Client as expert: A Delphi poll of clients’ subjective experience of therapeutic alliance formation variables

Abstract

This study extends prior research that identified client-derived variables believed to contribute to therapeutic alliance development. Forty-two clients participated in a three-round-modified Delphi poll in which the subjective importance of 74 client-derived, common alliance formation variables was rated using a six-point scale. Participants consensually identified 23 variables as highly important, five as moderately important and three as low in importance to the formation of a strong alliance. Variables related to validation and asking about parts of the client’s life other than the presenting concern were considered as most important by at least one-third of the sample. Limitations of the study are presented and implications for theory, measurement, practice and future research are briefly discussed.

Keywords:

• clients’ subjective experience
• client’s perspective
• Delphi method
• therapeutic alliance
• working alliance

Acknowledgements

This research was supported by a grant from the Social Sciences and Humanities Research Council of Canada. Special acknowledgements are given to Dr. Meris Williams for sharing her Delphi expertise during the design of this study.

Notes

1. However, all categorical variables contained at least one cell with an expected value of < 5, which renders the results of chi-square analyses suspect, thus Fisher’s exact test was used.

2. The top five variables rated by each participant were re-coded as a 6 to indicate which variables were considered the most critical variables and to maximize the variability to counter any potential ceiling effects (as many items were rated as 5 out of 5 by participants). Regardless, the results remain highly consistent if we relied on the original 1 to 5 scale, with the exception that mean item ratings were necessarily lowered without a six-point value. For example, the level of high consensus was 40.6% vs. 41.9% if we include the 6th point. In addition, with removal of the derived 6th point on the scale, 16/23 items originally categorized as high consensus and high importance remained in that category with six out of the seven original items remaining at high consensus but with a moderate rating of importance, due to a lowering of the scale’s maximal value by a linear transformation of one point. Thus, even if we did not derive point number 6 on the rating scale, the conclusions reached in this study would be nearly identical.

3. Interquartile range is a measure of variability suitable for non-parametric procedures based on the difference between the upper and lower quartiles of a sequentially ordered set of variables.

4. As is often done with published Delphi research and due to space constraints, we have not presented the ratings of variables for the first and second rounds; these are available by request from the first author.