Distinguishing the Themes Emerging from Masses of Open Student Feedback

EasyChair Preprint 5669

Abstract

Student feedback is one of the key methods for assessing the quality of teaching in higher education. Feedback is often collected using both Likert-type scales and open-ended questions. However, open-ended text answers are a difficult resource to utilize because of the manual work involved in qualitative analysis, and it is a challenge to gain insight of the underlying themes or issues behind the feedback. This paper presents a study in which we create and analyze topic models from open-ended student feedback. First,  6087 individual student evaluations were collected from university courses between two academic years, from 2016 to 2018. Then, topic models from the feedback texts were created using the Latent Dirichlet Allocation method with the R programming language and environment for statistical computing. After analyzing the resulting topic models, six categories of feedback were distinguished: 1) Positive comments about arrangements, 2) dissatisfaction in the teaching, 3) comments about course arrangements and deadlines, 4) lack of student motivation, 5) interest in the topic and understanding the material, and 6) comments about interesting, rewarding but challenging courses. Finally, this paper discusses the topic modelling results to provide an insight into the automatic analysis of student feedback.

Keyphrases: student evaluations of teaching, student feedback, text mining

@booklet{EasyChair:5669,
  author    = {Timo Hynninen and Antti Knutas and Maija Hujala and Heli Arminen},
  title     = {Distinguishing the Themes Emerging from Masses of Open Student Feedback},
  howpublished = {EasyChair Preprint 5669},
  year      = {EasyChair, 2021}}