Research Snippet – March 2009


Donald Robertson
Donald Robertson

Apologies for the delay while I settled into the research director role. We’re up and running now and I thought I’d begin with a different type of research snippet. We’ll look at some recent studies shortly – and NCH members can also read my article in this month’s Hypnotherapy Journal. However, first of all, let’s go back to the beginning. The beginning of modern psychotherapy outcome research can arguably be dated to the following widely-referenced study, whose authors introduced the statistical technique of meta-analysis.


In their seminal The Benefits of Psychotherapy (1980), Mary Lee Smith and her team of colleagues used the new technique of “meta-analysis” to process data from large numbers of controlled research studies on different modes of psychotherapy. Advanced statistical methods were used to derive an overall “effect size” for different forms of psychotherapy based on a review of all available research on psychotherapy prior to 1980. Their review identified 475 waiting list or placebo group controlled studies of adequate quality for inclusion in the meta-analysis.

Smith et al. found that the raw data from these studies showed most modalities of psychotherapy were effective, and that there was little difference between them in terms of their effect size. However, a small handful of treatments were found to have significantly above-average effect size,

Cognitive, cognitive-behavioural therapies, hypnotherapy, and systematic desensitisation appeared most effective. (Smith et al., 1980: 124).

These above-average approaches were all cognitive, behavioural, hypnotherapy, or some hybrid of those modalities. However, Smith et al. immediately proceeded to recalculate the figures by merging the different modalities of therapy into a small handful of broad categories and weighting the outcomes based on how “reactive” the types of measurement used were. It was argued that the measurements used in cognitive and behavioural therapies may have inflated the outcomes. When this was done, the superiority of cognitive and behavioural therapies was reduced, and specific measures relating to hypnotherapy dropped out of the process. Subsequent researchers, however, criticised this verdict and re-analsyed the figures using more stringent criteria which restored the intial superiority attributed to cognitive and behavioural therapies. Moreover, the way Smith et al. merged the different modalities into super-categories meant that different types of treatment became lumped together in their revised figures.

This study, therefore, with its internally-conflicted findings, can be seen to have prefigured the nature of debate over comparative psychotherapy outcome research over subseqeunt decades. Since the 1980s researchers have been divided between two broad camps. The first argues that specific techniques, mainly CBT but possibly also techniques such as hypnosis, are shown to be more effective than others. The other camp, which tends to be favoured by humanistic and psychodynamic therapists, interprets the outcome research as showing that all forms of psychotherapy are more or less equally effective, and that outcomes are mainly determined by the quality of the therapeutic alliance, i.e., the human relationship between therapist and client.

In any case, a number of other notable results can be derived from the huge meta-analysis conducted by Smith et al.

  • Measures of fear and anxiety were generally most responsive to psychotherapy.
  • Measures of personality traits and work or school performance were least responsive.
  • Depressed clients and clients with simple phobias were most responsive to psychotherapy.
  • Psychotics, complex neurotics and handicapped clients experienced least improvement.
  • The average effect size of 0.9 measured immediately post-treatment diminished to 0.5 on average at two year follow-up, perhaps an indication of relapse rates.
  • The years of experience of the therapist were unrelated to therapeutic outcomes.
  • When therapist and client were closely matched in terms of education and socio-economic class the benefits were greater.
  • Clients with higher IQ scores benefited more from therapy.
  • Female clients seemed to benefit more on average from therapy than men.
  • Individual and group therapy were equally effective.
  • Duration of therapy was unrelated to effectiveness, i.e., brief therapies of less than 10 sessions were just as effective as therapy conducted over many months or even years.
  • The more rigorous the research design, the higher the effect size measured.
  • Clients responding to advertisements seeking participants benefited more than those who sought treatment independently because of a problem, both benefited more than prison inmates or hospital in-patients. (1980: 124-12).

    Hence, intelligent female clients with depression or phobia, seeking to reduce anxiety, are perhaps the group most likely to benefit from psychological therapy, and brief group therapy with novice therapists might be just as effective for them as long-term individual therapy with an experienced practitioner, but the former is probably much cheaper.
    Smith et al. concluded that overall clients’ personal qualities probably contribute more to outcomes than the therapist’s theories or actions. Logically, this should suggest that the most important aspect of therapy is client selection for treatment rather than the choice of interventions used once treatment has started. It is surprising that so little attention has been paid to this aspect of psychotherapy.