Meta-analysis of the relation between mental health professionals' clinical and educational experiences and judgment accuracy : review of clinical judgment research from 1997 to 2010

摘要

Researchers have addressed many clinician and client attributes in relation to theaccuracy of judgments made by mental health professionals. One such moderatoraddressed clinicians’ judgment accuracy in relation to experience. Contrary to whatmany clinicians expect, a number of studies have failed to demonstrate a positivecorrelation between judgment accuracy and experience (e.g., Berman & Berman,1984; Ruscio & Stern, 2005; Schinka & Sines, 1974). In Spengler et al. (2009), therelationship between judgment accuracy and experience was assessed via a largescalemeta-analysis that examined studies of clinical judgment and experience from1970 to 1996. The result was a small but reliable, homogeneous effectdemonstrating a positive correlation between judgment accuracy and experience.The Spengler et al. meta-analysis found relatively few significant moderator effectsinfluencing the experience-accuracy effect, namely the type of judgment made byclinicians, the criterion validity of accuracy measures used, and publication source.In the present study, results from clinical judgment and experience studies from1997 to 2010 were combined in a meta-analysis. An update and extension allowedfor cross-validation of the Spengler et al. meta-analysis with more recent researchas well as an exploration of additional moderator variables, such as profession typeand inclusion of non-mental health participants. The overall effect was .16, with a95 percent confidence interval that was above zero (CI = .05 to .26). This overalleffect indicated experience significantly impacted judgment accuracy, consistentwith expectations. The overall effect was shown to be heterogeneous, indicating theQ statistic was sufficiently large to reject the null hypothesis regarding homogeneityof the effect size distribution. Exploratory analyses revealed the presence of twosignificant moderator variables, namely judgment type and publication source.Limitations included lack of variability of judgment type and difficulty with orcomplete inability to assess other potential moderators of interest, such as feedbackand utilization of test protocols for the stimulus measure. Other limitationsincluded utilization of a less exhaustive search strategy, in which some relevantstudies may have been missed. Despite limitations, the results of the present metaanalysislargely replicated those of the Spengler et al. meta-analysis.