The effects of therapist negativity

By Dr. Matthew Bennett copyright 1996

There are few research questions in psychology which are quite as complex or as important as psychotherapy outcome research. Since the time the firsts psychotherapists first began to work with the firsts psychotherapy patients, the process of therapy has retained an almost mythic air of mystery among the general public. For many decades, it seems, psychotherapists themselves appear to have been content to let it remain so…even among themselves.

Long after researchers began to learn to quantify the complex phenomena surrounding the many levels of human communication, the process of psychotherapy remained an enigma, as if protected from the standardization that has become the norm in medicine, law, and other professions.  Recent changes in cultural and professional reality have begun to demand changes in this attitude. As psychotherapists have become increasingly beholden to account for the quality and quantity of their services, there has been a special interest in methods of measuring the specific mechanisms and general effectiveness of the process of psychotherapy.

Much of the interest in psychotherapy outcome has focused on increasing understanding of successful “matchmaking” between patient and therapist and the dynamics of the patient-therapist relationship. The watershed Vanderbilt Psychotherapy Outcome studies have contributed much to illuminating the dynamic therapeutic relationship with a view to understanding when and why psychotherapy is most effective. A review of some of the most general basic findings to come out of such studies follows:

1. The therapeutic relationship resembles any human relationship in that it constitutes the interaction of two separate individuals who each bring their own personality and interactional styles to the relationship. For this reason psychotherapy outcome research has been a particularly complex arena of investigation, and it has been difficult to associate certain therapist characteristics, for example, with positive outcome: because the therapist is not the only person in the relationship, and nor is her or she the sole determining factor of successful outcome (Najavits & Strupp, 1994). In this sense it has been no less difficult
to predict successful therapy dyads than to predict successful marriages.
2. The interpersonal systems which arise during psychotherapy become a third variable beyond the individual influences brought by each party in the relationship. There is ample evidence that patients and therapists affect each other, such that the therapy relationship is, in effect, “greater than the sum of its parts.” This idea has its ancestry in the well known concepts of transference and countertransference, and has grown in sophistication to include such variables as shared values (Kelly & Strupp, 1992), internalized interpersonal styles (Harris, Quintana, Strupp, & Henry, 1994), and complementary introject (Henry, Schacht, & Strupp, 1990).
3. Specific modalities of the therapeutic relationship may act to influence therapeutic outcome. Since the therapeutic relationship (alliance) is by definition a purposive one, the main concern underlying research into the manifold aspects of the relationship is impact on therapeutic outcome. One of the most salient conclusions drawn from the Vanderbilt studies concerns complementarity (Strupp, 1993). Within the therapeutic alliance, as in all relationships, it has been shown that intrapsychic functioning influences and is influenced by interpersonal transactions such that “some behaviors (and self-concepts) are mutually reinforcing” (Talley, Strupp, & Morey, 1990, p. 182). Thus it can be seen that the dynamic interaction characterizing the therapeutic alliance is a complex system encompassing both patient and therapist self-images and interpersonal styles.
4. Given knowledge of the implications of self-image and interpersonal style of both client and therapist, it is possible to determine which alliance pairings are most-likely-to-succeed. Contemporary attachment theory posits that primal relationships with parents or caretakers shape successive relationships throughout life…including the therapeutic alliance. Clearly, then, a successful client-therapist match involves variables which include fundamental aspects of personality. Happily, it does not seem to be the case that only therapists of certain attachment styles can be paired with certain other patients. It is part of the training of a psychotherapist to learn to adapt his or her behaviors within the therapeutic alliance. Dolan, Arnkoff, & Glass (1993) take the position that “the therapist must have a thorough understanding of a client’s attachment patterns and internal models so as to avoid repeating those patterns with the client” (p. 409). However, these researchers do not necessarily mean that the effective therapist takes an interpersonal style diametrically opposed to the attachment style of the client. On the contrary, Dolan et al. show through case study that careful adaptation of therapist behaviors in the direction of the client’s attachment style can be conducive to successful outcome. For example, a very warm, intense therapeutic style may work well for one who is securely attached, but may cause preoccupied clients to distance themselves in self-protection.

It is the intention of the present study to illustrate such adaptive complementarity as a therapeutic process variable which contributes to positive outcome. A possible indication of the mechanism of adaptive complementarity was discovered inadvertently by Ramey (1996) who, in the course of another investigation, found that certain psychotherapy patients’ measures of anxiety actually decreased in the face of significant negativity expressed by their therapists. This study will attempt to show that the decrease in anxiety was reflective of a synergistic convergence between the patient’s internal working model and the therapist’s interpersonal behavior within the alliance. Thus the following two hypotheses are forwarded:

1. The patients who experienced decreased anxiety with therapist negativity had negative “self” images and positive “other” images, constituting preoccupied attachment style. Thus these patients will have higher “preoccupied” ratings than patients who experienced increased anxiety with therapist negativity.
2. The group which experienced decreased anxiety as a result of therapist negativity will be characterized by better therapy outcome than that which experienced increased anxiety.


Subjects and therapists were drawn from the Vanderbilt II psychotherapy research project (Strupp & Binder, 1984). Ten subjects were drawn from an available pool of 64 on the basis of two prerequisites: high therapist negativity as measured by the Vanderbilt Negative Indicators Scale (VNIS; Suh, Strupp & O’Malley, 1986) and increase or decrease in expressed anxiety as indicated by the SCL-90-R (Derogatis, 1977). The six patients who decreased in anxiety (including the same patients investigated in Ramey, 1996) constituted the experimental group; the remaining four who decreased in anxiety made up the control group. “High therapist negativity” was defined as a z-score of at least .6 on the VNIS at time of therapy; increases or decreases in anxiety were measured from pre-treatment to post-treatment SCL-90-R measures. All patients met screening requirements for inclusion in the Vanderbilt II study, including presence of significant interpersonal difficulties and an Axis I or II diagnosis as determined by the DSM-III (American Psychiatric Association, 1980). Average therapist negativity scores for the experimental group were 1.596 (range .623-3.670), and for the control group 2.083 (range .876-3.924); there was no significant difference between the groups on this measure. Average anxiety scores for the experimental group were 14.5 at pretreatment (range 7-21) and 3.167 at posttreatment (range 0-6). Average anxiety scores for the control group were 11.75 at pretreatment (range 0-32) and 15.25 at posttreatment (range 5-36).


The VNIS were ratings assigned by independent reviewers who observed video taped recordings of set selections of the therapy sessions. The SCL-90-R is a self report symptom inventory which is designed as a measure of general psychological distress. Therapy outcome decisions (good, moderate, poor) were based on significant overall symptom reduction as measured by the SCL-90-R.


Probability of type I error was maintained at .05 for all analyses.

An independent sample t-test on preoccupied attachment ratings revealed a significant difference between the anxiety-decrease (experimental) group and the anxiety-increase (control) group (p = .05). Therapeutic outcome results were compared for the two groups using chi-square analysis for independent samples: the groups were found to differ significantly with respect to therapeutic outcome (X2 = 7.916, df = 2, p < .05).


Statistical analysis of this small sample of psychotherapy patients fulfills the two hypotheses set offered in this study. Patients who reacted to therapist negativity with decreased anxiety were shown to have negative “self” and positive “other” attachment styles, supporting the hypothesis that complementary relationships between attachment style and therapist behavior were characterized by less “affective dissonance” and thus less anxiety. That this kind of complementarity is conducive to positive outcome is also strongly indicated in this study: the therapeutic outcome was most often good when therapist behavior matched internal interpersonal working models. When therapist behavior was not in accord with patient attachment style, greater affective dissonance (more anxiety) ensued, and the end result was more often a poor therapeutic outcome. These conclusions are in keeping with the findings of Dolan et al (1993).

Further research might well investigate the influence of therapist behavior – patient working model complementarity on other symptoms. Another possibly fruitful area of research is the effects of therapist negativity on therapy outcome with securely attached therapists. Such knowledge of the hitherto inviolable secrets of psychotherapy are certain
to increase its effectiveness as a healing art.