Bootstrap Confidence Intervals for Range-Restricted Coefficient Alpha

• In Event: Poster Session 7
  In Poster Session: Division D, Section 2 Poster Session

Abstract

Coefficient alpha is a biased estimator of its true unrestricted value, when a sample is subjected to range restriction. This bias can be adjusted by the homogeneity formula. However, it does not provide a procedure estimating the confidence intervals (CIs) of the corrected coefficient alpha. Thus, the present study seeks to apply the bootstrap procedure to obtain the CIs. Two Monte Carlo studies were conducted under different manipulated conditions, including measurement data, number of items, sample size, selection ratio, coefficient alpha, data distribution, and truncation type. Results showed that the bootstrap procedure was found to be accurate, thereby providing a useful method for estimating the CIs of the range-restricted coefficient alpha.

Authors

• Johnson Ching-hong Li, University of Alberta

• Ying Cui, University of Alberta

• Mark J. Gierl, University of Alberta

• Wai Chan, Chinese University of Hong Kong