AI for Qualitative Research A Hands-On Guide for Management Scholars.pdf
大小:(5.84 MB)

只需: RMB 9元  马上下载

2025最新资料！
170多页的大型资料包！
全部矢量文字，方便翻译学习！
AI for Qualitative Research A Hands-On Guide for Management Scholars
Traditionally, qualitative researchers have manually analyzed textual data, such as interviews, field notes, and archival documents, to uncover patterns and generate rich, context-sensitive understandings of the human experience (Dana & Dumez, 2015; Grodal et al., 2021). Today, we stand at the intersection of tradition and innovation. The advent of large language models (LLMs), such as OpenAI GPT models and Google Gemini models, heralds a transformative era for qualitative research by offering new tools that can process vast amounts of data with remarkable speed and efficiency. This  reource is designed as a practical guide for researchers eagerly integrating LLM-based algorithms into their qualitative work. We aim to equip you with the basic technical know-how and critical insights necessary to harness these models effectively while remaining mindful of their limitations. As you will discover throughout the  reource, there are practical advantages when implementing LLMs through programming applications. Coding facilitates data handling, increases control over LLM parameters, helps manage LLM limitations, and allows the combination of multiple models and techniques, improving data processing and analysis. We advocate for a foundational understanding of LLM technology, empowering researchers to become educated users able to leverage AI capabilities effectively and ethically. As such, this  reource provides explanations of the underlying technology, including LLMs’ limitations, as well as considerations when employing them in research projects. To facilitate hands-on learning and practical application of LLMs, this  reource provides a custom-designed dataset and sample code in a dedicated GitHub repository named Diana-GQ/ai_for_qualitative_analysis (https://github.com/Diana-GQ/ai_for_qualitative_analysis). GitHub is a web-based platform for sharing code and collaborating on development projects. For ethical considerations and data protection, we created a dataset of 1,115 synthetic social media posts representing ten imaginary personas. This dataset is exclusively for educational purposes, allowing you to practice and explore LLM applications in qualitative analysis without the risks associated with real-world private or proprietary data. The dataset can be downloaded and used as input to the code examples provided in the repository. While the data and code are freely available for educational use, they are protected by the same copyright agreement as the rest of the  reource and are not intended for commercial use. Although the  reource is written for noncoders, it requires a learning effort to acquire basic programming skills. For true beginners, we provide links to external resources where you can complement the knowledge we provide in this  reource. We recognize that implementing coding for the first time represents a major challenge. However, we believe that this is a worthy investment that enables you to take full advantage of the constantly expanding capabilities of LLMs. While the  reource’s intent is not to teach coding, we believe that this skill is critical for researchers to realize the full potential of emerging AI technologies. We encourage you to use all the resources provided with this  reource and experiment with the code scripts provided. To facilitate its use as a hands-on guide, the  reource is divided into two parts. The first part provides foundational knowledge of LLM technology, along with its application for management research and the associated ethical considerations. The second part provides concrete examples of how to use LLMs in qualitative analysis. Part 1 includes Chapters 2 to 4. It covers the basic knowledge needed when leveraging LLMs for qualitative analysis in management research. Chapter 2 provides a brief history of the origins of LLMs, their technology, and their limitations. Chapter 3 provides a literature review on the use of natural language processing (NLP) and LLMs in management research. Chapter 4 touches on the main ethical considerations when deploying research projects using LLMs as analytical tools. Part 2 includes Chapters 5 to 11. It involves coding examples and explanations to leverage LLMs for qualitative analysis. Part 2 explains and expands upon the method developed by Garcia Quevedo, Glaser, and Verzat in “Enhancing Theorization Using Artificial Intelligence: Leveraging Large Language Models for Qualitative Analysis of Online Data” (2025). Chapter 5 aims to familiarize noncoders with basic coding functions, enabling them to understand the coding logic, and introduces the systems and tools for using NLP and LLMs. Chapter 6 delves into the use of LLMs in qualitative analysis, following the method proposed by Garcia Quevedo et al. (2025). This method enhances qualitative inductive analysis by leveraging LLMs to efficiently select the most relevant data and gain deep insights from large datasets, preparing the groundwork for manual inductive analysis. It combines different NLP tasks via traditional and LLM-based models. Chapters 7 to 10 explain each of these tasks, providing the code and recommendations to implement them. Chapter 7 introduces basic data exploration techniques. Chapter 8 expands on the use of LLMs for classification. Chapter 9 delves into the use of LLMs for topic modeling, a specialized form of clustering. Chapter 10 explains information retrieval and retrieval-augmented generation (RAG) as a useful generative technique for retrieving and synthesizing information. The last chapter, Chapter 11, explores the expanded possibilities of LLMs, which are constantly evolving

1 Introduction 1
References 4
Part I
2 Overview of Artificial Intelligence, Machine Learning,
Natural Language Processing, and Large Language
Models 7
Natural Language Processing 8
A Brief History of AI and NLP 9
Transformer Architecture and LLMs 11
Embeddings and Text Generation 13
Open-Source, Open-Weights, and Proprietary LLMs 15
Limitations of LLMs 16
References 18
3 Natural Language Processing in Management Research 23
NLP in Qualitative Research Today 24
LLM Applications in Qualitative Research 27
Text Exploration and Summarization 27
Data Retrieval and Augmentation 28
Classification and Clustering Analysis 29
References 31
4 Ethical Considerations 35
Ethical Considerations Related to LLMs 37
Bias and Discrimination 37
Data privacy and Security 38
Explainability, Transparency, and Accountability 39
Ethical Considerations When Using LLMs in Qualitative
Research 40
Toward an Informed and Responsible Use of AI 42
References 43
Part II
5 Systems and Tools to Use NLP and LLMs: Getting
Started 49
Learning the Basics 50
Python Programming Language 50
Pandas Library 51
Integrated Development Environments 52
Coding Assistants and Agents 52
Jupyter Note reources 53
Application Programming Interfaces (APIs) 54
Getting Hands-On 54
Setting Up Your Development Environment 54
Accessing APIs for a Variety of Projects 55
How to Use the HuggingFace API 56
How to Use the OpenAI API 61
How to Use the Groq API 66
How to Use the Google Gemini API 68
Reference 73
6 Using LLMs in Qualitative Analysis 75
A Method for Relevant Data Selection 76
Implementing LLMs in Qualitative Analysis 77
References 78
7 Data Evaluation and Validation 81
Python Code for Exploration and Validation 83
Data Exploration and Validation via LLM-Based
Approaches 92
8 Classification 103
Understanding Classification with LLMs 105
Creating a Classification for a Research Project 107
Fine-Tuned LLMs for Common Classifications 108
Zero-Shot, One-Shot, and Few-Shot Classifications 114
One-Shot Classification 119
Few-Shot Classification 121
Considerations When Using LLMs as Classifiers 123
References 125
9 Clustering and Topic Modeling 127
Understanding Topic Modeling with LLMs 128
Leveraging Topic Modeling for Data Analysis 128
Considerations of Topic Modeling for Qualitative Analysis 141
References 144
10 Information Retrieval and Retrieval-Augmented
Generation 147
Understanding LLM-Based IR and RAG 148
Using IR and RAG 150
Considerations of IR and RAG for Qualitative Analysis 161
References 163
11 Perspectives on LLMs in Management and Qualitative
Research 165
The Advent of Multimodal Analysis 166
The Power of Reasoning Models 166
Automation with Agentic AI 167
Synthesizing Knowledge with Deep Research 167
Final Words 168
References 168
Index 171