Response styles as a threat to cultural equivalence

Abstract

This study aimed to identify how a response style (RS) affects differential item functioning (DIF) detections in cross-cultural comparison studies. RS is an individual’s systematic response tendency to survey questions. Following the representative indicators response style means and covariance structure (RIRSMACS) model introduced by Weijters et al.(2008), Tasaki and Shin (2017) developed an RS measurement model and quantified three RS types (acquiescence, extreme, and mid-point) in three nations (Japan, Korea, U.S.). Using the same data set, this study adapts their RS model to control for RS-related nuisance variances and to see if different DIF items emerge with or without the RS model. As a result of DIF analysis through the MGMACS approach, three nonuniform and five uniform DIF items were flagged on a 10-item psychological scale. By applying the RS model to DIF analysis, all the nonuniform DIF items and three uniform DIF items turned out to be non-DIF items. These results imply that RS-related nuisance variances may be a cause of DIF, and RIRSMACS can be used to control for RS.