Leveraging Large Language Models for Automating Inductive Qualitative Coding: A Comparative Study of Prompt Engineering Techniques
Bibliographic record
Abstract
This study explores the potential of Large Language Models (LLMs), like those from the GPT family, to automate inductive qualitative coding—a process of assigning codes to text excerpts and organizing them into categories, traditionally done manually in social science and software engineering research. Our primary question is: Can LLMs effectively automate inductive qualitative coding? To investigate this, we compared different prompt engineering techniques, including Zero-shot, Few-shot, and Chain-of-Thought (CoT) learning, in coding interview transcripts. While LLMs cannot fully replace human coders, they can aid the process with a human-in-the-loop approach. Few-shot learning showed consistent performance with moderate amounts data, while CoT proved most effective in reducing partial hallucinations. Initially aimed at full automation, our study pivoted to testing prompt strategies after realizing that a human-in-the-loop process would offer better accuracy and flexibility, given the challenges of context and token limits in LLMs. These findings suggest that tailored LLM with adequate prompting techniques can help assist researchers when performing qualitative analysis.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.088 | 0.287 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.004 | 0.008 |
| Open science | 0.003 | 0.007 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".