Does test preparation mediate the effect of parents' level of educational attainment on medical school admission test performance?

The finding that admission test performance correlates with parents' level of educational attainment raised concerns regarding their fair and valid use. Critics argued that this relationship may partly reflect socio-economic differences in test preparation. Using data from three cohorts of medical school applicants we directly tested the postulated mediation hypothesis. Latent class analysis was used to identify four classes of test-takers differing in their use of various test preparation methods. Mediation analyses revealed that although latent test preparation classes differed in admission test performance, test preparation was virtually unrelated to parents' level of educational attainment. This disconfirms the mediation hypothesis. The results were further corroborated by measurement invariance analyses indicating that although test preparation-based score gains were confined to the specific narrower traits, SES-related differences in subtest performance were fully explained by SES-related differences in higher-order traits. This pattern of finding is inconsistent with the hypothesis that SES-related differences in admission test performance largely reflect SES-related differences in the access to test preparation. However, it is consistent with theoretical accounts that attribute SES-related differences in admission test performance to processes that operate from infancy to adulthood, which eventually lead to actual knowledge and intelligence differences at the time-point of admission testing.