Go to Home Page

Other Methods Tried for Matching Clients and Therapists

General Description

In addition to the matching methods described on other pages on this site, several other matching methods were tried. They are summarized on this last page without much detail, because they either were very narrow recommendations for specific circumstances, or they did not work.

This is such an important and controversial area that it is fully discussed on the link above. The key issue is that there is a large percentage of therapists and therapy researchers who believe that there are certain treatments that have been proved to work for certain problems, and clients with these problems should be treated by therapists who specialize in these treatments. One of the name for this system is "empirically supported treatments", commonly known as "EST's." If I accepted this system, I wouldn't really need any other part of TMatch, since this would be the best method of matching clients to therapists. Needless to say, I don't accept the system of using EST's for matching. I summarize my reasons at the link above. In spite of not accepting matching diagnosis to treatments in general, there were two types that are so widely accepted that I did add them to the matching. They were that clients with phobias, agoraphobia and OCD should be matched to therapists who either use exposure techniques or refer to other therapists who use them, and clients with sexual dysfuncion should be matched to therapists who use certain behavioral techniques for these problems, or refer to other therapists who use them. This one aspect of matching has some potential to be retained for matching. This is explained in detail on the link above.

Findings from Project Match

Project Match was a multi-site research study which investigated the interaction of 21 different attributes of alcoholic clients matched to three different types of treatment: Motivational Enhancement Therapy (MET), Cognitive-Behavioral Therapy (CBT), or Twelve-Step Facilitation therapy (TSF) (Project Match, 1998). Most attributes did not show significant interaction effects. However, there were two effects with which the researchers had strong confidence. First, clients high in anger did better in MET than in the other two treatments. The researchers hypothesized that this was because MET is nonconfrontational and less likely to provoke them. Second, clients whose social network was more supportive of drinking did better with TSF than with MET. The researchers hypothesized this was because TSF attempts to get clients involved in Alcoholics Anonymous, which replaces social networks that support drinking with social networks that don't. Based on this, I added the two matching criteria listed below.

  1. clients who tend to have a lot of anger should be matched with therapists who practice therapy that is particularly non-confrontational, such as Motivational Enhancement Therapy (MET).
  2. Clients who have alcohol or other addiction problems, and also have social networks that support these addictions, should be matched with therapists who work with twelve step programs, or have other methods of replacing these social networks with more benign ones.

The first matching recommendation worked well enough to be retained, although the assessments used for client anger and therapist amount of confrontation need to be improved. The second matching recommendation did not work, because all therapists reported that they referred clients to 12-step programs when appropriate, and that they were sensitive to the need to replace clients' social networks that support addictions.

Affiliation and Control

In this theory of affiliation and control (Berzins, 1977; Kiesler, 1992), affiliation refers to friendliness, and ranges from very hostile to very friendly, while control measures dominance versus submissiveness. The matching recommendations are that dominant therapists should be matched to submissive clients, submissive therapists to dominant clients, friendly therapists to friendly clients, and hostile therapists to hostile clients. Neither of these ideas worked well in TMatch, mainly because the methods used to measure these qualities didn't work. In Tmatch, all therapists described themselves as completely friendly. If there are no hostile therapists, then the affiliation matching recommendation isn't useful. For measuring dominance vs. submissiveness, I made up four questions, and did not get a very large range of answers, so the matching didn't prove much one way or the other. This still might be an interesting area for futher testing, but only if a short valid test for control could be found. In the meantime, I am dropping this area of matching.

Client Tolerance for Treatment Complexity

In a study of matching clients with either a complex coping skills training or a relatively unstructured interpersonal therapy, Cooney, Kadden, Litt, and Getter (1991) found that patients with high cognitive impairment had better outcomes in interpersonal therapy, and those with low cognitive impairment did better in coping skills training. They hypothesized that the coping skills training was too complex for cognitively impaired clients. I extrapolated this hypothesis to the idea that clients would have different tolerances and preferences for complexity in therapy, and therapists' tendency toward complexity in therapy was matched with clients' tolerance for complexity. This did not work well in TMatch. There was no existing accepted test for this type of measurement, so I made up my own questions. The indications from the study of Tmatch were that clients had a very strong tendency to think they tolerated complexity very well, even when they didn't. This still seems like a promising area of matching, but it will have to wait for a better, more valid methods of assessment, especially for clients.

Clients' Perceptions of Reality

Goldfried (1991) found that therapists using different systems of therapy tended to have different ways of perceiving and communicating client's perceptions of reality. Cognitive-behavioral therapists communicated to clients that things were not as bad as the clients thought, while psychodynamic-interpersonal therapists communicated that things were worse than the clients thought. This finding led me to the hypothesis that clients who tended to see things as much worse than they really are should be matched with therapists who tended to communicate to clients that things are better than clients think, while clients who tended to see things as better than they really are (e.g., who might have problems that they are having trouble acknowledging) should be matched with therapists who tended to communicate that things are worse than clients perceive. Although this seems on the face a reasonable way to match, it was hard to carry out in TMatch because there was no existing accepted test for measuring these qualities. Therefore, I made up the questions. My questions did not produce clear and convincing differentiations among clients and therapists on these dimensions. This idea is worth trying again, but it would take a lot of research to produce valid assessment questions.

Gunderson's personality dimensions

Gunderson (1978) studied the effect on outcome of various personality matches between therapists and schizophrenic patients. His major finding was that therapists who were composed, contained, and stable, as opposed to frenetic and disorganized, did particularly well with anxious patients. His second major finding was that therapists who were comfortable with aggression had good results with hostile patients. He also found that therapists who were very comfortable with depression from a personal standpoint did well with depressed patients. From his study results, he also proposed that a charismatic or very optimistic therapist could be intimidating to a patient with a strong sense of failure. Although this study was only with schizophrenic patients, the results were clear enough and seem intuitively obvious enough that they warranted a try in TMatch. Therefore, I included the following matching condtions:

  1. Anxious clients should be matched to therapists who tend to be composed, contained, and stable, as opposed to frenetic and disorganized.
  2. Hostile clients should be matched to therapists who are particularly comfortable with aggression.
  3. Depressed clients should be matched to therapists who are comfortable with depression.
  4. Clients with strong senses of failure should be matched to therapists who are not charismatic or extremely optimistic.

The study of TMatch indicated the matching conditions 1 and 2 have some potential, and are worth retaining, at least for further testing. Condition 3 did not work because all therapists assessed themselves as being completely comfortable with depression. This is probably true, since any therapist not comfortable with depression would probably have quit the profession. Condition 4 did not work well for a similar reason. All therapists described themselves as fairly charismatic, and extremely optimistic. Charisma seems to not be a quality that can be easily self-rated. If there was a reasonable way of observer-rating therapist charisma, this idea would be interesting to try again, at least in terms of this particular trait. The trait of therapist optimism was useless for matching. It is likely that therapists by and large are optimistic, or they would not be able to remain as therapists.

Subjects Usually Talked About in Therapy

One of the methods used to try to give clients preferences for different therapy characteristics was a section in which clients and therapists each divided 16 different possible subjects of discussion during therapy into four classes, from most preferred to least preferred. Clients were matched to therapists for this criterion based on how much the orders of these talk-subjects matched each other. This matching criterion did not work. The study of TMatch indicated that therapists tended toward not directing or controlling what was discussed during therapy, so matching clients' preferences for what is discussed to what therapists get clients to discuss isn't productive. Also, therapists were confused by this section, and many rated the talk subjects by what they preferred the clients would discuss, which was not the intention for the list, and made it less useful. There were other problems with this criterion that are too complex to go into here. The conclusion was that this was an interesting criterion to try, but should not be retained for future versions of TMatch.

For more information, email Kenneth Frankel, Ph.D.


Berzins, J. I. (1977). Therapist-patient matching. In A. S. Gurman & A. M. Razin (Eds.), Effective psychotherapy: A handbook of research (pp. 222-251). New York: Pergamon.

Cooney, N. L., Kadden, R. M., Litt, M. D., & Getter, H. (1991). Matching alcoholics to coping skills or interactional therapies: Two-year follow-up results. Journal of Consulting and Clinical Psychology, 59, 598-601.

Goldfried, M. R. (1991). Research issues in psychotherapy integration. Journal of Psychotherapy Integration, 1, 5-25.

Greenberg, L. S., Elliott, R., Watson, J. C., & Bohart, A. C. (2001). Empathy. Psychotherapy, 38(4), 380-384.Greenberg, Elliot, Watson, & Bohart, 2001

Gunderson, J. C. (1978). Patient-therapist matching: A research evaluation. American Journal of Psychiatry, 135(10), 1193-1197.

Kiesler, D. J. (1992). Interpersonal circle inventories: Pantheoretical applications to psychotherapy research and practice. Journal of Psychotherapy Integration, 2(2), 77-99.

Project Match Research Group (1998). Matching alcoholism treatments to client heterogeneity: Project Match three-year drinking outcomes. Alcoholism Clinical and Experimental Research, 22(6), 1300-1311.