Unique effects and moderators of effects of sources on self-efficacy: A model-based meta-analysis.

Self-efficacy beliefs are strong predictors of academic pursuits, performance, and persistence, and in theory are developed and maintained by 4 classes of experiences Bandura (1986) referred to as sources: performance accomplishments (PA), vicarious learning (VL), social persuasion (SP), and affective arousal (AA). The effects of sources on self-efficacy vary by performance domain and individual difference factors. In this meta-analysis (k = 61 studies of academic self-efficacy; N = 8,965), we employed B. J. Becker's (2009) model-based approach to examine cumulative effects of the sources as a set and unique effects of each source, controlling for the others. Following Becker's recommendations, we used available data to create a correlation matrix for the 4 sources and self-efficacy, then used these meta-analytically derived correlations to test our path model. We further examined moderation of these associations by subject area (STEM vs. non-STEM), grade, sex, and ethnicity. PA showed by far the strongest unique association with self-efficacy beliefs. Subject area was a significant moderator, with sources collectively predicting self-efficacy more strongly in non-STEM (k = 14) compared with STEM (k = 47) subjects (R2 = .37 and .22, respectively). Within studies of STEM subjects, grade level was a significant moderator of the coefficients in our path model, as were 2 continuous study characteristics (percent non-White and percent female). Practical implications of the findings and future research directions are discussed. (PsycINFO Database Record

Influence of social class perceptions on attributions among mental health practitioners.

OBJECTIVE: A vignette-based study assessed the influence of social class attributions toward a hypothetical client's difficulty. METHOD: 188 licensed mental health professionals who were recruited through professional listservs completed an online survey after reviewing one of two versions of a vignette describing a hypothetical client that varied based on social class cues. RESULTS: As expected, this sample of licensed mental health practitioners detected social class differences based on the descriptors of the hypothetical client across the two vignettes. These perceived social class differences, however, did not impact participants' attributions toward the client for causing or solving her problems, level of Global Assessment of Functioning score ascribed to the client, or willingness to work with the client. CONCLUSIONS: There was no evidence that participants differentially ascribed attributions based on social class. Implications and directions for future research are provided.