University of Twente Student Theses

Login

Exploring pedagogical discourse data through LDA topic modelling, SSNA and thematic analysis

Steenbergen, Rik van (2021) Exploring pedagogical discourse data through LDA topic modelling, SSNA and thematic analysis.

Full text not available from this repository.

Full Text Status:Access to this publication is restricted
Abstract:Through modern technological development, the educational sector is no exception to the world wide increase and availability of large bodies of qualitative data. Specifically, text data from discourse is becoming increasingly easy to generate and store. However, the valorisation of larger bodies of qualitative data is fairly resource intensive. In this paper, we present a novel method for analysing large bodies of qualitative discourse data with the express purpose of adaptation and replication. We combined several quantitative and qualitative methods to relatively quickly and intuitively identify and analyse latent themes in a discourse dataset. In three iterative validation loops, we combined Latent-Dirichlet Allocation topic modelling, Semantic Social Network Analysis and thematic analysis with rounds of expert appraisal and a simple inter-rater procedure. We used model discourse from a European research project on video-supported collaborative learning (the video supported education alliance) in the presented method. Notably, we identified and analyzed four distinct topics, and the results of these analyses corresponded strongly to the theoretical framework of the project on video-supported collaborative learning. The results of the analysis are discussed on a methodological and pedagogical level. We conclude that the presented method to use thematic analysis on discourse units selected through SSNA on keywords generated by LDA resulted in distinctive themes connecting relevant topics in the studied dataset. We suggest this method or adaptations thereof could be of potential value to research and practice to quickly and intuitively identify and explore themes in larger qualitative datasets. To this end, we recommend methodological and statistical improvements to the presented novel procedure.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:54 computer science, 80 pedagogy, 81 education, teaching
Programme:Educational Science and Technology MSc (60023)
Link to this item:https://purl.utwente.nl/essays/85960
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page