This study aims at understanding different students approaches for solving assignments in MOOCs. It makes use of a large dataset of logs from students interaction with the MOOC platform Coursera on a course about functional programming with Scala. In total more than 10.000 students participated in the assignments. Learning approaches are divided in two categories: starting with video lectures (V) and start- ing with the assignment (A); and students are divided in three groups: those applying purely the approach V , those applying purely the approach A and mixed-approach student who can apply both approaches. We explore how our grouping correlates with assignment grades, number of submissions, time between submissions and overall performance. Significant difference has been found only on overall performance, while all three groups appear very similar on the other measures. Then we search correlations with approach changes for mixed-approach students. We observed that students are more likely to stay with the same approach, found significant difference on the starting time of learning activity sequences, but not on the time of student’s first assignment submission. We found no correlation between the approach choice and the grade or number of submissions on the previous assignment.