This article presents empirical evidence on educational investments by members of different castes and religion using household-level, cross-sectional data from West Bengal. It finds that scheduled caste households invest significantly less than other households in private coaching of children, even after controlling for all available socio-economic background variables. This result is posited to arise from two possible sources: from cultural factors and from positive discriminative practices. The article develops an empirical strategy to determine which type of factor is more significant and finds that cultural factors are more likely than positive discriminative practices to be the source of the lower spending.
