CSE Dissertation Defense: Kyle Harms

Apr 4
3:30 p.m.
5:30 p.m.
Jolley Hall, Room 309

"​Code Puzzle Completion Problems in Support of Learning Programming Independently"

Kyle Harms
Adviser: Caitlin Kelleher

Middle school children often lack access to formal educational opportunities to learn computer programming. One way to help these children may be to provide tools that enable them to learn programming on their own independently. However, in order for these tools to be effective they must help learners acquire programming knowledge and also be motivating in independent contexts. I explore the design space of using motivating code puzzles with a method known to support independent learning: completion problems. Through this exploration, I developed code puzzle completion problems and an introductory curriculum introducing novice programmers to basic programming constructs. Through several evaluations, I demonstrate that code puzzle completion problems can motivate learners to acquire new programming knowledge independently. Specifically, I found that code puzzle completion problems are more effective and efficient for learning programming constructs independently compared to tutorials. Puzzle users performed 33% better on transfer tasks compared to tutorial users, while taking 21% less time to complete the learning materials. Additionally, I present evidence that children are motivated to choose to use the code puzzles because they find the experience enjoyable, challenging, and valuable towards developing their programming skills. Given the choice between using tutorials and puzzles, only 10% of participants opted to use more tutorials than puzzles. Further, 80% of participants also stated a preference towards puzzles because they simply enjoyed the puzzle experience more than the tutorials. The results suggest that code puzzle completion problems are a promising approach for supporting independent learning of programming.