#108: Do Students Recognize Personalization? Comparing Personalized vs. Common-Solution Parsons Code Puzzles as Problem-Solving Scaffolding

Parsons code puzzles, which require students to arrange mixed-up code blocks into a correct program, can scaffold students who struggle while programming. Large language models can generate personalized Parsons puzzles based on students' existing incorrect code. However, the effects of such personalization remain underexplored, especially compared to the standard use of common-solution puzzles. We examined it with 85 students through a programming classroom study: students had access to personalized (PP) or only common-solution (CP) Parsons puzzles as scaffolding. We found that students who received a personalized puzzle rated the solution’s alignment with their own approach as 16% higher than with a common-solution puzzle. We also found preliminary evidence of a mediated pathway through which personalized puzzles improved scaffolding puzzle solution understanding via higher perceived solution alignment. This work offers insights for approachable AI-powered personalization during problem-solving.