Assessing heterogeneity in students’ performance. The case of the Massive Open Online Courses

Exploiting the conceptualization of MOCC performance and one of its main drivers –engagement– this paper aims to evaluate if and how much the role played by engagement in the prediction of student’s performance changes accord- ing to the students’ characteristics –gender, age, and country– or considering dif- ferent types of courses. To that end we explored the suitability of two tests, the parametric and permutation test, to assess whether the presence of heterogeneity in the sample has different effects on different parts of the conditional distribution of the students’ performance. Our results indicate that engagement varies according to the conditional quantiles of the performance and the specific segment defined by the categorical variables for which is estimated.