Developing coding schemes to analyze
problem-solving using NVivo

OVERVIEW

I worked with a Master’s student as an Undergraduate Research Assistant to run an experiment to test if AI-generated test questions were more engaging for engineering students than regular textbook questions. Each participant was allowed to pick between three engineering problems out of four categories: bending, extrusion, machining, and forging. Each student answered a problem generated by AI, a problem adapted directly from the textbook, and a problem with a combination of the two. After the experiment, the participants explained which questions they preferred and why.

Following the experiment, the lead researcher and I worked together to create coding schemes to analyze the students’ test results and recorded answers to the post-experiment interview questions. We separately coded the tests and interview transcripts and compared our results.