Learning to Program != "One-Size-Fits-All": Exploring Variations of Parsons Problems as Scaffolding

Abstract: Lowering the barriers to computer programming requires understanding how to scaffold learning. Parsons problems, which require learners to drag-and-drop blocks of code into the correct order and indentation, are proving to be beneficial for scaffolding learning how to write code from scratch. But little is known about the ability of other problem types to do so. This study explores learners' perceptions of a new programming environment called Codespec, which was developed to make computer programming more accessible and equitable by offering multiple means of engagement. Retrospective think-aloud interviews were conducted with nine programmers who were given the choice between Faded Parsons and Pseudocode Parsons problems as optional scaffolding toward solving write-code problems. The results showed that offering Faded and Pseudocode Parsons problems as optional scaffolds supported comprehension monitoring, strategy formation, and refinement of prior knowledge. Learners selectively used Faded Parsons problems for syntax/structure and Pseudocode Parsons problems for high-level reasoning. The costs noted included the time it takes to drag-and-drop the blocks and the confusion experienced when a solution diverges from a learners' mental model. Faded Parsons problems were also perceived as a desirable challenge. This study contributes to the field of computing education and human-computer interaction by extending the functionality of problem spaces that support Parsons problems and by providing empirical evidence of the effectiveness of using other problem types as scaffolding techniques.