Mixed-worded scales and acquiescence in educational large-scale assessments

Self-report data such as those regularly administered with questionnaires in the OECD’s educational large-scale assessments are subject to response biases such as acquiescence, i.e., the tendency to agree with questionnaire items regardless their content. Research has shown that acquiescence affects the psychometric quality of such data, posing a threat to validity. Using a simple index that can be computed in the presence of both positively and negatively keyed items, the author examined the prevalence, the individual-level correlates, the impact on associations between indicators, as well as the county-level consistency of acquiescence for 16 questionnaires administered in four study programmes (PISA, TALIS, SSES, and IELS). Findings suggest that variation in acquiescence exists both between and within countries, the latter of which is determined by factors largely in line with prior research. Impact on associations as well as high levels of country-level consistency are evident. Based on these findings, recommendations for the construction of questionnaires to be administered in future assessments are derived.