The psychometric properties of the difficult behavior self-efficacy scale.

The study was designed to estimate the psychometric properties of Hastings and Brown's (2002a) Difficult Behavior Self-efficacy Scale. Participants were two samples of physical educators teaching in Korea (n=229) and the United States (U.S.; n=139). An initial translation of the questionnaire to Korean and pilot study were conducted along with the larger study using a confirmatory factor analysis procedure. Internal consistency estimates (weighed Omega) for the five-item scale were 0.88 both the Korean and U.S. samples. The average variances extracted for the one factor were 0.59 for the total data set and 0.57 each for the Korean and U.S. samples. Confirmatory factor analysis supported a five-item, unidimensional model for self-efficacy for the total sample: Goodness of Fit Index (GFI)=0.97, Nonnormed Fit Index (NNFI)=0.95, Comparative Fit Index (CFI)=0.98, and Standardized Root Mean Square Residual (SRMR)=0.03. Only the Root Mean Square Error of Approximation (RMSEA=0.12) fell below criterion levels of acceptable fit, with similar fit indices occurring in separate analyses of the Korean and U.S. samples. Invariance testing across the two samples supported metric invariance (similarity of factor loadings) but not scalar invariance (U.S. means higher on all five items). The factor structure for the self-efficacy scale provides an initial estimate of validity and internal consistency for use with different teacher groups.