Anda di halaman 1dari 9

Jounial of Consulting and Clinical Psychology 1986, Vol. 54, No.

1, 39-47

Copyright 1986 by the American Psychological Association, Inc. 0022-006X/86/J00.75

Advent of Objective Measures of the Transference Concept


Lester Luborsky, Paul Crits-Christoph, and James Mellon
Department of Psychiatry, University of Pennsylvania

This is the first comprehensive review of research on measures of the transference since the concept was first put forward at the turn of the century. Such research is of two kinds: Q-sort questionnaire measures and psychotherapy process measures. The first type of measure, although reliable, has lacked validity information. The newer psychotherapy process measures of the last decade, for example, the core conflictual relationship theme method (CCRT), provide evidence for their reliability and validity. Considerable comparability was found between data from the CCRT method and operationalized propositions from nine of Freud's observations about transference. For example, the CCRT in relationship to the therapist was found to be highly similar to the CCRT in relationship to other people. Methodological issues and proposals that could further advance research on transference are examined.

For more than three quarters of a century there has been considerable consensus by psychoanalytic clinicians that transference is Freud's grandest clinical concept (Breuer & Freud, 1893-1895/ 1966; Freud, 1912/1966). During this period the concept has been continually used in everyday practice in psychoanalysis and in all of the psychodynamically oriented psychotherapies. In these therapies the nature of transference, consistent with Freud's definition (Freud, 1912/1966), refers to the patient's expression of attitudes and behavior derived from early conflictual relationships with significant parental figures in the current relationship with the therapist. In each session in such therapies, the therapists make inferences about the state of the transference as a basis for understanding the patient and for guiding their interventions (Luborsky, 1984). Despite this heavy clinical reliance on the concept, there has been until recently an impoverishment in clinically useful and even research-useful clinical quantitative methods. These research methods fall into two categories: (a) questionnaire measures and (b) psychotherapy process measures. The questionnaire approach began about 35 years ago but never got very far; the process measure approach, although new, already shows clinical and research value.

Questionnaire Measures of Transference


To achieve an operational version of the concept, transference was purposely denned by Chance (1952) in terms of the similarity

This work was supported in part by United States Public Health Service Research Scientist Award MH 40710 to Lester Luborsky. It was first specifically supported in part by a grant from the Fund for Psychoanalytic Research of the American Psychoanalytic Association. More recently some of the research has been supported in part by National Institute of Mental Health Grants MH40472 and MH39673. Among those who played a vital role in the research were Arthur Auerbach, Stephanie Ming, Keith Alexander, Frederic J. Levine, Anna Rose Childress, Kenneth D. Cohen, Paul van Ravenswaay, Leslie Alexander, Laura Dahl, Fu-Chin Lee, Lynn Tomko, KatherineCrits-Christoph, and Anita V. Hole. Correspondence concerning this article should be addressed to Lester Luborsky, Department of Psychiatry, 207 Piersol Building, Gl, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania 19104.
39

between the patient's description of a significant parent and the patient's description of the psychotherapist. In Fiedler's studies (e.g., Fiedler & Senior, 1952), transference was denned in terms of a comparison of the patient's description of the ideal person with the patient's prediction of the therapist's self-description and by both of these with similar measures completed by the therapist. In Apfelbaum's (1958) method, transference was intended to be tapped by a questionnaire (Q-sort) on the patient's expectations about the qualities of the therapist who would later be assigned to the patient. The patients were grouped in terms of three types of expectations reported in their initial (preassignment) Q-sort: Cluster A (therapist will give nurturance); Cluster B (therapist will be a model); and Cluster C (therapist will be a critic). Each of these types tended to be maintained to the end of treatment, as indicated by high test-retest reliability. Such stability was listed by Freud (1912/1966)asa characteristic of transference. Rawn (1958,1981) developed Q-sort-based scales and applied them to four sessions of one patient's analysis. He took the unusual further step of comparing these results with those from clinical observations and noted signs of convergence. Crisp (1964a, 1964b, 1966) continued this line of research with Qsorts of questionnaire items. The items rated were about father and therapist figures; the estimate of transference was based on a comparison of these ratings. One finding indicated that attitudes toward the therapist tend to change with or to precede changes in symptoms, Subotnick (1966a, 1966b) developed the method further. He used two separate sets of Q-sorts: attitudes toward parents and attitudes toward the therapist at various points in therapy. Similarity was found between the attitudes toward parents and therapist (by high loadings on the factors common among the Q-sorts). All of these six sets of studies are based on the questionnaire approach, usually in the form of the Q-sort method. They suffer from the use of a method with questionable validity. These measures need to be compared with measures of transference based on psychotherapy sessions of the kind described in the next section. Furthermore, as Fisher and Greenberg (1977) pointed out, the studies do not distinguish appropriate from inappropriate attitudes toward the therapist, a distinction usually considered to be required by the concept of transference.

40

L. LUBORSKY, P. CRITS-CHRISTOPH, AND J. MELLON

Psychotherapy Process Measures of Transference


Clearly a more direct approach to the concept of transference was necessary. Because the concept was originally derived from psychotherapy, measures might more appropriately be constructed from sessions, such as the following three types of methods: systematic clinical formulation, rating, and content coding.

Systematic Clinical Formulation Methods


A formulation, as used here, means a description of a patient's main relationship patterns. Formulations were routinely made in the Menninger Foundation Psychotherapy Project (Wallerstein, 1985; Walleistein & Robbins, 1956) by a termination research team from an evaluation of the therapist's process notes and from interviews with the patient, the therapist, and the patient's relatives. The Chicago Consensus Group (Seitz, 1966) performed the best known study of free discursive formulations of transference by independent groups of clinicians. They reported a lack of consensus, but in fact, any conclusion was difficult to justify because their method was not capable of coping objectively with the comparison of different judges' formulations. The Mayman and Fans (1960) formulation system was based on clinical recognition of common elements across several early memories. Although their system is an important precursor to present day methods, it lacks explicit principles for guiding clinical judgments and making estimates of reliability. The basic observation that an enduring set of relationship patterns are repeated throughout a person's life has also been included in script theory. Carlson's (1981, p. 502) account of Tomkins's (1979) script theory described the script as "the individual's rules for predicting, interpreting, responding to, and controlling experiences governed by a 'family' of related scenes." Tomkins's (1979, p. 228) theory also identifies one "nuclear scene," or sometimes several, that manifest these rules. Carlson (1981) provided an example of a person's nuclear scene that recurs after 30 years. Her interpretation of this scene is the usual psychodynamic clinical one, that is, that the scene is a pattern setter for later relationship episodes. A more precise comparison of Tomkins's script theory with the core conflictual relationship theme (CCRT) and other similar methods would be facilitated by a scoring system for the script theory.

ference likely, r - .46, p < .01) when the judgment was based on the amount of transference expressed in relation to each person referred to in the segment, (c) The segments with high transference showed much more affect than did the segments with low transference (Lower, Escoll, Little, & Ottenberg, 1973). (d) The concept space used by each of eight psychoanalyst judges could be reliably identified (Luborsky, Crabtree, Curtis, Ruff, & Mintz, 1975) by a factor analysis of their ratings of 23 transference-related concepts on the same segments, (e) Consistency over time was found for ratings of transference and resistance made on a postsession checksheet (Graff & Luborsky, 1977) for four psychoanalytic patients, two more improved and two less improved. The two more improved patients showed a pattern of increasing amount of transference. These results imply that a revision is needed in the common theory that in successful psychoanalysis, transference is eradicated.

Content Coding Methods


Two recently developed measures of transference are closer to representing the clinical concept than any described so far. Both of these use guided clinical judgment: Clinicians must follow a set procedure and are not free to make formulations in whatever categories they find most compatible. The method by Gill and Hoffman (1982b) estimates the frequency of the patient's experience of the relationship with the therapist; the method by Luborsky (1976, 1977) estimates the frequency and the content of the CCRT.

Patient's Experience of the Relationship With the Therapist Method (PERT)1


This guided clinical judgment method using transcripts of psychotherapy (Gill & Hoffman, 1982a, 1982b) provides guidelines for a tally of the frequency of communications regarding the patient's manifest experience of the relationship with the therapist as well as presumed implicit references to the experience. It has two main divisions: (a) experiences of the relationship that are manifestly about the relationship with the therapist (r) and (b) experiences of the relationship that are not manifestly about the relationship with the therapist or have no specific designation (x). Further divisions include communications that express awareness of the presumed parallel between experiences outside and experiences inside the session (e.g., xr). Another category is derived from the inference of a judge rather than from what is directly expressed by the patient (e.g., Jxr). The PERT system also estimates the degree to which the therapist's interventions deal with the main aspects of the PERT (e.g., XR). For both the PERT and the therapist's interventions, the judge also indicates the basis for the inferences. One kind of reliability has been reported so far: the agreement between two judges on the total for each code within the entire session. The correlation for the r code was .89 and for the xr code, .63. An example of a session as scored by the PERT system is the best way to further explicate the method (later we will refer back to the same example to illustrate the CCRT scoring). This example is from the treatment of a young man in psychoanalysis

Rating Methods
In the clinical formulation methods just described, agreement is harder to estimate than it is in the studies using rating methods. However, the first rating method studies were limited to ratings of the amount but not the type of transference. One of these (Strupp, Chassan, & Ewing, 1966) reported only slight agreement among five independent judges in their ratings of amount of transference in whole sessions. In the first study of the Analytic Research Group of the Institute of the Pennsylvania Hospital (Luborsky, Graff, Pulver, & Curtis, 1973) of one psychoanalytic patient, clinicians judged the amount of transference and related variables in 30 five-min segments. Several findings are noteworthy: (a) The agreement was low (r - .26) when simply the amount of overall transference in a segment was rated, (b) The agreement was higher (for trans-

This is our acronym for the Gill and Hoffman (1982b) system.

OBJECTIVE MEASURES OF TRANSFERENCE (from GUI & Hoffman, 1982b, p. 151 ff). Judges chose these segments from the transcript as relevant to the PERT. Each of the following segments is a narrative episode about a relationship with a different person. In order to facilitate the comparison with the CCRT method, the person with whom the patient is interacting in each episode is named after the relationship episode (RE) number, as would be done for the CCRT system. RE no. 1: A guy (man friend). He came over to drink beer, and to have this conversation which was a little difficult. I pretended to be enjoying it, enjoying him, you know, in the spirit of good fellowship and shit and stuff, but I really wanted to bewell, I didn't want to be reading, but you know, I felt that this was the thing that, that was keeping me from reading and that hassled me. I really nicking resented it a lot. You know among my friends, they're respecting and always have really respected my wanting to do my own thing. . . But you know, with a guy like this (clears throat), he's just in another world totally from that. And, you know, he wouldn't understand if I said that, you know, he would be insulted and that kind of shit You know it was kind of a hassle. This would be provisionally scored an x because it is a communication about matters manifestly other than the patient's experience of the relationship with the therapist. According to Gill and Hoffman (1982b), the episode may allude to resentment of the involvement that the patient feels that the analyst demands of him. However, such an inference based on a single episode is not regarded as supported until more direct evidence for it emerges. RE no. 2: Therapist. This morning I, like didn't particularly feel like coming here, you know. Because like, I don't know, I felt some kind of, you know, I felt like I didn't need it. I guess I was just, you know, my spirits were a little raised. If only now I could get out of the bag of feeling that I have t o . . . . A coding of Jxr for RE no. 1 can now be justified on the basis of the episode about the therapist, and that would be the final score. The previous x episode can be thought of retrospectively as including implicit elaborations of the present issue about the experience of the relationship with the therapist, elaborations that the patient resists making because he feels that he may insult the therapist. RE no. 3: Woman (a potential date). This woman bag is real bad for me. The woman that I'm going ape shit over now is a woman I've never even seen. I think I told you about the telephone operator who I made a date with. And 1 keep calling her back, but she's never home and I talk to her roommate. That's been a big hassle because I keep, you know, hoping she'll be there, that she'll talk to me. I've never even seen this girl, I mean, I don't know anything about her, you know. Again a coding of Jxr appears to be warranted, and a similar theme appears to be further exemplified. Again the basis for the Jxr coding is the explicit statement by the patient about the therapist (RE no. 2). A bit later, more about the same episode with the woman (a potential date) is presented: When I finally got through to her roommate yesterday and found out that she wasn't going to be in, like all the woman obligations just went off me. I knew that there was nothing I could do to find a woman and, you know, there was kind of a relief.

41

Another Jxr coding would be given for the following example. RE no. 4: Some girl (fantasy). The fantasy that really turned me on was this, you know, heterosexual sadistic fantasy. I mean just really fucking the shit out of some girl and her being, you know, prostrate before my, you know, my massive oigan. And uh, I described to you before how, you know, the idea of girls doing things that are really, had a lot of indignity about them. RE no. 5: Therapist. Well, now I'm getting that same feeling that, you know, I'm sort of talking about worthless shit. Because, and you know, my basis for thinking that is the fact that you haven't said anything. Jeez, we go through this same nonsense every session, it's just amazing to me. I'm sort of ashamed that my mind isn't a little more creative, to think of different hassles. You know, it's sort of boring going through the same hassle four times a week, for what at this point seems like a timeless period. For each example, possible responses by the therapist are indicated by Gill and Hoffman (1982b) that illustrate the types of inferences that their judges make, for example, after RE no. 5, the therapist might have said, "Maybe you feel that I'm sitting here enjoying my power over you while you are forced to go through all kinds of contortions to gratify me." These examples, although brief, do illustrate the PERT system. This system does not provide a quantitative method for scoring the inferences about the types of content of the PERT, although the judge does write out inferences about the content and, at the end, a summary account of the transference and of the therapist's contribution to its explication.

Core Conjlictual Relationship Theme Method (CCRT)


Description of the method. This method (Luborsky, 1976) is also a system to guide clinical judgment of the content of the central relationship patterns in psychotherapy sessions. Although the CCRT judges read the entire session, the primary data to be scored are narrative episodes about relationships that patients commonly tell during psychotherapy sessions. These are parts of sessions in which the patient explicitly narrates experiences in relationships. Typical narratives are about father, mother, brothers, sisters, friends, bosses, and the therapist. These relationship episodes are identified by a separate set of independent judges before the transcripts are given to the CCRT judges. A minimum of 10 relationship episodes are usually used as a basis for scoring the CCRT. The use of narratives as well as some of the scoring categories bear a resemblance to the Thematic Apperception Test (Murray, 1938). The CCRT judge reads the relationship episodes in the transcript and identifies the types of each of the three components within each episode: (a) the patient's main wishes, needs, or intentions toward the other person in the narrative; (b) the responses of the other person; and (c) the responses of the self. Within each component the types with the highest frequency across all relationship episodes are identified; their combination constitutes the CCRT. The steps in the CCRT method represent a formalization of the usual inference process of clinicians in formulating transference patterns. The clinician-judge first identifies the wishes and responses to the wishes in each of the REs and from these makes a preliminary CCRT formulation (Steps 1 and 2), and

42

L. LUBORSKY, P. CRITS-CHRISTOPH, AND J. MELLON

then the same judge reidentifies and reformulates (Steps 1' and!1): Step 1. Identify the types of wishes (W) and responses (RO, response from other; RS, response from self) in each relationship episode (RE). Step 2. Formulate a preliminary CCRT based on the frequency of each of the types of each component. Step 1'. Reidentify, where needed, the types of W, RO, and RS based on the Step 2 preliminary CCRT. Step 2'. Reformulate, where needed, based on the recount of all Ws, ROs, and RSs in Step 1'. It should be noted that the CCRT judges work independently of each other. Judges are trained by first reading the CCRT manual (Luborsky, 1983) and trying several standard practice cases, receiving feedback from the research team about their performance after each one. Although we have preferred to use experienced clinicians with a psychoanalytic orientation as judges, some graduate students have also performed well as judges because the task does not require that the judge be committed to a particular school of therapy. In fact, the CCRT may belong in a family of related conceptualizations of relationship patterns, which includes Tomkins's (1979) concept of a nuclear script, Meichenbaum and Gilmore's (1984) concept of core organizing principles, and other concepts reviewed by Singer (1984). The simplest way to illustrate the CCRT scoring system as

well as its similarity and contrast with the PERT system is to briefly apply the CCRT system in abbreviated form to the five narratives from Gill and Hoffman (1982b) reproduced earlier (see Table 1). As can be seen in the CCRT formulation at the bottom of Table 1, the most frequent form of the wish component in these five REs is to be free of obligations imposed by others (in Step 2' each judge identified this wish in four of the five REs). The most frequent RO is that the other person does not respond to the wish. The most frequent RS is to feel hassled and to feel compelled to give in. Reliability of the CCRT (agreement among judges). Even in the first trials of the CCRT method (Luborsky, 1977), considerable agreement among judges was found. In Levine and Luborsky (1981), 16 graduate psychology student judges individually scored the CCRT for one patient (Mr. B.N.). When the scoring of each of these judges was compared with composite scoring of 4 research judges (who also individually scored the CCRT), good agreement was found (average correlation of .88). Furthermore, agreement was shown by the method of mismatched cases: Agreement was greater when the components to be compared were drawn from the Mr. B.N. case itself rather than from two other purposely mismatched cases. A larger reliability study was carried out on eight patients each scored by three independent judges (Luborsky, Crits-Christoph, et al., 1985). Determining agreement involved the use of

Table 1 Comparison of Core Conjlictual Relationship Theme (CCRT) Components Selected by Two Judges
Component Relationship episode no. 1: Guy Wish Response from other Response from self Judge I Judge 2

To be free of the unwanted visitor He wouldn't understand; he would be insulted 1 feel hassled, resentful, compelled to suffer his presence

To be respected for wanting to do my own thing He interfered; he didn't respect me; he wouldn't understand; he would be insulted I feel hassled, resentful; I submitted to him

Relationship episode no. 2: Therapist Wish Response from self Relationship episode no. 3: Woman Wish Response from other Response from self Relationship episode 4: Some girl (fantasy) Wish Response from other Response from self Relationship episode no. 5: Therapist Wish Response from other Response from self CCRT formulation' Wish Response from other Response from self
1

To be free of having to come to the session I didn't feel like coming but come anyway

To not feel that I have to come to the session I feel compelled to come

To be free of the obligation to have to reach this woman She can't be reached I feel hassled

To get in touch with her yet not to feel I have to She doesn't respond I feel hassled; I feel relieved

To have the girl under my sexual domination She is forced to submit Self-blame

To be sexually dominant She submits I am strong and overpowering

To be free of having to come for so long He gives no response 1 feel hassled and blame myself

To not go through the "hassle" in therapy He does not respond I feel worthless and ashamed

To be free of being obligated and imposed on/4 Does not respond/3 Feel hassled/3; compliance/3

To feel free from obligations and the control of others/4 Does not respond/2 Feel hassled/2; feel compelled to submit/2

Number after slash refers to the number of relationship episodes, within these five relationship episodes, that contain the theme component.

OBJECTIVE MEASURES OF TRANSFERENCE a second set of two judges who compared the CCRT formulations of each of the three CCRT judges and were asked to indicate whether the formulations were basically similar or different. Formulations were judged to be similar if the identical words or words with similar meanings (e.g., anxious and afraid) were used by the different CCRT judges. This task showed good interjudge agreement (96%). We then calculated how often the three CCRT judges came up with similar formulations across the eight cases. The data revealed that on the wish component, the three CCRT judges had similar formulations 75% of the time (6 out of 8); two of the three judges had similar formulations 100% of the time. For the negative RO, 63% of the time the three judges arrived at similar formulations, whereas two out of three judges arrived at similar formulations 88% of the time. For the negative RS, the three judges reached similar formulations 38% of the time, whereas two of the three reached similar formulations 88% of the time. We should emphasize that for good reliability to be achieved, judges should be well-trained in the use of the method (by following the manual and comparing their work with our set of practice cases). Correspondence of the CCRT with Freud's observations of transference. One way to examine the usefulness of the CCRT as a measure of transference is to compare its results with nine observations Freud (1912/1966) made about transference. The basis for much of this novel comparison is a study (Luborsky, Crits-Christoph, et al., 1985; Luborsky, Mellon, et al., 1985) of eight patients' psychotherapy sessions. Each patient's CCRT was scored independently twice: by three judges on a minimum of 10 REs drawn from two sessions early in treatment and by three different judges on 10 REs from two sessions late in treatment (about 1 year later). Each of the following sections takes up one of Freud's nine observations in terms of the relevant CCRT results. 1. Number of transference patterns. Freud (1912/1966) stated that each patient has one transference pattern, "(or several such)" (p. 100). Thus he thought there was one main pattern, but his parenthetical addition shows he was not sure. The CCRT results for the early sessions of the eight patients correspond with his initial impression: There is one main theme, but often a lesser frequency theme is apparent as well. Averaging across the eight patients, the main wish was judged to be present in 80% of each patient's REs, whereas a secondary wish was present, on the average, in only 16% of each patient's REs (Luborsky, CritsChristoph, et al., 1985). 2. Uniqueness of transference patterns. Another observation of Freud (1912/1966, p. 99) was that each patient has a special form of transference pattern. The CCRT results are consistent with this observation. Some of the main wishes from our sample include to be strong, free, and independent; to be close to and open to others; to be responsible and in control; to be seen and treated as a special person; and to be assertive and not acquiesce to authority. Although some degree of similarity was observed among the wish component of different patients, examination of the response components as well reveals a relatively distinct pattern for each patient (Luborsky, Mellon, et al., 1985). 3. Erotic basis of transference pattern. Freud (1912/1966, p. 99) stated that the pattern applies to the "conduct of his [or her] erotic life"the pattern that becomes the transference pattern begins in the child's early years, governs the conduct of the erotic

43

life and then generalizes to every new person. One implication of these views (from Luborsky, Crits-Christoph, et al., 1985) will be examined here as an illustration. Independent judges rated the degree to which each of the REs dealt with explicitly erotic versus explicitly nonerotic interactions. The CCRT was derived for each group of REs. We then had a second set of judges rate paired comparisons of the two categories of the CCRTs to establish their degree of similarity. The results indicated a fair degree of similarity (mean similarity of 4.78 on a 7-point scale where 1 = completely different and 7 = completely identical). This finding implies that the erotic relationships do express a version of the CCRT that is similar to the CCRT from the nonerotic relationships. 4. Awareness of the transference pattern. Freud (1912/1966, p. 100) observed that the transference pattern is composed of a portion that is kept out of awareness and a portion that is in awareness. Our initial attempt at the development of a measure (Crits-Christoph & Luborsky, 1984) relied on guided clinical judgments of the extent to which the patient was aware of the CCRT pattern that was derived from a given session. Results indicated (Crits-Christoph & Luborsky, 1984) that a patient's level of awareness of the CCRT varied considerably from session to session and that there was usually at least one part of the CCRT that was judged to be not in awareness. 5. Consistency of transference pattern aver time. Freud (1912/1966, p. 100) believed that the pattern "is constantly repeatedconstantly reprinted afreshin the course of the person's life." Luborsky, Crits-Christoph, et al. (1985) compared CCRTs scored from sessions early in treatment with the same patient's CCRTs scored from sessions late in treatment, approximately 1 year later. Considerable consistency over time was found: The average similarity on a 1 to 7 scale (1 = not similar, 1 = completely identical) of early and late CCRTs for each patient was 5.7. This is in comparison to a mean similarity of 4.0 for early CCRTs of each patient paired with late CCRTs of different patients. 6. Change in the transference pattern. On this topic Freud (1912/1966, p. 100) stated, "It is certainly not entirely insusceptible to change." The wording of the observation implies that Freud (1912/1966) considered the pattern to have considerable stability as well as some latitude for change. The CCRT results correspond with this observation (Luborsky, Crits-Christoph, et al., 1985). For the eight patients, we examined the change from early sessions to late-in-treatment sessions of the percentage of REs in which the CCRT was present. For example, for early sessions, the average percentage of REs that contained the main positive response from others category was 10.0%. This increased to an average of 17.3% in later sessions. The amount of change in the CCRT, however, varied depending on whether the patient improved in treatment (see the Validity: CCRT improvement versus standard improvement measures section). 7. Transference pattern in relation to the therapist. Freud (1912/1966, p. 100) observed that the therapist becomes "attached to" one of the "stereotype plates." His words mean that the relationship with the therapist in the course of the treatment becomes like the general transference pattern, that is, the stereotype plate. This hypothesis was tested (Luborsky, Crits-Christoph, et al., 1985) by deriving CCRT formulations separately for REs involving the therapist and for those involving other people (i.e.,

44

L. LUBORSKY, P. CRITS-CHRISTOPH, AND J. MELLON

not the therapist). Judges blindly rated the similarity of the therapist CCRT to the other people CCRT for each patient and, for comparison purposes, also rated the similarity of the therapist CCRT of each patient to the other people CCRTs of the seven other patients in the study. The correct pairings of therapist CCRTs with other people CCRTs were given an average similarity rating of 6.5 on a 7-point scale (7 = completely identical) whereas the incorrect pairs (i.e., therapist CCRT for one patient matched with an other people CCRT of another patient) were given a lower average similarity rating (4.6). These results constitute the first quantitative confirmation of the existence of a parallel, based on psychotherapy sessions, of the patient's experience of the relationship with the therapist and the patient's experience of relationships with other people. S. Early origins of the transference pattern. In Freud's (1912/ 1966, p. 99) view, the transference pattern derives from the "combined operation of his [or her] innate disposition with the influences brought to bear on him [or her] during his [or her] early years." This observation implies that there should be a parallel between the relationship pattern with the therapist and the one with the early parental figures. The term transference seemed fitting to Freud (1912/1966) because it was a central part of his concept that attitudes and behavior developed in the relationships to the early parental figures were transferred to other relationships. Luborsky, Crits-Christoph, et al. (1985) compared CCRTs scored from REs involving a memory of an interaction with early parental figures versus the overall CCRT scored from all other REs. A high degree of similarity was evident (mean rating on a 7-point scale = 6.4) for early memory of parent CCRTs paired with the same patient's overall CCRT. When the early memory of parent CCRT for each patient was matched with overall CCRTs from other patients, less similarity was evident (mean similarity = 3.6). The accuracy of recall by the patient of the early parental relationships, however, presents potential problems for data derived from psychotherapy alone. One way to deal with this limitation is to examine those rare patients in psychotherapy for whom early observational data are also available. 9. Transference inside versus outside of the treatment setting. Freud (1912/1966, p. 101) stated, "It is not a fact that transference emerges with greater intensity and lack of restraint during psychoanalysis than outside it" Through this statement Freud affirmed that transference is evident both inside and outside of the analysis. Using the CCRT method, the first study of this observation is under way (van Ravenswaay, Luborsky, & Childress, 1983). To evaluate the transference pattern from data obtained outside of the psychotherapy, a method was constructed called the Relationship Anecdotes Paradigm (RAP) Test (Luborsky, 1978). It is a specially designed interview in which the person is asked to tell relationship anecdotes about specific other people. We have found so far on six patients that there is considerable consistency in the CCRT derived from narratives within treatment sessions with the CCRT derived from outside-of-treatment narratives, a finding consistent with Freud's observation about the generality of transference. The inside and the outside sources both reveal virtually the same wish as well as many of the same responses from the other person and responses from the self. The parallel implies that the transference pattern is expressed in treatment but exists outside of therapy as well.

In summary, there is much concordance between Freud's nine observations about transference and data from the CCRT measure. For each of Freud's observations for which an operational proposition was readily derivablesix out of nine by conservative countthe CCRT data provided results consistent with the observations. This implies some usefulness for the CCRT measure as a measure of the transference concept. Validity: CCRT improvement versus standard improvement measures. One way to assess the validity of the CCRT is to relate it to other measures. Using data from the study of eight patients (Luborsky, Mellon, et al., 1985), we hypothesized that change in the CCRT from early to late in treatment should be related to independent measures of the outcome of treatment. The measure of change in CCRT was the difference score between the early treatment pervasiveness of each CCRT component (i.e., the percentages of relationship episodes that contained each main wish, negative RS, negative RO, positive RS, and positive RO) and the late-in-treatment pervasiveness of the same CCRT components. Two independent outcome measures were selected as criteria, one from the perspective of the patient, the Hopkins Symptom Checklist total score, and one from the perspective of an external clinical judge, the Health-Sickness Rating Scale (Luborsky, 1975). Both measures were obtained at the beginning of treatment and at termination in the Penn Psychotherapy Project (Luborsky et al., 1980). Change in the pervasiveness of the main negative RS was significantly correlated with change in HealthSickness Rating Scale, r(6) = -.81, p < .05, as was change on the main wish, r(6) = -.73, p < .05. Change in the main positive RO was significantly correlated with change on the Hopkins Symptom Checklist, r(6) = -.79, p < .05. The direction of all of these correlations was as expected, that is, increase in the frequency of positive components or decrease in negative components of the CCRT was associated with more favorable outcomes. A study by Baguet, Gerin, Sali, and Marie-Cardine (1984) has also shown a relation of change on the CCRT to change on the Health-Sickness Rating Scale. Another kind of validity also has been explored: validity defined as meaningful relation of the CCRT to other phenomena. The core content of what is forgotten during momentary forgetting in psychotherapy is similar to the CCRT content (Luborsky, in press). Comparison of the CCRT method with other methods. The PERT measure (Gill & Hoffman, 1982b) is the best known similar measure; the actual consistency of the PERT and CCRT methods is under investigation (Kaechele, Thomae, & Luborsky, 1983). The Plan Diagnosis Method (Weiss & Sampson, in press) may also be a similar measure, although its authors do not explicitly describe it as a measure of transference. It involves a clinical description of four components: the patient's goals for therapy, the inner obstacles (pathogenic beliefs) preventing the patient from reaching the goals, the tests in the treatment through which the patient confirms or disconfirms the obstacles, and the insights that will be helpful. The clinical description of each component is reduced to a list of brief statements. In a reliability study (Rosenberg, Silberschatz, Curtis, Sampson, & Weiss, 1983), the list was expanded to include statements that were not relevant to the particular patient, just as was done in the method of mismatched cases (Levine & Luborsky, 1981). An independent team was able to reliably differentiate case-relevant versus case-irrelevant statements.

OBJECTIVE MEASURES OF TRANSFERENCE

45

The Dynamic Focus Method (Schacht & Binder, 1982) provides guidelines for formulating the treatment focus of the therapist's interventions. This system has four components called (a) acts of self, (b) expectations of others, (c) consequent acts of others toward self, and (d) consequent acts of self toward self. These four components appear to be similar to the three components in the CCRT. Acts of self, for example, include the wishes. The expectations of others and consequent acts of others toward self are both included in the RO in the CCRT system. Consequent acts of self toward self are essentially the same as the RS in the CCRT system. Teller and Dan! (1981) are developing a language-based (and eventually computer-assisted) data analysis system for psychotherapy sessions. They demonstrated the principle that even within a single session with only a few frames, that is, recurrent structures within relationship episodes, one can tap into basic underlying relationship schemata. Another promising method that may capture transference phenomena is configurational analysis (M. Horowitz, 1979). In this method, the data from process notes and transcripts of sessions are examined from three points of view: state, relationship, and information. Of the three, the system for the analysis of the relationship is most like that of the CCRT. Evidence for reliability for the state analysis is provided (M. Horowitz, 1979).

Individual differences among patients in response to transference interpretations are impressive. In a study (Luborsky, Bachrach, Graff, Pulver, & Christoph, 1979) of three patients selected to represent a range of outcomes, 16 interpretations were chosen for each patient. These were reliably rated by independent judges for 250 word units before and after each interpretation. For each patient there was a clear parallel among positivity of the immediate response to interpretations, the helping alliance, and the outcome of treatment. Each of the new measures of transference has been presented as a guide to the therapist in formulating and responding interpretively. The categories of the PERT measure are set up in terms of Gill's (1982) and Gill and Hoflman's (1982b) belief in the importance of therapist responding interpretively to indications of PERT. In the CCRT measure, the principles for formulating the CCRT and deciding on the best response based on this formulation are explicated in Luborsky (1984) for both the CCRT method and a simplified clinical version that can be used in everyday practice. Similarly, Silberschatz (1984) reported using the Plan Diagnosis Method as a guidance system for judges. A significant correlation was obtained between the judges' rating of consistency of the therapist's response to the Plan Diagnosis Method with the immediate and long-term benefits to the patient.

Benefits of Transference Interpretations


Much of the research interest in fashioning measures of the transference arose because the theory of the curative action of psychoanalytic psychotherapy stresses that the therapist's interventions should deal with the transference. Nevertheless, only a few studies, noted next, have tried to measure the beneficial impact of therapist's interpretations. The Menninger Foundation Psychotherapy Project (Kernberg et al., 1972) suggested that when the focus on transference is high, patients diagnosed as borderline psychotic and treated by skillful therapists improve significantly more. Malan's study (1976) found a trend for the more successfully treated patients to be those for whom an emphasis was placed on interpretations dealing with transference in relation to parents and therapists. Because Malan's data consisted of only the therapists' process notes, Marziali (1984) replicated his work by using recordings of 26 patients. The frequency of interpretations with therapistparent links and therapist-parent-other links correlated significantly with several of the outcome scales as well as with the sum of five psychodynamic scales (e.g., the therapist-parent and therapist-parent-other measures both correlated .57, p < .01, with the sum score). In Crisp's work (1964a, 1964b, 1966) it was found that attitudes toward the therapist often change along with or before changes in the patient's symptoms. The implication is that the understanding (probably aided by the therapist's interventions) precedes symptom change. The earliest precise analysis of the immediate effects of transference interpretations within psychodynamic psychotherapy sessions was by Garduk and Haggard (1972). They compared the immediate effects during the 5 min after transference interpretations versus after noninterpretations for four cases. Interpretations were shown to elicit more defensive associations, more transference-related material, more understanding, and more affect than were noninterpretations.

Methodological Issues in the Measurement of Transference


Research on such a complex concept as transference presents challenging methodological problems for investigators attempting to develop a reliable and valid measure. One research issue concerns the choice of the data base on which to score a transference measure. The use of psychotherapy transcripts and recordings is warranted not only because these are the data from which the concept was originally generated but also because psychotherapy provides a real-life enactment of the transference with the therapist that others can observe. Disadvantages of using material from psychotherapy sessions are the large amount of time needed for transcription and the intricate process of reviewing the transcripts to arrive at formulations or to make ratings. Although questionnaire methods save research time, the validity of such measures needs to be demonstrated. Decisions about the data base do not stop with the use of psychotherapy sessions. It is necessary to identify a scorable unit so that judges can concentrate their efforts on relevant material. The Schacht and Binder (1982) Dynamic Focus Method uses patient utterances (a single, uninterrupted turn at talk) as the unit of scoring. For the CCRT method, a procedure of delineating relationship episodes has been adopted. The assumption is that transference patterns will be evident in many of them. However, it will be important to determine the extent to which a set of such units is representative for a patient and whether there are differences in adequacy of the units that affect the scoring. In order to mitigate some of these problems, Schacht and Binder (1982) developed a special interview to elicit more relevant utterences to score. Measures that involve determining a clinical formulation (e.g., CCRT, Plan Diagnosis, and Dynamic Focus) require a methodological decision on guided versus unguided clinical judgment In the unguided approach, the clinician is free to choose any principle and any level of abstraction for his or her formulations. The guided approach, in contrast, specifies some principles and

46

L. LUBORSKY, P. CRITS-CHRISTOPH, AND J. MELLON vidual transference themes in a group psychotherapy: Application of the core relationship method]. Psychotherapies, 1-2, 43-49. Benjamin, L. S. (1974). Structural analysis of social behavior. Psychological Review, 81, 392-425. Breuer, J., & Freud, S. (1966). Studies on hysteria. In J. Strachey (Ed. and Trans.), The standard edition of the complete psychological works qfSigmund Freud (Vol. 2). London: Hogarth Press. (Original work published 1893-1895.) Carlson, R. (1981). Studies of script theory: I. Adult analogs of a childhood nuclear scene. Journal of Personality and Social Psychology, 40, 501 -

some of the levels of abstraction in advance. Holt (1978) reviewed guided versus unguided clinical judgments and concluded that guided approaches yield greater reliability and validity. For transference, attempts to use unguided judgments have given ambiguous results (e.g., Seitz. 1966). The more recent measures all provide some degree of guidance to the judge. Even with a guided clinical method the problem of variability in wording and amount of detail in each judge's formulations still exists. This variability is especially true of ideographic methods, such as the CCRT, because these allow for unique descriptions of each case. The CCRT method minimizes variability in wording across judges by encouraging the judges to stay close to the wording used in the transcripts, by keeping the level of inferences to moderately clearly inferable ones, and by extracting only the redundant themes across many REs. The nomothetic approach escapes the problem of variability in wording by asking judges to fit formulations into preset categories. For example, the Schacht and Binder (1982) Dynamic Focus Method requires judges to code formulations into the categories provided by the "Structual Analysis of Social Behavior" (Benjamin, 1974). Although the idiographic approach is closer to the process of making transference formulations in clinical practice, the use of preset categories provides many clear benefits to the researcher. These include the likely higher reliability and ease of application and quantification compared with free-form patient-specific categories. The noncomparable data from freeform methods make statistical evaluation difficult. Whether the gains of preset categories offset the loss of uniqueness in description of each case remains a question for future research. Further assessment of the validity of the new measures of transference is needed. In terms of discriminant validity, it must be demonstrated that these measures provide information beyond that given by general personality measures (e.g., the Inventory of Interpersonal Problems; L. Horowitz, Weckler, & Doren, 1983) and diagnostic assessment systems (e.g., the Schedule for Affective Disorders and Schizophrenia; Endicott & Spitzer, 1978). Other kinds of validity questions must also be raised: What types of criteria should transference measures be expected to predict? How should such relationships be assessed? How do measures derived from narratives told in psychotherapy relate to observations of interpersonal interactions? These questions are now being addressed, but further work is needed. In conclusion, the oldest line of quantitative research based on questionnaire methods did not gain much acceptance. Only within the last 10 years have measures been based on psychotherapy sessions. It is time now to move ahead with further development of these methods, as well as to examine how the earlier questionnaire and the newer session methods are interrelated. In the event that the two approaches agree, we will have to acknowledge that the advent of objective transference measures was much earlier than was recognized in the last review of the field (Luborsky & Spence, 1978).

510.
Chance, E. (1952). The study of transference in group therapy. International Journal of Group Therapy, 2, 40-53. Crisp, A. (1964a). An attempt to measure an aspect of transference. British Journal of Medical Psychology, 37, 17-30. Crisp, A. (1964b). Development and application of a measure of transference. Journal of Psychosomatic Research, S, 327-335. Crisp, A. (1966). Transference, symptom emergence and social repercussion in behavior therapy: A study of 54 treated patients. British Journal of Medical Psychology, 39, 179-196. Crits-Christoph, P., & Luboisky, L. (1984, September). Development of a measure of self-understanding of core relationship themes. Paper presented at the National Institute of Mental Health conference on Methodologic Challenges in Psychodynarnic Research, Washington, DC. Endicott, J., & Spitzer, R. (1978). A diagnostic interview: The schedule for affective disorders and schizophrenia. Archives of General Psychiatry, 37, 837-844. Fiedler, F., & Senior, K. (1952). An exploratory study of unconscious feeling reactions in fifteen patient-therapist pairs. Journal of Abnormal and Social Psychology, 47, 446-453. Fisher, S., & Greenberg, R. (1977). The scientific credibility of Freud's theories and therapy. New \brk: Basic Books. Freud, S. (1966). The dynamics of the transference. In J. Strachey (Ed. and Trans.), The standard edition of the complete psychological works ofSigmwd Freud (Vol. 12, pp. 99-108). London: Hogarth Press. (Original work published 1921) Garduk, E., & Haggard, E. (1972). Immediate effects on patients of psychoanalytic interpretations. Psychological Issues, Monograph 28, 185. Gill, M. (1982). Analysis of transference: Theory and technique. Psy-

chological Issues, Monograph 53, 1-193. Gill, M., & Hoffman, I. (1982a). Analysis of transference: Studies of nine audio-recorded psychoanalytic sessions. Psychological Issues, Monograph 54, 1-229. Gill, M., & Hoffman, I. (1982b). A method for studying the analysis of aspects of the patient's experience of the relationship in psychoanalysis and psychotherapy. Journal of 'the American Psychoanalytic Association, 30, 137-167. Graff, H., & Luborsky, L. (1977). Long-term trends in transference and resistance: A quantitative analytic method applied to four psychoanalyses. Journal of the American Psychoanalytic Association, 25, 471-

490.
Holt, R. R. (1978). Methods in clinical psychology: Prediction and research (Vol. 2). New York: Plenum Press. Horowitz, L., Weckler, D., & Doren, R. (1983). Interpersonal problems and symptoms: A cognitive approach. In P. C. Kendall (Ed.), Advances in cognitive-behavioral research and therapy (Vol. 2, pp. 82-127). New \brk: Academic Press. Horowitz, M. (1979). States of mind: Analysis of change in psychotherapy.

References
Apfelbaum, B. (1958). Dimensions of transference in psychotherapy. Berkeley: University of California Press. Baguet, J., Gerin, P., Sali, M, & Marie-Cardine (1984). Evolution des themes transferentiels individuels dans une psychotherapie de groupe (Application de la methode du relationnel central) [Evolution of indi-

New York: Plenum Press. Kaechele, H., Thomae, H., & Luborsky, L. (1983, July). A comparison of two transference-related measures. Paper presented at the meeting of the Society for Psychotherapy Research, Sheffield, England. Kernberg, Q, Burstein, E., Coyne, L., Appelbaum, A., Horwitz, L., & Voth, H. (1972). Psychotherapy and psychoanalysis. Bulletin of the Menninger Clinic, 36, 1-178.

OBJECTIVE MEASURES OF TRANSFERENCE Levine, F. J., & Luborsky, L. (1981). The core conflictual relationship theme method: A demonstration of reliable clinical inferences by the method of mismatched cases. In S. Tuttman, C. Kaye, & M, Zimmerman (Eds.), Object and self A developmental approach (pp. 501526). New York: International Universities Press. Lower, R., Escoll, P., Little, R., & Ottenberg, P. (1973). An experimental examination of transference. Archives of General Psychiatry, 29, 738-

47

lationship paradigms. American Journal of Orthopsychiatry, 30, 507-

520. Meichenbaum, D., & Gilmore, J. B. (1984). The nature of unconscious


processes: A cognitive-behavioral perspective. In K. Bowers & D. Meichenbaum (Eds.), The unconscious reconsidered (pp. 273-298). New York: Wiley. Murray, H. (1938). Exploration in personality. New York: Oxford University Press. Rawn, M. (1958). An experimental study of transference and resistance phenomena in psychoanalytically-oriented psychotherapy. Journal of Clinical Psychology, 14, 418. Rawn, M. (1981). A note on unwitting replication: Quantitative studies of transference and resistance twenty years apart. Journal of Clinical Psychology, 37, 782. Rosenberg, S., Silberschatz, G., Curtis, J., Sampson, H., & Weiss, J. (1983, July). The plan diagnosis method: A new approach to establishing reliability for psychodynamic formulations. Paper presented at the meeting of the Society for Psychotherapy Research, Sheffield, England. Schacht, T, & Binder, J. (1982). Focusing: A manual for identifying a circumscribed area of work for time-limited dynamic psychotherapy (TLDP). Unpublished manuscript, Vanderbilt University. Seitz, P. (1966). The consensus problem in psychoanalytic research. In L. Gottschalk & A. Auerbach (Eds.), Methods of research in psychotherapy (pp. 209-225). New York: Appleton-Century-Crofts. Silberschatz, G. (1984, September). Effect size and the unit of measurement in psychodynamic psychotherapy research. Paper presented at the National Institute of Mental Health conference on Methodologic Challenges in Psychodynamic Research, Washington, DC. Singer, J. L. (1984, August). Transference and the human condition: A cognitive-affective perspective. Paper presented at the meeting of the American Psychological Association, Toronto, Ontario, Canada. Strupp, H., Chassan, J., & Ewing, J. (1966). Toward the longitudinal study of the psychoanalytic process. In L. Gottschalk & A. Auerbach (Eds.), Methods of research in psychotherapy (pp. 361-400). New\brk: Appleton-Centry-Crofts. Subotnick, L. (1966a). Transference in child therapy: A third replication. Psychology Record, 16, 265-277. Subotnick, L. (1966b). Transference in client-centered play therapy. Psychology, 3, 2-17. Teller, V., & Dahl, H. (1981). The framework for a model of psychoanalytic inference. Proceedings of the Seventh International Joint Conference on Artificial Intelligence, 1, 394-400. Tomkins, S. (1979). Script theory: Differential magnification of affects. In H. E. Howe, Jr., & R. A. Diensbier (Eds.), Nebraska symposium on motivation (Vol. 26, pp. 201-236). Lincoln: University of Nebraska Press. van Ravenswaay, P., Luborsky, L., & Childress, A. (1983, July). Consistency of the transference in versus out of psychotherapy. Paper presented at the meeting of the Society for Psychotherapy Research, Sheflield, England. Wallerstein, R. (1985). The Menninger Foundation psychotherapy research project. New \brk: Guilford Press. Wallerstein, R., & Robbins, L. (1956). The psychotherapy research project of the Menninger Foundation: IV. Concepts. Bulletin of the Menninger Clinic, 20, 239-262. Weiss, J., & Sampson, H. (in press). Thepsychotherapeuticprocess. New York: Guilford Press. Received December 3, 1984 Revision received May 13, 1985

741.
Luborsky, L. (1975). Clinicians'judgments of mental health: Specimen case descriptions and forms for the Health-Sickness Rating Scale. Bulletin of the Menninger Clinic, 39, 448-480. Luborsky, L. (1976). Helping alliance in psychotherapy: The groundwork for a study of their relationship to its outcome. In J. L. Claghom (Ed.), Successful psychotherapy (pp. 92-116). New York: Brunner/Mazel. Luborsky, L. (1977). Measuring a pervasive psychic structure in psychotherapy: The core conflicutal relationship theme. In N. Freedman & S. Grand (Eds.), Communicative structures and psychic structures (pp. 367-395). New York: Plenum Press. Luborsky, L. (1978). The Relationship Anecdotes Paradigm Test (RAP Test): A TAT-like method using actual narratives. Unpublished manuscript. Luborsky, L. (1983). The core conflictual relationship theme method: Guide to scoring and rationale. Unpublished manuscript. Luborsky, L. (1984). Principles of psychoanalytic psychotherapy: A manual for supportive-expressive (SE) treatment. New York: Basic Books. Luborsky, L. (in press). Recurrent momentary forgetting: Its content and context. In M. Horowitz (Ed.), Emotional and cognitive factors in unconscious processes. Chicago: University of Chicago Press. Luborsky, L., Bachrach, H., Graff, H., Pulver, S., & Christoph, P. (1979). Preconditions and consequences of transference interpretations: A clinical-quantitative investigation. Journal of Nervous and Mental Disease, 167, 391-401. Luborsky, L., Crabtree, L., Curtis, H., Ruff, G., & Mintz, J. (1975). The concept "space" of transference for eight psychoanalysts. British Journal of Medical Psychology, 48, 1-6. Luborsky, L., Crits-Christoph, P., Mellon, J., Alexander, K., Cohen, K., Childress, A., Levine, F., & Hole, A. V. (1985). Freud's grandest clinical concept of transference: Further confirmation by the CCKT method. Unpublished manuscript. Luborsky, L., Graff, H., Pulver, S., & Curtis, H. (1973). A clinical-quantitative examination of consensus on the concept of transference. Archives of General Psychiatry, 29, 69-75. Luborsky, L., Mellon, J., Alexander, K., van Ravenswaay, P., Childress, A., Levine, F, Cohen, K. D., Hole, A. V., & Ming, S. (1985). A verification of Freud's grandest clinical hypothesis: The transference. Clinical Psychology Review, 5, 231-246. Luborsky, L., Mintz, J., Auerbach, A., Christoph, P., Bachrach, H., Todd, T, Johnson, M., Cohen, M., & O'Brien, C. P. (1980). Predicting the outcomes of psychotherapy: Findings of the Penn Psychotherapy Project. Archives of General Psychiatry, 37, 471-481. Luborsky, L., & Spence, D. (1978). Quantitative research on psychoanalytic therapy. In S. L. Garfleld & A. E. Bergin (Eds.), Handbook of psychotherapy and behavior change: An empirical analysis (pp. 331368). New York: Wiley. Malan, D. (1976). Toward the validation of dynamic psychotherapy. New York: Plenum Press. Marziali, E. (1984). Prediction of outcome of brief psychotherapy from therapist interpretive interventions. Archives of General Psychiatry, 41, 301-304. Mayman, M., & Paris, N. (1960). Early memories as expressions of re-

Anda mungkin juga menyukai