Score Reliability for Scales Consisting of Ordinal Items: A Bayesian Structural Equation Modeling Approach

• In Event: 34.079 - Roundtable Session 14
  In Roundtable Session: 34.079-11 - Ordinal and Binary Data: From Differential Item Functioning to Classification and Regression Tree

Abstract

When item scores are ordered categorical, the relationship between the observed items and the latent factors is no longer linear. To calculate the score reliability for scales with ordinal variables, nonlinear reliability has been proposed in the previous literature. However, the nonlinear reliability estimation has been found to be inaccurate when the sample size is small and the categorical distribution is extreme. This simulation study investigates the nonlinear reliability estimated using Bayesian structural equation modeling in hopes of improving the estimation accuracy of the nonlinear reliability coefficients in small samples with asymmetric distributions of ordinal variables. This study is implemented with 2 scale lengths, 2 numbers of categories, 4 sets of thresholds, and 4 sample sizes with different prior specifications.

Authors

• Yanyun Yang, Florida State University

• Yan Xia, Arizona State University - Tempe