Abstract

This study aims to investigate how different levels of student performance are distinctively characterized in a scenario-based democratic communication assessment, applying a series of text-mining analyses. Responses written by 543 middle school students were categorized into higher, middle, and lower levels according to their scores, and the performance characteristics of the groups were compared. The main findings are as follows. First, the keyword frequency analyses showed that words describing the problem situation in the scenario were frequent at the lower level, while words describing methods and procedures for solving the problem were frequent at the upper level. Second, as a result of keyword network analysis, words describing the situation were loosely connected at the lower level, while words describing the principles of problem solving showed higher centrality. Third, the topic modeling analysis showed that at the lower level, the presented problem situation emerged as a latent topic, while at the upper level, mutual respect and problem solving methods emerged as latent topics. Finally, we discuss the usefulness and future challenges of using text mining techniques to scrutinize qualitative differences in student performance in constructed response questions.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call