The Efficacy of Personal Construct Therapy as a Function of the Type and Severity of the Presenting Problem

Abstract

A recent quantitative review of the personal construct therapy (PCT) outcome literature revealed reliable but somewhat modest effects for this type of therapy. Examination of moderator variables has shown that the efficacy of PCT might vary as a function of other factors, such as whether or not the treatment was tested with a clinical population. In the present study, these findings were expanded on by exploring the relation between the type and severity of presenting problems and treatment outcome as reported in the controlled PCT outcome literature. Overall, results revealed that the efficacy of PCT did not differ substantially across different types of problems, but effect sizes tended to be significantly smaller when more severe problems were being treated.

Notes

1. Standard care refers to a form of treatment that did not include an active psychological component and was customarily administered for the population of interest or the particular setting in which the study took place. For example, in one study of a PCT treatment for patients with musculoskeletal pain (Haugli et al., 2003), participants in the control group met with a physician and adhered to whatever treatment suggestions he or she made, which was standard protocol for patients with this problem.

2. The influence of publication bias in the PCT literature has been explored in previous publications (Holland et al., 2007; Metcalfe et al., 2007; Viney et al., 2005). In general, it appears that unpublished studies have yielded somewhat smaller effect sizes.

3. Waitlist, no treatment, standard care, and inactive treatment controls were all grouped as “no active treatment” because of the similarity of the effect sizes calculated across these different comparisons. In fact, effect sizes based on comparisons with placebo treatments and standard care on average tended to be somewhat larger than those based on waitlist or no treatment comparisons, which is the opposite of what might be expected.

4. This overall effect size differs somewhat from that reported in a previous publication (Holland et al., 2007). This difference is due to the updating of this data set with the results of a recent study by Winter and colleagues (2007).

5. Watson & Winter's (2005) study was not coded for the type of presenting problem, because the sample included participants presenting with a multitude of diverse problems. Hence, this study was not included in any of the analyses involving this variable.

6. It should be noted that this effect size was based on one study (Bannister et al., 1975) and was calculated entirely with estimation procedures, which in this case mostly involved in-putting an effect size of zero for measures that were reported as not yielding significant differences between the treatment and control group at posttreatment. Thus, this effect size should be interpreted cautiously.

7. In References, ∗ indicates reference included information about a PCT outcome study that was used in the review. Some references represent the same study. In addition, Viney, Metcalf, and Winter's (2005) review provided extra information that had previously been unpublished for multiple PCT outcome studies. Superscripts indicate the problem type assigned to each study, whereby 1 = trauma and stress, 2 = physical problems and aging, 3 = anxiety and fear, 4 = problematic behavior, and 5 = psychosis and delusions.