Ends E.J., Page, C.W.
Quarterly Journal of Studies on Alcohol: 1957, 18, p. 263–277.
Early US study uncovered for the Alcohol Treatment Matrix found a Rogerian client-centred therapy characterised by non-directive, empathic listening beneficially changed self-perceptions of alcohol-dependent patients and reduced relapse compared to approaches based on learning theory or psychoanalysis.
Summary This seminal study uncovered for the Alcohol Treatment Matrix found a Rogerian, empathic client-centred approach (for which see matrix cell B4) reduced relapse compared to approaches based on learning theory or psychoanalysis, seemingly because it led to the most balanced changes in perceptions of self and of ideal self.
The participants were 96 “inebriate” patients aged 25 to 45 admitted as inpatients to the Willmar (Minnesota) State Hospital in the USA, selected from all such patients on the basis that their intellectual level enabled them to actively participate in the study. They were formed into 16 groups of six patients each for group therapy. Then sets of four groups were allocated to four types of therapeutic approaches, delivered over five weeks in 15 sessions by four therapists, each trained in all the approaches. Pre-study training included expert feedback on session recordings. The four therapists each took one of the groups allocated to each approach, an attempt to eliminate differences between therapists as a cause of the outcomes, leaving the therapeutic approach as the main influence.
One of the approaches was a therapeutically neutral social discussion format deliberately avoiding the specific elements which characterised the other approaches. The four groups allocated to this formed a control set.
Elaborated by the researchers themselves, the approach they expected to turn out best was based on learning theory, which also underpins cognitive-behavioural therapy. Another approach was the non-directive or client-centred therapy developed by Carl Rogers, a foundation for modern-day motivational interviewing. How this and the other approaches were implemented in the study is not detailed, but as formulated by its originator, the Rogerian method entails the communication of genuineness, unconditional positive regard (no ‘ifs’ qualifying the therapist’s acceptance of the patient) and empathic understanding. Explicit advice or direction is not considered to be among these “necessary and sufficient conditions” for therapeutic progress, though the therapist might seek to provoke change by communicating new ways for the patient to see and understand their problems. The final approach was a variant of psychoanalysis, based on investigating the interaction of conscious and unconscious elements in the mind and bringing repressed fears and conflicts into the conscious mind by techniques such as dream interpretation and free association.
The authors were most interested in how these approaches affected the psychological health of the patients, something they were able to assess in the 63 who completed the therapies. Their method was to assess as an average for each group the degree of movement from starting treatment to its end in scores patients recorded on a self-administered psychological test. It involved sorting 100 cards displaying statements like, “I understand myself,” and “I just don’t respect myself,” in the order in which they were seen by the patient as best characterising themself and the person they would like to be. Clinical psychologists had previously completed this exercise with a view to sorting the cards in a way most indicative of psychological health. At issue was the degree to which patients’ ‘self’ and ‘ideal self’ card-sorts moved towards or way from this healthy pattern over the course of therapy. Also assessed was the patients’ progress in controlling their drinking and their behaviour over the 18 months after treatment ended. In both cases only the 63 patients who completed their therapy programme were included in the analyses. Drop-out was fairly even across the therapies, the final numbers of patients ranging from 15 to 17 in each group.
Unexpectedly, the researchers’ favoured learning-theory approach seeming to produce the least healthy changes, and the Rogerian approach the most healthy:
Learning-theory therapy ... does not demonstrate any therapeutic usefulness by reversing any of the deleterious processes which occurred spontaneously in the control group (with the single exception of index 6), but rather tends to accelerate such processes. A second point is that group meetings in which the therapist is not actively ‘therapeutic’ have no therapeutic value for most of the participants, and result in minimal changes, but changes that are not constructive and generally harmful ... [W]hile superficially equivalent in terms of gain in healthiness of the self-concept and gain in self-acceptance, a significant and perhaps crucial difference exists between the effects of analytic therapy and client-centered therapy. That difference lies in the means by which the increased agreement between self and ideal were brought about. In the case of the analytic group, change was achieved by a one-sided modification of the self; in the case of the client-centered group it resulted from a coordinated reworking of both self and ideal. If one believes that misbehavior, mental illness or maladjustment are manifestations of the whole personality, rather than of one aspect of it, then the process of coordinated restructuring seems to represent sounder therapeutic movement, or personality growth.
To test whether these in-treatment changes were reflected in later behaviour change among the patients, nearly all who completed treatment were followed up by assessors did not know to which approach they had been allocated. Interviews six months after and hospital records one and one-and-a-half years after treatment ended were used to assess whether there had been “no evidence of further alcoholic episodes or misbehavior”, evidence interpreted as the patient having greatly improved. Lesser degrees of improvement were classified as “possibly improved” and “apparently no change”. Also recorded was whether the patient had been readmitted to the hospital during the 18-month follow-up. At 7 of 15 greatly improved and 3 readmitted, best results occurred after Rogerian therapy chart. Next at 6 and 5 respectively was the psychoanalytic approach, and way behind were the learning-theory approach and the control group, few of whose patients had no evidence of relapse and around half of whom had been re-admitted. In other words, the pattern of relapse-related data gathered up to one-and-a-half years after treatment matched the pattern of during-treatment changes in psychological health. Both indicated that though it consisted mainly of empathic listening with no specific techniques to promote behaviour change, the Rogerian approach was the most effective.
commentary Blinding of the follow-up assessors, the training and supervision of the therapists, and their rotation across the therapies, make this for its time a methodologically advanced study. The poor performance of the researchers’ favoured learning theory programme seems to rule out ‘researcher allegiance’ as an influence on the findings. However, these can only be considered to apply to the two-thirds of patients who stayed long enough in the hospital to complete the therapies. Though the researchers performed statistical tests on the follow-up results, these took no account of the clustering of the patients in groups, undermining findings of statistical significance. The main question mark over the trial is whether the 16 groups were allocated at random to the four different approaches or in some other way which helped reduce the chances that the outcomes were due to the types of patients allocated to each therapy. The report on the study does not specify the allocation method, though it is clear that allocation of patients to the 16 groups was not random, making the random allocation of the groups themselves even more important. The study is, however, among those in a bibliography which required that studies “used a procedure (typically randomization) designed to yield equivalent groups before treatment”.
Last revised 23 February 2017. First uploaded 19 February 2017
REVIEW 2018 Cohesion in group therapy: a meta-analysis