Inequalities in test scores between Indigenous and non-Indigenous youth in Canada

This paper documents a robust achievement gap between the math scores of Indigenous and white youth in Canada between 1996 and 2008. Using data from the restricted-access National Longitudinal Survey of Children and Youth we show that after controlling for a rich set of observables, students who self-identify as Indigenous perform 0.31 standard deviations lower on a standardized math test compared to their white counterparts. We find that this test gap emerges by the age of 12, and it did not decline between 1996 and 2008, despite the recommendations of the 1996 Royal Commission on Aboriginal Peoples to ameliorate the public education system for Indigenous students. Counterfactual estimates from the decomposition method of Lemieux (2002) suggest that the test gap among the lowest performing students would have been eliminated if Indigenous students faced the same level of and returns to observable characteristics as white students. This exercise does not result in a narrowing of the test gap in the upper tail, suggesting that unobservables, rather than observables, are driving the majority of the test gap among high achieving students.