valued for their utility 1. Murderous rage toward and fear of annihilation by the object 2. Rage and mortification following narcissistic injury 1. Self-esteem dependent on idealization and/or devaluation of objects 2. Feelings of grandiosity and/or inferiority in relation to objects 1. Abuse 2. Rejection
Searching
Controlling
Triangular
Mature 1. Capacity to express love for both sexes 2. At least one intimate dyadic relationship 3. Capacity for equitable (noncompetitive) relationships 4. Capacity to take risks and be assertive 5. Object constancy despite disappointment 6. Tolerance of bad and appreciation of good in others 1. Capacity to mourn objects not attainable or lost 2. Affection, love, concern, and warmth towards objects of both sexes 1. Self-esteem based on equitable receiving from and giving to objects
1. Repeated “falling in love” 2. Attempts to relive lost relationships 3. Initial optimism followed by pessimism 4. Fear of loss and abandonment 5. Distancing behavior
1. Well-meaning attempts to control 2. Possessive behavior 3. Strong negative reaction and defiance or compliance to other’s control 4. Formal stilted relationships
1. Repetitive rivalrous relationships 2. Inordinate competition 3. Opposite sex viewed as a lover or parent 4. Frequent placation of others in authority 5. Ill at ease with success
1. Craving and longing for the lost object 2. Undue optimism followed by undue pessimism
1. Anger in reaction to inability to control or possess the object 2. Ambivalence and disappointment in relationships 1. Self-esteem maintained by successfully controlling and possessing the object
1. Fear of retaliation by the injured third party 2. Feelings of triumph followed by guilt
1. Positive self-esteem dependent on real or fantasized relationship with a substitute for the lost object
1. Intensive attachment 2. Over indulgence
1. Over control 2. Discouragement of assertiveness
1. Self-esteem dependent on real or fantasized conquering of a third party and acquiring the object
1. Oedipal competition 2. Oedipal triumph and guilt
1. Loving parenting 2. Identification with both parents
351
QUALITY OF OBJECT RELATIONS TABLE 3 Essential Questions for the Five Levels and Three Time Periods of the Quality of Object Relations Scale Levels Time Periods
Primitive
Searching
Controlling
Triangular
Childhood (family of origin)
Do you remember much abuse or disorder in your childhood?
Was there a major event that changed the course of your childhood?
How would you describe your parents in terms of their being controlling of your thoughts, feelings, or behavior?
Adolescence: Period of transition (peers, parents)
Tell me about any incident you can recall where you felt hurt or betrayed by peers at school or socially
Were there times as a teen or young adult where you found yourself “falling in love”? How did those experiences turn out?
How did you engage in “rebellion” as a teen or young adult?
Adulthood (partners, children)
In what way would you say your intimate relationships have been stormy or hurtful?
Looking back over your relationships, have they been a bit like a roller coaster; lots of ups and downs?
Do you find yourself taking care of your partner; sort of looking out for him/her, trying to help him/her with his/her problems?
Was there anything special or unique about your relationship with your (opposite sex) parent that you did not share with your (same sex) parent? Can you tell me about any incident where you and another were both vying for the attention of a third person? How did that work out? How would you say you’ve been competitive with people that you are close to?
Treatment outcome. Outcome assessment at pretherapy and posttherapy was represented by the following13 variables. All but the last variable (independent assessor’s average rating of severity of disturbance) were provided by patient self-report. Interpersonal distress was represented by the overall score from the 64-item Inventory of Interpersonal Problems (Horowitz, Rosenberg, Baer, Ureño, & Villaseñor, 1988). Social, sexual, and family dysfunction were measured by a modification of the Social Adjustment Scale interview (Weissman, Paykel, Siegal, & Klerman, 1971). Depression was assessed by the 13-item short form of the Beck Depression Inventory (Beck & Beck, 1972), anxiety by the 20-item Trait Anxiety Scale (Spielberger, 1983), and general symptomatic distress by the Global Severity Index of the Symptom Checklist–90–Revised (Derogatis, 1977). Self-esteem was measured by Rosenberg’s (1979) 10-item Self-Esteem Scale. Life satisfaction was measured by a single item rated on a 7-point Likert-type scale that ranged from 1 (completely satisfied) to 7 (completely dissatisfied). Mature defenses and maladaptive defenses were represented by two factors that were derived from a principal components analysis of residual change scores (posttherapy–pretherapy) for the three subscales of the 40-item Defensive Style Questionnaire (Andrews, Singh, & Bond, 1993). The patient’s average rating and an independent assessor’s average rating of severity of disturbance for a set of the patient’s individualized target objectives for therapy were also included.
Mature In what ways did members of your family express love and concern for each other?
Were there things about your high school friends that irritated you? How did it affect your relationships? Do you have good male and female friends?
Because of moderate to high correlations between residual change scores (posttherapy–pretherapy) of the outcome variables, a principal components analysis with orthogonal rotation of the residual change scores was used to reduce the 13 variables to a smaller number of outcome factors. Three factors (eigenvalues > 1) emerged, which accounted for 61% of the variance. The factors (General Symptomatology and Dysfunction, Social-Sexual Maladjustment, and Nonuse of Mature Defenses and Family Dysfunction) and their corresponding outcome variables are listed in Table 4. These are the same factors that were generated in the follow-up outcome study (Piper, McCallum, Joyce, Azim, & Ogrodniczuk, 1999). They were retained to maintain continuity.
RESULTS Preliminary Analyses Preliminary analyses involved examining the relationship between each of the alliance variables and QOR. For pattern of alliance, we correlated the QOR scores and the alliance slopes provided by the HLM analyses. For both patient-rated alliance and therapist-rated alliance, the correlations were not statistically significant. Similarly, for average level of alliance, the correlations were not statistically significant. These findings indicate considerable independence between QOR and the alliance variables for the
352
PIPER, OGRODNICZUK, JOYCE TABLE 4 Rotated Factor Solution From the Principal Components Analysis of the 13 Outcome Variables
Outcome Factor and Variable 1. General Symptomatology and Dysfunction (38% of variance) Anxiety General symptomatic distress Depression Self-esteem Interpersonal distress Life Satisfaction Target severity (assessor rated) Target severity (patient rated) Maladaptive defenses 2. Social-Sexual Maladjustment (12% of variance) Social dysfunction Sexual dysfunction 3. Nonuse of Mature Defenses and Family Dysfunction (10% of variance) Mature defenses Family dysfunction
Loading
.84 .80 .78 .75 .74 –.72 .63 .62 .59 .84 .74
–.74 .47
sample of this study. Additional preliminary analyses examined whether there were differences between the 34 high-QOR patients (overall score of 5 or higher) and the 110 low-QOR patients (overall score below 5) on five demographic variables (age, gender, marital status, educational status, employment status) and on presence of Axis I disorders and Axis II disorders. Only marital status was significant, χ2(2, N = 144) = 6.89, p = .032. The effect size r was .22. More high-QOR patients (62%) lived with a partner than low-QOR patients (36%). Also as part of the preliminary analyses, for the interpretive patients and for the supportive patients, we created a correlation matrix that reported the simple Pearson correlations among the eight central variables of this study (patient-rated pattern of alliance, patient-rated average alliance, therapist-rated pattern of alliance, therapist-rated average alliance, QOR, general symptomatology and dysfunction, social sexual maladjustment, and nonuse of mature defenses and family pathology). These two matrixes are presented in the Appendix.
large General Symptoms factor, F(1, 64) = 5.03, p = .03. The effect size r was .26. Figure 1 presents regression line estimates that illustrate the interaction. For high-QOR (5 or higher) patients, the greater the increase in patient-rated alliance over the course of therapy, the better the outcome. For low-QOR (less than 5) patients, the greater the decrease in patient-rated alliance over the course of therapy, the better the outcome. For this interaction effect, it was important to know whether there was a significant relationship between QOR and the initial level of alliance, that is, whether high-QOR and low-QOR patients began at similar levels of alliance. Thus, correlations were calculated between QOR and two alliance variables: patient-rated alliance at Session 1 and patient-rated alliance average for Sessions 1 through 7 (initial third of therapy). Both correlations were small and nonsignificant. This indicated that high-QOR patients and low-QOR patients began with similar levels of alliance ratings. In addition, a significant main effect of QOR, favoring high-QOR patients, was found for the Social-Sexual outcome factor, F(1, 63) = 13.34, p = .001. The effect size r was .42. A similar, nearly significant main effect of QOR was found for the Nonuse of Mature Defenses and Family Pathology outcome factor, F(1, 65) = 3.71, p = .058. The effect size r was .23. QOR, Patient-Rated Average Alliance, and Outcome (Interpretive Therapy Patients) A three-step, hierarchical regression analysis approach was also used to examine the interaction effect of the average level of alliance and QOR on treatment outcome. In each analysis, the order of entry of the variables was average
QOR, Patient-Rated Pattern of Alliance, and Outcome (Interpretive Therapy Patients) For each of the two samples of patients (interpretive therapy, supportive therapy), we used a three-step, hierarchical regression analysis for each of the three outcome factors to examine the interaction effect of pattern of alliance and QOR on treatment outcome. In each analysis, the order of entry of the variables was pattern of alliance (slopes), QOR, and the interaction of the two. In interpretive therapy, a significant interaction effect for patient-rated pattern of alliance and QOR was found for the
FIGURE 1 Interaction of quality of object relations (QOR; high = 5 or greater, low = less than 5) and slope of patient-rated alliance in interpretive therapy.
353
QUALITY OF OBJECT RELATIONS level of alliance, QOR, and the interaction of the two. A main effect of average level of alliance was found for the Nonuse of Mature Defenses and Family Pathology outcome factor, F(1, 66) = 4.15, p = .05. The effect size r was .24. Higher average alliance was associated with better outcome. A main effect of QOR was found for each of the three outcome factors: General Symptoms, F(1, 63) = 6.25, p = .02; Social-sexual, F(1, 63) = 12.74, p = .001; and Nonuse of Mature Defenses and Family Pathology, F(1, 65) = 4.38, p = .04. The effect sizes, r, were .29, .41, and .24, respectively. Higher levels of QOR were associated with better outcome. There were no significant interactions between patient-rated average level of alliance and QOR. To test the first part of Hypothesis 2, Pearson correlations were calculated between patient-rated average alliance and the three outcome factors for high-QOR patients. The correlations were nonsignificant for all three factors: General Symptoms, r(14) = –.30, p = .30; Social-Sexual, r(14) = –.23, p = .40; and Nonuse of Mature Defenses and Family Pathology, r(14) = –.20, p = .45. QOR, Therapist-Rated Pattern of Alliance, and Outcome (Interpretive Therapy Patients) There were no significant findings. Thus, there were no significant interaction effects to support Hypothesis 1: General Symptoms, F(1, 65) = .72, p = .40; Social-Sexual, F(1, 63) = .00, p = .996; and Nonuse of Mature Defenses and Family Pathology, F(1, 65) = .35, p = .56. The effect sizes, r, were .10, .00, and .07, respectively. QOR, Therapist-Rated Average Alliance, and Outcome (Interpretive Therapy Patients) There were no significant regression effects. However, to test the second part of Hypothesis 2, Pearson correlations were calculated between the average level of therapist-rated alliance and the three outcome factors for high-QOR patients. The correlations were nonsignificant for all three factors: General Symptoms, r(1, 14) = –.12, p = .66; Social-Sexual, r(1, 14) = –.23, p = .39; and Nonuse of Mature Defenses and Family Pathology, r(1, 14) = .04, p = .89. QOR, Patient-Rated Pattern of Alliance, and Outcome (Supportive Therapy Patients) There were no significant findings. QOR, Patient-Rated Average Alliance, and Outcome (Supportive Therapy Patients) A significant main effect for patient-rated average alliance was found for the General Symptoms outcome factor, F(1, 68) = 10.52, p = .002. The effect size r was .37. Higher average alliance was associated with better outcome.
QOR, Therapist-Rated Pattern of Alliance, and Outcome (Supportive Therapy Patients) A significant main effect for pattern of alliance was found for the Social-Sexual outcome factor, F(1, 68) = 5.42, p = .023. The effect size r was .27. The greater the increase in alliance over time, the better the outcome. QOR, Therapist-Rated Average Alliance, and Outcome (Supportive Therapy Patients) There were no significant findings.
DISCUSSION The findings of this study have provided further evidence of the importance of QOR as a moderator of relationships between central variables in short-term individual psychotherapy. Specifically, QOR emerged as a moderator of the relationship between patient-rated pattern of alliance and outcome in general symptomatology and dysfunction in short-term, interpretive individual psychotherapy. This outcome factor accounted for the most variation in our set of outcome variables. For high-QOR patients, the greater the increase in patient-rated alliance over therapy, the better the outcome. For low-QOR patients, the greater the decrease in patient-rated alliance over therapy, the better the outcome. Given the consistently reported finding in the literature of a direct relationship between the average level of alliance and favorable outcome, the finding for low-QOR patients appears to be counterintuitive. Why should a decreasing level of alliance be related to better outcome? An explanation for the findings that focuses on the nature of interpretive psychotherapy and the nature of QOR can be formulated. Interpretive therapy is a challenging form of treatment for patients. It is essentially patient driven. The patient is responsible for beginning each session and deciding what follows. Neither the content nor process is predetermined, which creates a climate of uncertainty and unpredictability. The patient is expected to address sensitive and painful topics. When the therapist does speak, it is often in the form of a confrontation or interpretation about conflictual events that to some degree have been unconscious. The climate of interpretive therapy is intended to promote regressive processes including transference reactions. The inappropriate aspects of transference reactions can then be examined and ideally better understood. Different types of patients can be expected to respond to the challenges of interpretive work in different ways. Low-QOR patients have a history of disappointing relationships. Although new relationships are usually met with high expectations, they usually turn out to be nongratifying and abusive. The patient often ends up feeling rejected and disillusioned. In the absence of therapy, the pattern repeats itself
354
PIPER, OGRODNICZUK, JOYCE
in their everyday lives. In interpretive therapy, the pattern can be expected to repeat itself, at least in part, for a period of time. Although low-QOR patients begin interpretive therapy with high expectations, as reflected in moderate to high alliance ratings, the demanding climate, in particular the transference focus of the therapist, can be perceived as critical and threatening as reflected in lower alliance ratings over time during therapy. This pattern of alliance is important for the outcome of therapy because it reveals the problematic relationship patterns that are associated with the low-QOR patient’s difficulties. Greater benefit from therapy is likely to follow when these issues are reexperienced in therapy and can be fully explored. Other low-QOR patients, however, respond to the challenges of therapy in less characteristic ways. These patients may come to idealize the relationship with the therapist to tolerate the anxiety caused by the transference work in therapy. This is reflected in increasingly higher ratings of the alliance throughout therapy. Such a reaction to a difficult interpersonal relationship is not characteristic of these patients. Thus, they do not reexperience their negative relationship patterns in therapy and defensively may focus on more superficial issues in therapy to preserve their perceived positive relationship with the therapist. The result is that the patient’s underlying issues are not dealt with fully and continue to trouble the patient. When the termination of the short-term therapy inevitably occurs, their idealized conception of the therapist likely becomes shattered with an associated exacerbation of symptoms. In contrast, high-QOR patients have a history of experiencing relatively positive relationships. Nevertheless, they tend to experience conflicts concerning interpersonal control and competition with others. In interpretive therapy, the patient similarly can be expected to experience conflicts with the therapist over issues related to control and competition. The transference reaction and its repetitive nature are highlighted, interpreted, and explored by the therapist. Most high-QOR patients can be expected to appreciate this exploration as reflected in increasing alliance ratings. A result of examining the patient’s core relationship issues is greater benefit from therapy. Some high-QOR patients may respond to the challenges of interpretive therapy differently. They may resent the therapist’s interpretation of their controlling and competitive behavior. Essentially, they are unable to relinquish control to the therapist and continue to relate to the therapist in a competitive manner. These patients end up devaluing their relationship with the therapist as reflected in their decreasing pattern of alliance ratings. As a result, the patient is less likely to benefit from therapy. In summary, we suggest that it is important to consider whether the therapy relationship reflects the patient’s characteristic relationship patterns and reactions outside of therapy. A therapy relationship that is consistent with the patient’s typical pattern allows the patient and therapist to work on the primary issues underlying the patient’s problems and difficulties.
Greater benefit from therapy is expected to follow. A therapy relationship that is inconsistent with the patient’s characteristic relationships and reactions does not allow the therapist to work with the patient’s core relationship issues. They will likely continue to trouble the patient after treatment. In addition to the interaction effect involving QOR and patient-rated pattern of alliance, a main effect for patient-rated average level of alliance was found in interpretive therapy. The greater the average level of alliance, the greater the benefit on the outcome factor representing nonuse of mature defenses and family dysfunction. A similar main effect for patient-rated average alliance was found in supportive therapy. The greater the average level of alliance, the greater the benefit on the outcome factor representing general symptoms and dysfunction. These two findings are consistent with the extensive literature that has reported a direct relation between average level of alliance and favorable outcome (Horvath & Symonds, 1991; Martin et al., 2000; Orlinsky et al., 2004). Also consistent with substantial evidence from previous clinical trials that the research team (Piper, Azim, Joyce, McCallum, Nixon, et al., 1991; Piper et al., 1994) and others (Cook et al., 1995; Ford et al., 1997; Honig et al., 1997) have conducted were the main effects involving QOR. The greater the QOR, the more favorable the outcome. As indicated earlier in this article, we believe that in general, high-QOR patients are better able tolerate and work with the demanding, depriving, and anxiety-arousing features of interpretive therapy. No significant findings emerged for the therapist-rated alliance variables in interpretive therapy. The direct relationship between increase in alliance and outcome for low-QOR patients in the Piper et al. (1995) study did not emerge as hypothesized in this study. A lack of consistency between findings for patient-rated alliance scores and therapist-rated alliance scores has characterized the average alliance literature. It appears to characterize the more limited literature of pattern of alliance as well. In this study, the correlation between the patient-rated and therapist-rated average level of alliance scores was small, with less than 13% shared variance. It is possible that the patients and therapists were responding to different cues. In the case of supportive therapy, the only other significant finding involved the Social-Sexual outcome factor. There was a main effect for therapist-rated pattern of alliance. The more the therapist perceived an increase in the alliance, the better the outcome. In the more gratifying, less confrontational climate of supportive therapy, the therapist’s ratings of alliance likely reflected general progress in working on the everyday problems of the patient. The absence of an interaction effect is consistent with the considerable differences in the nature of the two forms of therapy and the greater likelihood of negative transference developing in interpretive therapy. The presence of a greater number of significant effects for interpretive therapy, in particular the interaction effect involving the pattern of alliance, may in part reflect the impact
355
QUALITY OF OBJECT RELATIONS that assessing QOR has on the treatment process. A pretherapy examination of the nature and quality of important relationships in the patient’s life sets the stage for a sensitive examination of relationships in interpretive therapy, especially the transference relationship. Thus, conducting a QOR assessment may constitute a useful method of preparing patients for interpretive therapy. This is a benefit in addition to using the QOR assessment to select patients (high-QOR) who are likely to establish a stronger alliance and achieve better outcomes in interpretive therapy. Recent reviews of object relations theory and research (Huprich & Greenberg, 2003; Piper & Duncan, 1999) have highlighted differences among various conceptions of object relations such as whether they refer to internal processes, external behavior, or both, and the many different measures of object relations. Researchers who have focused on internal events with projective measures have been more interested in concepts of psychopathology. Researchers who have focused on external events with behavioral measures have been more interested in important clinical events such as the therapeutic alliance and therapy outcome. The field would benefit from future research that clarified the relationships among the different concepts and measures and their relationships to psychopathology and clinical events such as the strength of the therapeutic alliance and therapy outcome. Several limitations of this study should be noted. Most important, our comparative trial was not specifically designed to evaluate the question of whether QOR moderated the relationship between the pattern of the alliance and treatment outcome. Because it was essentially an exploratory study that produced counterintuitive findings, replication is needed. It is of course possible that the significant finding was due to sampling error. In addition, the ideas that were offered to explain the particular moderator effects were based on theory rather than a process analysis of actual patient and therapist behavior during the therapy sessions. Finally, the rater reliability of the assessment of QOR was satisfactory but not as high as it has been found to be in subsequent studies. In conclusion, this study provided additional evidence of the strength of QOR as a moderator variable. In combination with considerable evidence of QOR as a moderator variable in previous studies, it suggests that this possibility should be considered in future studies of short-term psychotherapy. ACKNOWLEDGMENTS This research project was supported by Grant 6609–1765–60A from the National Health Research and Development Program, Health and Welfare Canada, Ottawa, Ontario, Canada. We thank Douglas R. Ginter, John S. Rosie, and David Shih, who served as assessors; J. Fyfe Bahrey, Satwant K. Duggal, Andrea Duncan, Scott C. Duncan, Dianne R. Kipnes, William J. M. Nickerson, John G. O’Kelly, and B. Jill Spaner, who served as therapists; and Hillary Morin, who served as research coordinator.
REFERENCES American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author. Andrews, G., Singh, M., & Bond, M. (1993). The defense style questionnaire. Journal of Nervous and Mental Disease, 181, 246–256. Azim, H. F. A., Piper, W. E., Segal, P. M., Nixon, G. W. H., & Duncan, S. (1991). The quality of object relations scale. Bulletin of the Menninger Clinic, 55, 323–343. Beck, A. T., & Beck, R. W. (1972). Screening depressed patients in family practice: A rapid technic. Postgraduate Medicine, 52, 81–85. Connolly, M. B., Crits-Christoph, P., Shappell, S., Barber, J. P., Luborsky, L., & Shaffer, C. (1999). Relation of transference interpretations to outcome in the early sessions of brief supportive-expressive psychotherapy. Psychotherapy Research, 9, 485–495. Cook, B., Blatt, S. J., & Ford, R. Q. (1995). The prediction of therapeutic response to long-term intensive treatment of seriously disturbed young adult inpatients. Psychotherapy Research, 5, 218–230. de Carufel, F. L., & Piper, W. E. (1988). Group psychotherapy or individual psychotherapy. Patient characteristics as predictive factors. International Journal of Group Psychotherapy, 38, 169–188. Derogatis, L. R. (1977). SCL–90 administration, scoring, and procedures manual I. Baltimore: Johns Hopkins University Press. First, M. B., Gibbon, M., Williams, J. B., & Spitzer, R. L. (1990). Mini-SCID. Toronto, Ontario, Canada: Multi-Health Systems. First, M. B., Gibbon, M., Williams, J. B., & Spitzer, R. L. (1991). SCID–II PQ and AutoSCID II. Toronto, Ontario, Canada: Multi-Health Systems. Ford, J. D., Fisher, P., & Larson, L. (1997). Object relations as a predictor of treatment outcome with chronic posttraumatic stress disorder. Journal of Consulting and Clinical Psychology, 65, 547–559. Gunderson, J. G., Najavits, L. M., Leonhard, C., Sullivan, C. N., & Sabo, A. N. (1997). Ontogeny of the therapeutic alliance in borderline patients. Psychotherapy Research, 7, 301–309. Hellerstein, D. J., Rosenthal, R. N., Pinsker, H., Samstag, L. W., Muran, J. C., & Winston, A. (1998). A randomized prospective study comparing supportive and dynamic therapies. Journal of Psychotherapy Practice and Research, 7, 261–271. Høglend, P. (1993). Transference interpretations and long-term change after dynamic psychotherapy of brief to moderate length. American Journal of Psychotherapy, 47, 494–507. Høglend, P., & Piper, W. E. (1995). Focal adherence in brief psychotherapy: A comparison of findings from two independent studies. Psychotherapy, 32, 618–628. Honig, M. S., Farber, B. A., & Geller, J. D. (1997). The relationship of pretreatment representations of mother to early treatment representations of their therapist. Journal of American Academy of Psychoanalysis, 25, 357–372. Horowitz, L., Rosenberg, S. E., Baer, B. A., Ureño, G., & Villaseñor, V. S. (1988). Inventory of interpersonal problems: Psychometric properties and clinical applications. Journal of Consulting and Clinical Psychology, 56, 885–892. Horvath, A. O., & Symonds, D. B. (1991). Relationship between working alliance and outcome in psychotherapy: A meta-analysis. Journal of Counseling Psychology, 38, 139–149. Huprich, S. K., & Greenberg, R. P. (2003). Advances in the assessment of object relations in the 1990s. Clinical Psychology Review, 23, 665–698. Joyce, A. S., & Piper, W. E. (1990). An examination of Mann’s model of time-limited individual psychotherapy. Canadian Journal of Psychiatry,35, 41–49. Klee, M. R., Abeles, N., & Muller, R. T. (1990). Therapeutic alliance: Early indicators, course, and outcome. Psychotherapy, 27, 166–174. Mallinckrodt, B. (1996). Change in working alliance, social support, and psychological symptoms in brief therapy. Journal of Counseling Psychology, 43, 448–455. Mann, J. (1973). Time-limited psychotherapy. Cambridge, MA: Harvard University Press.
356
PIPER, OGRODNICZUK, JOYCE
Martin, D. J., Garske, J. P., & Davis, M. K. (2000). Relation of the therapeutic alliance with outcome and other variables: A meta-analytic review. Journal of Consulting and Clinical Psychology, 68, 438–450. Miller, J. M., Courtois, C. A., Pelham, J. P., Riddle, P. E., Spiegel, S. B., Gelso, G. J., et al. (1983). The process of time-limited therapy. In C. Gelso & D. H. Johnson (Eds.), Explorations in time-limited counselling and psychotherapy (pp.175–184). New York: Teachers College Press. Ogrodniczuk, J. S., & Piper, W. E. (1999). Measuring therapist technique in psychodynamic psychotherapies: Development and use of a new scale. Journal of Psychotherapy Practice and Research, 8, 142–154. Ogrodniczuk, J. S., Piper, W. E., Joyce, A. S., & McCallum, M. (1999). Transference interpretations in short-term dynamic psychotherapy. Journal of Nervous and Mental Disease, 187, 572–579. Orlinsky, D. E., Ronnestad, M. H., & Willutzki, U. (2004). Fifty years of psychotherapy process-outcome research: Continuity and change. In M. J. Lambert (Ed.), Bergin and Garfield’s handbook of psychotherapy and behavior change (5th ed., pp. 307–389). New York: Wiley. Paivio, S. C., & Patterson, L. A.(1999). Alliance development in therapy for resolving child abuse issues. Psychotherapy, 36, 343–354. Piper, W. E., Azim, H. F. A., Joyce, A. S., & McCallum, M. (1991). Transference interpretations, therapeutic alliance and outcome in short-term individual psychotherapy. Archives of General Psychiatry, 48, 946–953. Piper, W. E., Azim, H. F. A., Joyce, A. S., & McCallum, M. (1993). Manual for time-limited, short-term individual therapies: Interpretive and supportive forms. Unpublished manuscript. Piper, W. E., Azim, H. F. A., Joyce, A. S., McCallum, M., Nixon, G. W. H., & Segal, P.S. (1991). Quality of object relations vs. interpersonal functioning as predictors of therapeutic alliance and psychotherapy outcome. Journal of Nervous and Mental Disease, 179, 432–438. Piper, W. E., Boroto, D. R., Joyce, A. S., McCallum, M., & Azim, H. F. A. (1995). Pattern of alliance and outcome in short-term individual psychotherapy. Psychotherapy, 32, 639–647.
Piper, W. E., & Duncan, S.C. (1999). Object relations theory and short-term dynamic psychotherapy: Findings from the Quality of Object Relations Scale. Clinical Psychology Review, 19, 669–685. Piper, W. E., Joyce, A. S., Azim, H. F. A., & Rosie, J. S. (1994). Patient characteristics and success in day treatment. Journal of Nervous and Mental Disease,182, 381–386. Piper, W. E., Joyce, A. S., McCallum, M., & Azim, H. F. A. (1993). Concentration and correspondence of transference interpretations in short-term psychotherapy. Journal of Consulting and Clinical Psychology, 61, 586–595. Piper, W. E., Joyce, A. S., McCallum, M., & Azim, H. F. A. (1998). Interpretive and supportive forms of psychotherapy and patient personality variables. Journal of Consulting and Clinical Psychology, 66, 558–567. Piper, W. E., McCallum, M., & Joyce, A. S. (1996). Manual for assessment of quality of object relations. Unpublished manuscript. Piper, W. E., McCallum, M., Joyce, A. S., Azim, H. F., & Ogrodniczuk, J. S. (1999). Follow-up findings for interpretive and supportive forms of psychotherapy and patient personality variables. Journal of Consulting and Clinical Psychology, 67, 267–273. Piper, W. E., McCallum, M., Joyce, A. S., Rosie, J. S., & Ogrodniczuk, J. S. (2001). Patient personality and time-limited group psychotherapy for complicated grief. International Journal of Group Psychotherapy, 51, 525–552. Raudenbush, S., Bryk, A., Cheong, Y., & Congdon, R. (2001). HLM 5 hierarchical linear and nonlinear modeling. Lincolnwood, IL: Scientific Software. Rosenberg, M. (1979). Conceiving the self. New York: Basic. Spielberger, C. D. (1983). Manual for the State-Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press. Weissman, M. M., Paykel, E. S., Siegel, R., & Klerman, G. L. (1971). The social role performance of depressed women: A comparison with a normal sample. American Journal of Orthopsychiatry, 41, 390–405. Westerman, M. A., Foote, J. P., & Winston, A. (1995). Change in coordination across phases of psychotherapy and outcome: Two mechanisms for the role played by patients’ contribution to the alliance. Journal of Consulting and Clinical Psychology, 63, 672–675.
APPENDIX Correlation Matrix for Alliance Variables, QOR, and Outcome Factors for Patients in Interpretive Therapy (Above the Diagonal) and for Patients in Supportive Therapy (Below the Diagonal) Patient-Rated Alliance Slope Patient-rated alliance slope Patient-rated alliance average Therapist-rated alliance slope Therapist-rated alliance average QOR score General symptomatology Social-sexual maladjustment Nonuse of mature defenses
Patient-Rated Alliance Average
Therapist-Rated Alliance Slope
Therapist-Rated Alliance Average
Nonuse of Mature Defenses
QOR Score
General Symptomatology
Social-Sexual Maladjustment
–.16
.07
.21
–.05
— .27*
.26*
.22
.14
—
.16
.36**
.02
–.16
–.24*
.13
.24*
.04
—
.34**
–.13
.15
.01
.11
.02 –.02
.25* .23
.14 –.14
— .00
.03 —
–.10 –.30**
–.11 –.26*
–.01 –.40**
–.15
–.37**
–.02
–.22
.02
.01
–.19
–.09
–.06
–.17
–.11
–.04
.27*
—
.32**
.26*
–.08
.52**
—
.01
.05
.38**
.31**
Note. Because of missing data, n varies from 66 to 72. QOR = quality of object relations. *p ≤ .05, **p ≤ .01.
William E. Piper Department of Psychiatry University of British Columbia 2255 Wesbrook Mall Vancouver, British Columbia, Canada E-mail:
[email protected]
Received July 28, 2003 Revised October 15, 2003
V6T 2A1
—