Lessons Learned from Available Parsons Puzzles Software

Abstract

Since their original inception in 2006, Parsons puzzles have repeatedly been shown to be an effective tool for supporting the development of students’ programming skills. There have been so far very few software tools designed to help students practice with these puzzles. In this work, we propose to first review existing tools from the traditional perspective of the requirement to better serve students, but also by adopting a similar perspective including the needs of instructors and Computing Education researchers. We also introduce a new tool, EvoParsons, and show how it proposes to address some of the limitations and opportunities that were identified.

EvoParsons is a proof of concept implementation designed to interchangeably use both instructor-designed Parsons puzzles, but also automatically generated ones. Simply enabling such possibility already taught us the critical importance of providing an easily extensible Parsons puzzle implementation. Computing Education researchers, who are interested in applying artificial intelligence techniques, benefit greatly from open source access to Parsons puzzle software. However, we found these implementations to be even more useful if they rely on language-agnostic technologies such as REST web-based APIs.

These needs differ when Computing Education researchers are more interested in analyzing the students interactions with the puzzles. In such scenarios, simplifying the deployment of multiple server versions (aimed, for instance, at supporting different experiments or groups) is essential. We identified Docker containers as a way to address such needs efficiently. In addition, such researchers also need to have easy access to very detailed logs of the interactions that occurred between students and the puzzles. Relying on a well-established format facilitate further analysis of such logs by a wide variety of tools.

From the students’ perspective, our preliminary evaluation of EvoParsons also revealed the importance of identifying and mitigating aspects of the user interface that might increase the user-fatigue phenomenon and thus make the student-puzzle interactions less meaningful. Accessibility of the software, which for many students nowadays also includes tablets or smart phone, was also found to be an essential feature absent from existing tools.

Last but not least, from the perspective of instructors, making Parsons puzzles easily available to their students is the primary goal. Again, ease of deployment of server-side components plays a central role in this that can also be addressed by the use of Docker containers. However, it is just as essential for instructors to be able to easily modify the puzzles managed by the software. Parsons puzzle software should therefore facilitate the contribution of new puzzles, either via uploading files describing puzzles in a straightforward format, or by providing an instructor interface to the software allowing to manage such contributions. We explored the former possibility so far.

We conclude our comparison between EvoParsons and existing tools by reviewing preliminary student feedback on the use of the EvoParsons system.