The international surveys of pupil achievement - PISA, TIMSS, and PIRLS - have been widely used to compare socioeconomic gradients in children's cognitive abilities across countries. Socioeconomic status is typically measured drawing on children's reports of family or home characteristics rather than information provided by their parents. There is a well established literature based on other survey sources on the measurement error that may result from child reports. But there has been very little work on the implications for the estimation of socioeconomic gradients in test scores in the international surveys, and especially their variation across countries. We investigate this issue drawing on data from PISA and PIRLS, focusing on three socioeconomic indicators for which both child and parental reports are present for some countries: father's occupation, parental education, and the number of books in the family home. Our results suggest that children's reports of their father's occupation provide a reliable basis on which to base comparisons across countries in socioeconomic gradients in reading test scores. The same is not true, however, for children's reports of the number of books in the home - a measure commonly used - while results for parental education are rather mixed.
educational inequality socioeconomic status measurement error international comparisons PISA PIRLS