Investigating Impacts of Measurement Noninvariance on Classification Based on Scale Scores (Poster 47)

• In Event: Sunday Poster Session 1:15 pm
  In Poster Session: Educational Statisticians SIG Poster Session

Abstract

This study investigates impacts of measurement non-invariance in items on classification based on scale scores. A simulation study is designed to assess impacts of measurement non-invariance under different data conditions and different classification scenarios. The manipulated factors include reliability of scale scores, latent factor distributions in each group, percentage of non-invariant items, type of non-invariance, magnitude of non-invariance, and direction of non-invariance. Preliminary results revealed that the impact of non-invariance depends on the degree of non-invariance at scale score level and varies between different categories.

Authors

• Lu Liu, Florida State University

• Yanyun Yang, Florida State University