Re-reviewing item parameter equivalence in TIMSS 2015 from a sociocognitive perspective

Periodical
Educational Assessment, Evaluation and Accountability
Volume
33
Year
2021
Page range
27-48
Relates to study/studies
TIMSS 2015

Re-reviewing item parameter equivalence in TIMSS 2015 from a sociocognitive perspective

Abstract

From a sociocognitive perspective, item parameters in a test represent regularities in examinees’ item responses. These regularities are originated from shared experiences among individuals in interacting with their environment. Theories explaining the relationship between culture and cognition also acknowledge these shared experiences as the source of human cognition. In this context, this study argues that if human cognition is a cultural phenomenon and not everywhere the same, then item parameters in cross-cultural surveys may inevitably fluctuate across culturally different populations. The investigation of item parameter equivalence in TIMSS 2015 supports this argument. The multidimensional scaling representation of similarity in the item parameters across countries in TIMSS 2015 shows that the item parameters are more similar within Arab, Western, East Asian and post-Soviet country clusters and are remarkably less similar between these clusters. Similar fluctuation structure across countries in discrimination and difficulty parameters in mathematics and science subjects point to the vital role of cultural differences in item parameter nonequivalence in cross-cultural surveys. The study concludes that it is very difficult for cross-cultural surveys to achieve the highest level of measurement invariance that guarantees meaningful scale score comparisons across countries.