Can distributed practice improve students' efficacy in learning their first programming language?

Learning how to program has become required for many majors in higher education. However, programming is not easily learned, especially for non-engineering students. To improve students' learning efficiency, we applied distributed practice to a C programming class with 69 college students in first grade, but have students decide the space of practice by themselves. By mining the relationships between practice patterns and learning performance, we found that students who practiced with high frequency significantly outperformed those who practiced infrequently. The frequency of practice was a strong predictor of both homework grades (p=0.001) and midterm exam grades (p=0.023). By contrast, the total amount of practice had very little effect on learning performance. The result shows that distributed practice is a better learning strategy than massed practice in C programming language learning. But the optimized space of practice in this domain has not been completely revealed yet.