Heterogeneity in reporting behavior in surveys of Czech students in PISA 2012 : analysis according to region, type, and size of school
Abstract
Differences in reporting behavior among students can bias data (not only) in international surveys. Using PISA 2012 data, we analyze the differences in reporting behavior of Czech students with selected characteristics (gender, parents' education) at various schools (region, school type, and school size) and the relationship between their reporting behavior and test results.
We identify reporting behavior using acquiescence response style identification, item nonresponse analysis, overclaiming technique (OCT), and anchoring vignettes. There were notable differences in reporting behavior particularly between regions, school types, and genders, such as in the evaluations of the lowest level classroom management vignette.
Student test results are markedly related to their reporting behavior at both individual and school levels; for example, the correlations between reading literacy and OCT accuracy are 0.49 and 0.71, respectively. Future research may focus on different approaches, e.g., item response time analysis.