Abstract

Two formats, traditional Likert scale single statements and two-alternative forced choice statements, were administered to samples from the United States and China to examine whether format influences response biases. Examination of response anchor usages and decision making processes from the two samples using an item response tree model yielded conclusions consistent with findings of prior studies--traditional Likert scale single statements exhibited cross-cultural response style differences: Chinese respondents engage in significantly less endpoint endorsement. Such response style differences can produce measurement nonequivalence across cultures. In contrast, a two-alternative forced choice format with Item Response Theory scoring yielded latent trait score estimates substantially correlated with estimates from other measures, demonstrating convergent validity, but with much smaller differences in trait variance across cultural groups. Thus, the two-alternative forced choice format appears to be an effective format for reducing the impact of cross-cultural response style bias. Implications for practice and extensions for further research are discussed.