The dissertation focuses on two areas of research, methodological questions regarding observer based measurements and how factors related to the therapist and to the patient in psychotherapy together affect the outcome of therapy.
Specifically the dissertation contains one article investigating the effect of having observers rate multiple constructs relevant to psychotherapy in order to assess the psychotherapy treatment process accurately. This first article forms the methodological foundation of the dissertation as the two other articles used the assessment method investigated in this first study to obtain data for further analysis. The article used generalizability (G) theory to investigate sources of variance, particularly if assessing two versus assessing one instrument simultaneously introduced variability in the scores that was not due to differences across the object of measurement. The study concluded that only minimal amounts of variance were introduced by assessing two instruments simultaneously, and that the assessment procedure was sound.
The dissertation includes two articles investigating psychotherapy process. The data is observer based assessments from a trial comparing short-term dynamic psychotherapy (STDP) and cognitive therapy (CT) for cluster C personality disorders. In the original publication (Svartberg, Stiles & seltzer, 2004) the researchers found no differences in outcome between the two treatments. The Svartberg, Stiles and Seltzer (2004) study was concerned with differences in mean effectiveness, and did not investigate variance in outcomes due to differences in interventions, therapists or patients.
The second article therefore investigated the relationship between therapists’ effort to orient patients towards affect, patients’ sense of self, and affect activation in the patient. Observations from sessions were nested within patients. This created a multilevel longitudinal design, allowing for multilevel mixed model analysis. The analysis also separated effects into within- and between effects. That is, both the effect of an individual’s standing on a measure relative to other individuals (between) and the effect of an individual diverting from the expected level for him or her at specific measurement point were investigated. The results indicated that the affective experiencing of a patient was predicted by the interaction of the therapist orienting a patient more towards affect than what the therapist was expected to do for a particular patient, and the patient experiencing a higher level of sense of self than what was expected for that patient in that session. The results have implications for therapist wanting to increase the affective experiences of their patients.
The third article investigated the relationship between therapists avoiding affect early in therapy (i.e. first four sessions), the bond as a component of the working alliance in session four, and outcome of therapy. Multiple regressions were applied to the total sample and to STDP and CT cases. The relationship between the variables differed in the treatments. In CT affect avoidance was positively related to both bond and symptom reduction. In STDP however, affect avoidance suppressed the relation of bond to symptom reduction, and also negatively influenced symptom reduction. The bond is often referred to as a common factor in psychotherapy, but appears to work differently in STDP and CT. This has implications for how one understands common factors. Although the bond may be a common factor, the manner in which it interacts with treatments to achieve benefits may not be common. Further, the finding can be used to inform therapists in the respective treatments on how to work with affect and bond. In CT, avoidance of affect increased the bond and contributed to outcome, whereas in STDP avoiding affect can be understood as helping the patient like the therapist better, but does not help the patient get better.