Path analysis of attachment style and therapeutic assessment outcomes

John M Poston, Purdue University

Abstract

This intervention study examined the predictive utility of attachment anxiety and avoidance on clients' benefit/outcome from Therapeutic Assessment (TA; Finn, 1996, 2007). This study also examined whether defensive exclusion, a concept from Bowlby's (1988) attachment theory, helped explain overall benefit/outcome. Defensive exclusion occurs when an individual selectively ignores internal and external relational cues to suppress attachment system activation (Bowlby, 1988). In this study, 39 college students participated in a manualized, 3-session TA intervention and completed a demographic questionnaire and five measures: the Experiences in Close Relationship scale (ECR; Brennan, Clark, & Shaver, 1998), as a measure of attachment anxiety and avoidance; the Outcome Questionnaire-45.2 (OQ-45.2; Lambert et al., 1996), as a measure of psychological distress; the Assessment Questionnaire-2 (AQ-2; Finn, Schroeder, & Tonsager, 1995), as a measure of participants' new self-understanding and relationship with the examiner; the thought-listing technique (Cacioppo & Petty, 1981), as a measure of defensive exclusion; and the Counselor Rating Form-Short version (CRF-S; Corrigan & Schmidt, 1983) as a measure of counselor credibility. The OQ-45.2 was administered pre-/post-, for comparison purposes. Path analysis was conducted to test the hypothesized model fit. Specifically, it was hypothesized that (a) attachment anxiety and avoidance would negatively predict participants' improvement in psychological distress, new self-understanding, and positive relationship with the examiner; (b) attachment avoidance would have greater predictive utility than attachment anxiety; and (c) defensive exclusion would explain the relationship between attachment avoidance and TA outcomes. Also, a second hypothesis was tested regarding Holland Codes, a client variable shown to have predictive significance in testing (Kivlighan & Shapiro, 1987). Specifically, a partial F-test was used to test whether Holland codes added significantly to a model containing attachment variables as predictors. Results indicated that (a) a minimally re-specified model adequately fit the data (χ2 = 1.18, p = 0.88; RMSEA = 0.00 [95% CI: 0.00, 0.12], CFI = 1.00, TLI = 1.00); (b) attachment anxiety (β = 0.31, p < 0.01) and relationship with the examiner (β = 0.57, p < 0.01) significantly predicted participants' new self-understanding; (c) predictor variables collectively explained 4% of the variance in change in psychological distress, 42% of the variance in new self-understanding, and 3% of the variance in relationship with the examiner; and (d) Holland codes did not add significantly to a model already containing attachment styles (F [6, 30] = 1.30, p = .29). Implications for TA theory, research, and practice are discussed.

Degree

Ph.D.

Advisors

Hanson, Purdue University.

Subject Area

Counseling Psychology|Clinical psychology

COinS