A method for obtaining standard errors and confidence intervals of composite reliability for congeneric items

A method for obtaining standard errors and confidence intervals of composite reliability coefficients based on the bootstrap method is proposed. Using a structural equation modeling framework for estimating composite reliability of congeneric measures (Raykov, 1997), the approach takes repeated samples from a given sample and then estimates reliability. This approach extends earlier research on the distribution of scale reliability that is primarily concerned with Cronbach's coefficient alpha, and represents a general method based on fewer assumptions. It also permits testing hypotheses about composite reliability. Its use is demonstrated on simulated data.