Abstract

We use data from the international assessments PISA, TERCE, and ICCS to evaluate the invariance of student socioeconomic background scales among the countries participating in these studies. More specifically, we examine whether measures that were developed regionally exhibit better psychometric properties than other measures that were designed to work in a larger and more diverse number of education systems. First, we test the invariance of socioeconomic status (SES) scales across all the countries participating in each study, and then, we run the same test including only Latin American countries. The results suggest that none of the SES scales are invariant at the scalar level among all the countries participating in each study, and therefore comparisons between nations should be made with caution. As for the second group of analyses, our results show that when the analysis is restricted to Latin American countries, the invariance levels improve considerably. Finally, the levels of invariance achieved by each scale in each study and for each group of countries are discussed, as well as the type of comparisons that can be made given these results.