Concerns about fairness in educational testing has prompted the development of methodological tools to assess the possibility of bias in test scores associated to sociodemographic characteristics of examinees. These tools include differential validity and prediction. The first estimates the association between tests scores and a criterion (sucha as first year academic performance in the university), comparing such association among sociodemographic groups. The second technique compares the predictive validity of tests among groups, testing whether prediction errors are similar among groups. Using Chilean university entrance tests (Mathematics and Language PSU) and high school grades as predictors, and first year university grades as the criterion, differential validity and prediction was assessed, considering gender and type of school as grouping variables. Results show a similar pattern to US previous studies, indicating a consistent but small trend to underprediction of women’s performance.

Equity, evaluation, admission tests, sociodemographic characteristics, university education.