Disentangling Language-Related Item Biases in Stress Research With People of Turkish Origin in Germany: An Application of the CCT Procedure

Psychological stress experiences play a major role in the development of ethnic health disparities. Investigating such relationships often requires the use of questionnaires in different languages, however, this involves the risk of biased measurements. Such biases may be even more likely to occur the closer the construct being measured is to cultural experiences. We adapted the culture, comprehension, and translation bias (CCT) procedure (Bader et al., 2021) to test three language-related item biases in the measurement of stress experience among people of Turkish origin in Germany (i.e., language choice, comprehension, and translation bias) in the Perceived Stress Scale (PSS) and the Multidimensional Acculturative Stress Inventory (MASI). One thousand three hundred four adult persons of Turkish origin living in Germany participated in an online survey with a quasi-experimental design: participants either chose the German version, were assigned the German version, or were assigned the Turkish version of the study questionnaire. Participants completed the 10-item PSS, 25-item MASI, and other measures, and provided sociodemographic data. The adapted CCT procedure revealed none of the language-related biases for the PSS, whereas for the MASI a language choice bias could be demonstrated for one item, a comprehension bias for one item, and a translation bias for five items. The results suggest that language-related item biases are trait-specific and that the CCT procedure is suitable for testing other item biases beyond those tested by Bader et al. Testing such item biases may improve the study of stress experiences in the context of ethnic health disparities. Public Significance Statement Bader et al. (2021) proposed a procedure that uses multiple pairwise comparisons to detect measurement noninvariance in order to disentangle item biases in a cross-cultural context. We adapted their approach and identified language-related item biases in a measure of acculturative stress, but not in a scale that assesses general stress experiences. The measurement of constructs may be more susceptible to such biases the closer the construct is to culture-related experiences.