When Qualitative Research Meets Large Language Model: Exploring the Potential of QualiGPT as a Tool for Qualitative Coding

定性研究以深入探索复杂现象而闻名，往往需要耗费大量时间进行分析，尤其是在编码阶段。现有的定性评估软件通常缺乏自动编码功能、用户友好性和成本效益。大型语言模型（LLMs）如GPT-3及其后继者的出现标志着增强定性分析的转型时代。本文介绍了QualiGPT，这是一个旨在解决使用ChatGPT进行定性分析所面临挑战的工具。通过对传统手动编码和QualiGPT在模拟和真实数据集上的表现进行比较分析，结合归纳和演绎编码方法，我们证明了QualiGPT显著提高了定性分析过程。我们的研究结果表明，QualiGPT提高了定性编码的效率、透明度和可访问性。该工具的性能是通过使用互评可靠性（IRR）度量进行评估的，在各种编码场景中，结果表明人类编码者和QualiGPT之间存在实质性的一致性。此外，我们还讨论了将人工智能整合到定性研究工作流程中的影响，并概述了增强人工智能与人类协作的未来方向。

Qualitative research often involves time-intensive manual coding, lacking automatic coding capabilities, user-friendliness, and cost-effectiveness. This paper aims to introduce QualiGPT, a tool developed to address these challenges and enhance qualitative analysis.

QualiGPT utilizes Large Language Models like GPT-3 to improve efficiency, transparency, and accessibility in qualitative coding. It significantly enhances the qualitative analysis process by providing automatic coding capabilities and user-friendly features.

QualiGPT's performance was evaluated using inter-rater reliability measures, indicating substantial agreement between human coders and QualiGPT in various coding scenarios. The tool was tested on both simulated and real datasets, incorporating inductive and deductive coding approaches. The paper also discusses the implications of integrating AI into qualitative research workflows and outlines future directions for enhancing human-AI collaboration in this field.

Recent related studies include 'Using Machine Learning to Facilitate Qualitative Data Analysis: A Systematic Review' and 'Artificial Intelligence and Qualitative Research: A Scoping Review'.