The interplay between rich and big data in programming education research

To conduct solid research on how students learn programming, we need both ‘rich data’ and ‘big data’. In the past decades, researchers have been collecting both types of data, such as large datasets of programs written by students, containing numerous mistakes, but also more fine-grained data, such as verbalizations of what students were thinking when solving a challenging programming problem. While there is an interplay between these two types of data, they are typically used to answer different questions. There are also several existing datasets available for conducting programming education research, however, these are more often ‘big’ rather than ‘rich’, and it is not trivial to find and use them. In this talk I will show several examples of my research, in which I have (re)used datasets to study aspects of how students learn to program, discussing the need for collecting, analyzing and sharing big as well as rich data.