Scaffolding Expertise: Evaluating Scaffolds for Block-Based Coding Among Experts and Novices
Yifan Zhang, Teomara A. Rutherford · 2024
Coding for computer programming is a common way to support and assess computational thinking (CT) skills, a set of problem-solving skills essential for computer science (CS) and STEM more broadly. We designed a coding platform, Fox and Field, to determine what kinds of scaffolds can encourage novice coders to behave more like experts. We compared actions between 221 upper-division undergraduate CS/engineering (expert, n=106) and social science (novice, n=115) majors randomized to scaffolding conditions from four universities in the United States. Overall, experts used statistically significantly more practices aligned with CT skills, such as those that increased code efficiency (e.g., non-right angle turns, loops). This difference disappeared when novices were scaffolded by being told to use fewer codes or by priming with critical features. Both experts and novices were equally likely to be swayed by purposefully distracting features within the platform, such as a drawn path to deflect them from the most efficient solution. Results present initial evidence regarding which features of coding platforms can direct students to exercise CT-linked practices, leading to recommendations regarding platform development to better support learning.