Both volunteer and nonvolunteer subjects were randomly assigned to one of two conditions. In the choice condition, subjects were led to believe that they had a choice between two hypothetical relaxation treatments that both they and another subject would receive. Subjects in the second condition were denied the choice of treatment, receiving the treatment chosen by another subject. All subjects, regardless of choice, actually received the same taped relaxation treatment. As predicted, volunteer subjects who were given a choice between treatments significantly valued the treatment more and reported the treatment to be significantly more effective than volunteer subjects who had no choice in the matter. Nonvolunteer subjects, viewing choice as less important than volunteer subjects, were not significantly affected by the subsequent responsibility manipulation, whereas volunteering may be a reliable source of bias, the degree to which the volunteer feels responsible for the success of the outcome appears to predict the direction of the bias. These findings are relevant to both the clinician and to clinical outcome research.

The act of voluntarily engaging in therapeutic treatment, coupled with feelings of responsibility for its success, may serve to affect one's perception of the actual outcome of the treatment. The individual who freely chooses to invest time, effort, and money in psychotherapy, as well as the therapist who invests time, money, ego, and training in the practice of psychotherapy, may be prone toward a positive outcome bias.
Cognitive dissonance theory (Aronson, 1969; Brehm & Cohen, 1962; Festinger, 1957) pre-dicts that subjective evaluation is partly a function of a person's effort, investment, and feeling; of responsibility for choice. The need to justify one's efforts and important choices may be satisfied by perceiving the chosen alternative as more valuable. A client who freely chooses a therapeutic treatment and feels responsible for its success may, according to dissonance theory, perceive it as more valuable. A treatment that is highly valued should be more effective. Schroeder (1960) found that acceptance of responsibility was positively related to client improvement. Fee-paying clients have been found to benefit more from psychotherapy than non-fee-paying clients (Goodman, 1960; Rosenbaum, Friedlander, & Kaplan, 1956),
This research was designed to better under-stand those factors that may lead to a system-atic source of bias in evaluating the value and effectiveness of psychotherapy. The two factors studied here, volunteering and responsibility for choice, were selected since they appear to be important in producing cognitive dissonance. A person who freely volunteers for treatment may consider the treatment more valuable than a nonvolunteer. The volunteer's subsequent choice concerning treatment and the feelings of responsibility for that choice may further the need to justify the treatment as a good one. Therefore, it was hypothesized that individuals who volunteered for treatment would be more affected by feelings of responsibility for choice-as indicated by their perceived effectiveness of treatment and the value they assigned to treatment-than non-volunteers. It was predicted that volunteers who were placed in the position of responsibility for choice by being asked to decide between two treatments--would view the treatment as more valuable and more effective than the volunteers with no choice. It was expected that those volunteers who were denied choice would tend to devalue treatment as in psychological reactance (Brehm, 1966).

METHOD

SUBJECTS

The subjects, 15 volunteers -7 males and 8 females and 15 nonvolunteers 12 males and 3 females were pooled from an undergraduate psychology class at Temple University. Initially, the experimenter informed the students of a relaxation training session and solicited volunteers. A volunteer sign-up sheet was obtained, and a list of those students who were present was supplied by their professor. Later in the week, the professor told the class that they would receive extra credit for participating in psychology experiments and that participation in at least one study was encouraged. Those students who did not volunteer at first but who later participated in the experiment were considered “nonvolunteers.”

PROCEDURE

Prior to scheduling subjects for treatment, both volunteers and nonvolunteers were randomly assigned to cither of two treatment groups. One treatment group was given a choice between two hypothetically different relaxation treatments, and the other group received treatment chosen by fellow subjects. The subjects were then scheduled for treatment in pairs so that both conditions would be represented. In those cases when, due to scheduling difficulties, only one condition was represented, a confederate acted as the subject in the other condition.
On entering the laboratory, the paired subjects were informed of two supposedly different types of audiotaped relaxation treatments--"neuroglandular" and "cardiovascular." The experimenter then asked the subject in the choice treatment group to pick the treatment tape he/she would prefer to receive. Surprisingly, many of the subjects had a definite preference. After the subjects made a choice, the appropriately labeled tape was played for both subjects on a tape recorder. Both tapes were, in fact, the same 20-minute tape of Jacobsen's (1938) method of systematic relaxation.
Several self-report bipolar Likert scales were administered: (a) "How relaxed are you right now?" (1 = "not at all" and 10 = "completely," pretreat-ment and posttreatment), (b) "How valuable was the treatment?" (1="not at all" and 10="extremely," posttreatment), and (c) "How interested are you in having another session ?" (1= "not at all" and 4= “very much so," posttreatment). This last question served as a check on the reliability of the volunteering variable. Those who initially volunteered were expected to favor further treatment. Subjects were debriefed following the experiment, and those who had indicated an interest in further treatment were assured that additional treatment would be made available.

RESULTS

A 2 X 2 (volunteer X responsibility) analysis of variance design was used to analyze the data on the subjects' perceived value of the relaxation treatment they had received. The results indicated a significant Volunteer x Responsibility interaction, F (1, 26) = 4.222, ? < .05, with no significant main effects. As predicted, a simple effects analysis showed that volunteers who had a choice of treatment significantly rated the treatment as more valuable (M = 9.4) than those volunteers who had the treatment chosen for them (M = 7.0), F( 1,26) = 5.36, ? < .03. There was no significant difference between nonvolunteers who had a choice (M = 8.3) and nonvolunteers who did not have a choice (M = 8.9), F(1, 26) < 1.

A 2 X 2 x 2 (volunteers x responsibility x time) analysis of variance design was used to analyze the repeated-measures data. The prediction was unidimensional involving a one-tailed test because data from a pilot study had shown that subjects become more relaxed over time when given the opportunity to do so. Overall, the subjects did report becoming more relaxed, F(1; 26) = 14.65, ? < .0005 (pretreatment M = 6.8, posttreatment M = 8.7). None of the other main effects or two-factor interactions were significant. The expected three-factor interaction was significant. F (1, 26) = 3.20, ? <.04, indicating that not all groups were reporting increased relaxation proportionately. Simple main effects of treatment over time indicated that volunteers with choice reported the greatest effects from treatment, F(1, 26) = 18.46, ? < .0005 (pre-treatment M = 6.0, posttreatment M = 9.1), whereas volunteers without choice reported a slight but nonsignificant increase in relaxation, F(1, 26) = 2.73, ? < .25 (pretreatment M = 6.6, posttreatment M = 7.9). They were the only group to show no significant effects. Nonvolunteers with choice, F (1, 26) = 6.65, ? < .02 (pretreatment M = 7.4, posttreatment M = 9.3), reacted similarly to nonvolunteers without choice, F (1, 26) = 7.60, ? < .02 (pretreatment M= 7.3, posttreatment M = 9.4). Both groups indicated a significant increase in relaxation.
A 2 X 2 (volunteer X responsibility) analysis of variance of the subjects' interest in returning for another session revealed only a significant volunteer main effect, F(1, 26) = 7.04, ? < .025; r =.46.

Volunteer subjects, regardless of their positions of responsibility for choice in the first treatment session, were interested in coming for more treatments (M = 3.7). The nonvolunteers were less likely to volunteer to return (M = 2.9). This result supports the reliability assumption of the volunteering factor.

The perceived value of treatment showed a significant positive correlation with the self-reported change in relaxation, r (28) = .40, ? < .025. The sex of subjects did not correlate with any of the dependent measures-(a) change in relaxation (r = -.09), (b) value (r = .O1), or (c) interest in further treatment (r = -.15).

Discussion

Those individuals who voluntarily sought therapeutic treatment were significantly affected by the degree to which they experienced responsibility concerning the choice of treatment. Volunteers who were given a choice of treatment, and whose choice also affected someone else, reported highly significant positive change in their degree of relaxation and valued the treatment significantly more than volunteers who were not given a choice of treatment. Volunteers who were denied choice were the only subjects reporting no significant effects from treatment. This resistance to treatment may be an example of psychological reactance (Brehm, 1966). Nonvolunteers were not significantly affected by the subsequent manipulation of choice between treatments.

Volunteers, regardless of their positions of responsibility for choice in the experiment, were still likely to volunteer for continued treatment, whereas nonvolunteers were much less likely to be interested in further sessions. This result supports Rosnow and Rosenthal's (1974; Rosenthal & Rosnow, 1975) argument that volunteering is a reliable and systematic source of bias. However, the direction of the bias is not always clear. These results suggest that the direction of the bias might be predictable if more information concerning the volunteer's position of responsibility were known. When volunteers were in positions of greater responsibility for the success of treatment, there appeared to be a positive bias. However, volunteers who had been denied responsibility of choice, perhaps feeling manipulated, produced negatively biased data.

This research illustrates a methodological problem inherent in clinical outcome research. Outcome research often depends on the data of those individuals, both clients and therapists, who choose to be in the therapeutic situation. However, although feelings of responsibility may be a nuisance variable for the researcher, it may be invaluable for the therapist, in that a client's feelings of responsibility for treatment success may influence the effectiveness of therapy.

References

Aronson, E. The theory of cognitive dissonance: A current perspective. In L. Berkowitz (Ed.) Advances in experimental social psychology (vol. 4). New York: Academic Press, 1969.

Brehm, J. W. A theory of psychological reactance. New York: Academic Press, 1966.

Brehm, J. W. & Cohen, A. R. Explorations in cognitive dissonance. New York: Wiley, 1962.

Festinger, L. A theory of cognitive dissonance. Stanford: Stanford University Press, 1957.

Goodman, N. Are there differences between fee and non-fee cases? Social Work, 1960, 5, 46-52. Jacobsen, E. Progressive relaxation. Chicago: University of Chicago Press, 1938.

Rosenbaum, M., Friedlander, J., & Kaplan, S. Evaluation of results of psychotherapy. Psychosomatic Medicine, 1956, 18, 113-132.

Rosenthal, R., R Rosnow, R. L. The volunteer subject. New York: Wiley, 1975.

Rosnow, R. L., & Rosenthal, R. Taming of the volunteer problem: On coping with artifacts by benign neglect. Journal of Personality and Social Psychology, 1974, 30, 188-190.

Schroeder, P. Client acceptance of responsibility and difficulty of therapy. Journal of Consulting Psychology, 1960, 24, 467-471.