Examining Wording Effects on a Self-Concept Scale From TIMSS 2015

• In Event: 15.059 - Fairmont Royal York Roundtable Session One
  In Roundtable Session: 15.059-7 - Specialized Structural Equation Modeling Applications: Heterogeneous Cross-Lagged Data, Generalized Structured Component Analysis, and Item Wording Effects

Abstract

In order to reduce potential response bias, it is suggested that self-reported scales consist of both positively and negatively worded items. This mixed format may have an undesirable effect, called the wording effect. In this study, we apply Bayesian statistics to compare two models for the mixed-format mathematics self-concept scale to investigate the impact of the wording effect on this scale. Based on reliability values, explained common variance, and ratios of factor loadings, we find that the bi-factor model is the preferred model and that the wording effect is substantial for the scale.

Authors

• Minsun Kim, University of Missouri - Columbia

• Ze Wang, University of Missouri