Abstract Currently student feedback is mainly evaluated with quantitative methods since qualitative analysis has been highly effort intensive. In this article, we present a process for tapping into the resource of responses to open-ended feedback questions by using a topic-modelling approach that goes beyond listing modelling outcomes. The objective of this study is to present a streamlined, yet rigorous, process for analysing large amounts of written feedback that connects qualitative findings to existing literature, theories and quantitative feedback. The topic models are created using the Latent Dirichlet Allocation (LDA) method, after which qualitative and quantitative evaluation methods are used to validate the topic outcomes. The proposed process can help educators analyse teaching quality on programme- or institution-wide level, or on single courses with a very large number of students. The process systematizes and combines existing processes, is repeatable, and can serve as a basis for richer analysis for educators. In student evaluation of teaching (SET) research, it advances the state of the art in applied topic modelling by demonstrating how to validate the topics via thematic analysis and by connecting them to theoretical frameworks and quantitative data. Previous topic modelling studies in this field follow mainly descriptive approaches. We demonstrate the process with feedback data collected from 6087 student evaluations of university courses and confirm that quantitative feedback variables can be used to validate qualitative feedback topic-modelling outcomes and thematic analysis provides a more in-depth explanation of the topics. We additionally find that the proposed topic modelling approach discovers new constructs of SET that cannot be distinguished from quantitative SET measures. The main limitation of the study is that the proposed process is novel and requires further evaluation to establish its full validity.

中文翻译：

---

利用书面学生反馈提高教学质量：简化流程

摘要 由于定性分析工作量很大，目前学生反馈主要采用定量方法进行评估。在本文中，我们展示了一个过程，通过使用超越列出建模结果的主题建模方法来挖掘对开放式反馈问题的响应资源。本研究的目的是提出一个简化但严格的过程，用于分析大量书面反馈，将定性发现与现有文献、理论和定量反馈联系起来。主题模型是使用潜在狄利克雷分配 (LDA) 方法创建的，然后使用定性和定量评估方法来验证主题结果。提议的过程可以帮助教育工作者在项目或机构层面分析教学质量，或有大量学生的单一课程。该过程将现有过程系统化并结合起来，具有可重复性，并可作为教育工作者进行更丰富分析的基础。在学生教学评价 (SET) 研究中，它通过展示如何通过主题分析验证主题并将它们与理论框架和定量数据联系起来，推进了应用主题建模的最新技术。该领域以前的主题建模研究主要遵循描述性方法。我们使用从大学课程的 6087 名学生评估中收集的反馈数据来演示该过程，并确认定量反馈变量可用于验证定性反馈主题建模结果，主题分析提供了对主题的更深入解释。我们还发现，所提出的主题建模方法发现了无法与定量 SET 度量区分的新 SET 构造。该研究的主要限制是所提出的过程是新颖的，需要进一步评估才能确定其完全有效性。