Supervisory working alliance trajectories and client outcome in Chinese trainees

ABSTRACT

We investigated the developmental trajectories of supervisory working alliance over time and their correlations with Chinese counseling trainees’ client symptom relief. Participants were 89 beginning counseling trainees from a master’s level training program in China. Results showed three clusters of developmental trajectories, including “Stable and Increasing,” “Rupture and Repair,” and “Bonding and Distancing.” Trainees with the first trajectory had greater client symptom relief. These results suggested the possible distinct clusters of developmental trajectories of SWA across beginning counseling trainees, and highlighted the potential benefit of maintaining a stable and strong SWA with trainees that could facilitate better client outcomes.

KEYWORDS:

• Supervisory working alliance
• developmental trajectory
• client symptom relief
• growth mixture modeling
• Chinese beginning counseling trainees

Disclosure statement

It is declared that the authors did not have any conflict of interest that might have influenced the results of this study.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/07325223.2022.2114968

Notes

1. Initially, we fitted a three-level model (session nested within trainees, who were nested within supervisors). The Intra-Class Correlations for SWA were .01 at the supervisor level, .63 at the trainee level, and .36 at the session level, suggesting that there were minimal variances of SWA that were explained at the supervisor level. Further, when estimating the growth model with linear and quadratic terms in Mplus 8.0 software (Muthen & Muthen, 2017), the program reported estimation errors which, upon further inspection, seemed to be caused by the extremely small variances at the supervisor level for the growth parameters. Based on these results, we decided to run a two-level model (session at the within level, trainee at the between level, see Figure 1(a)), while employing the “type = complex” command in Mplus to correct standard error estimation biases due to clustering at Level-3 (supervisor level).

2. Several criteria were used to compare these solutions (Lanza & Rhoades, 2013). First, the most appropriate model should have small Akaike Information Criteria (AIC) estimates and Bayesian Information Criteria (BIC) estimates. Second, all generated classes should have appropriate class sizes (more than 10% of the total sample size) and are conceptually interpretable so that they represent meaningful classes. Further, Nylund et al. (2007) recommended using the bootstrapped likelihood ratio test (BLRT) to assess whether the likelihood ratio values for solutions with different numbers of classes show statistically significant differences. However, BLRT is not available in Mplus for mixture modeling when the “type = complex” command is used to account for higher level clustering. Therefore, we relied on the first two criteria in deciding the best number of latent classes.

3. To statistically compare the three classes in terms of their client improvement coefficients, statistical texts recommended against conducting post-hoc analyses of variances (ANOVA) as if the latent class membership were an observed categorical variable (Asparouhov & Muthen, 2014; Lanza & Rhoades, 2013). Therefore, we did not conduct an ANOVA on the three group-specific mean values reported above and instead included the average client improvement coefficient as a therapist-level predictor in the GMM model, specifying a logistic regression of latent class membership onto client improvement at the between-trainee level.

One might feel that it is counterintuitive that client symptom improvement index is entered here as a “predictor variable” of SWA trajectory class membership, rather than as an “outcome variable” being predicted by class membership (as it is conventionally treated in statistical analyses). This is because, statistically, Mplus does not allow regressing the manifest client symptom improvement index onto the latent class membership variable “cb.” Moreover, conceptually, this study is correlational in nature and we cannot derive directionality or causality between the class membership of SWA developmental trajectory and the client improvement index from the conducted analyses. Thereby regressing Y onto X, or X onto Y, is conceptually equivalent and should be interpreted as X and Y being correlated or associated, rather than one leading or causing the other. Therefore, whether regressing the class membership to client improvement index, or the other way around, does not impact the substantive interpretation of the results that they are (or are not) associated with each other. In fact, using client symptom change as a “predictor” variable has also been done in existing studies, for example, Kivlighan and Shaughnessy (1995).

4. The likelihood change associated with one SD increase/decrease could be conceptualized as a proxy of effect size in this context of logistic regression.

Additional information

Notes on contributors

Xu Li

Xu Li is an assistant professor in Counseling Psychology at the Department of Educational Psychology in University of Wisconsin-Milwaukee. Xu Li earned his B.S. in Mathematical Sciences and his M.Ed. in Clinical and Counseling Psychology at Beijing Normal University in China. He then moved to the U.S. and obtained his Ph.D. degree in Counseling Psychology at the University of Maryland, College Park. His research focuses on the process and outcome of individual and group counseling, and the development and training of therapist trainees in diverse cultural contexts.

Chaihua Lin

Chaihua Lin is currently a faculty member in the Faculty of Psychology, Beijing Normal University, in Beijing, China. She obtained her M.Ed. in mental health counseling from University of Miami and her Ph.D. in counseling psychology from Purdue University in the U.S. Her research interests are multiculturalism, group counseling, interpersonal relationships, college student adjustment, and counselor training and development.

Manxuan Wu

Manxuan Wu is currently a doctoral student in Counseling Psychology at the Department of Educational Psychology in University of Wisconsin-Milwaukee. Manxuan earned her B.S. in Applied Psychology at China Women’s University in Beijing and her M.A. in educational psychology at the University of Minnesota in the U.S.

Feihan Li

Feihan Li obtained her Ph.D. and M.Ed. in counseling psychology from University of Missouri-Columbia, before becoming a faculty member in the Faculty of Psychology, Beijing Normal University, in Beijing, China. Her research interests are college student career development, international student cultural adjustment, and counselor training and development.