Impact of Hint Content on Performance and Learning: A Study with Primary School Children in a Scratch Course

The implementation of computational thinking concepts in primary school curricula usually includes programming activities. As primary school teachers often lack subject knowledge, they may struggle to help students during these programming activities. Additional support can be provided by automated program analysis, for example in terms of hints on conceptual knowledge related to bad coding patterns observed, or procedural hints on how to solve the task at hand. However, care has to be taken since these hints need to balance (1) helping students to perform a specific task successfully, while nevertheless (2) ensuring a learning effect beyond the specific task. To understand the effects of different types of hints we therefore conducted a study with 36 children aged 7–12 in 10 programming courses. After being introduced to basic programming structures in three units, the children were tasked to debug six Scratch programs using different types of hints, where we observed that procedural hints have the strongest impact on performance. In order to examine an impact on the transfer of learned knowledge, we observed the children’s difficulties during the successive fifth unit, in which they created their own projects. The results of the fifth unit show that having received a procedural hint on a specific pattern during the fourth unit leads to slightly fewer bad related code patterns but also to slightly fewer good code patterns. Considering these results together with the subjective perceptions of the children, we can derive insights into how to best support performance and learning using (automated) feedback.