SEM-Based Composite Reliability Estimates of the Crisis Acuity Rating Scale with Children and Adolescents

Abstract

A Likert-type rating scale of crisis acuity was completed by on-call clinicians with 227 children and adolescents admitted consecutively to a youth crisis unit of a rural community mental health center. Traditional methods of estimating reliability for the scale were precluded. In particular, the scale was not found to be unidimensional when subjected to confirmatory factor analysis. This rendered Cronbach's alpha coefficient inappropriate. A six-factor oblique model corresponding to apriori subscales was supported. SEM-based reliability estimates for this confirmed factor structure using a phantom composite was detailed. The results yielded a point estimate of .845 with a bootstrap (1000 samples) 90% confidence interval between .819 and .867.