Comparison of Estimated Variance Components Between Different Designs Using Multivariate Generalizability Theory Approach (Poster 28)

• In Event: AERA Poster Session Wednesday 10:50 a.m.
  In Poster Session: Measurement, Psychometrics, and Assessment: Poster Session I

Abstract

Estimation of variance components may differ in different designs. This study explores how estimated variance components of a well-being scale differ in nested country design, undifferentiated fixed country design, and separated fixed design using a generalizability theory framework. Findings suggest that estimated variance components are the greatest for the undifferentiated design and smallest for the nested design, which indicates that the randomizing facet of the country design would lower the reliability compared to fixing the facet of the country. Although randomizing countries have lower reliability, the generalizability coefficients illustrate the applicability of the scale across countries. Variance components for the dimension of resiliency are more applicable than motivation across countries, showing smaller differences for resiliency than the motivation dimension.

Author

• Jeongmin Ji, University of Iowa