LIVE Research Projects
Human–AI Partnering in Qualitative Data Analysis
A project that explores how generative AI can be incorporated into qualitative data analysis
Project at a Glance
This project explores how generative AI can be incorporated into qualitative data analysis to augment, rather than automate, human analysis. Across both structured analytic coding and reflexive interpretive inquiry, the work examines how human–AI partnering can support deeper engagement with data, surface overlooked patterns / alternative interpretations and use AI uncertainty as a diagnostic signal. The goal is to expand researchers’ capacity for rigorous analysis while keeping human judgment at the center.
Project Description
Qualitative analysis involves closely examining large amounts of rich data to generate insight. Different modes of qualitative analysis place different demands on researchers, from applying defined categories consistently to developing situated interpretations through iterative engagement with data. These processes are often time-consuming and difficult to sustain across large and complex datasets, making qualitative analysis a productive setting for exploring how researchers can develop ways to partner with generative AI without replacing human judgment.
The project investigates two distinct modes of qualitative analysis. In structured analytic coding, categories and coding criteria are explicitly defined, so delineating their boundaries and applying categories consistently are critical. Here, the work uses human and AI disagreement as well as AI uncertainty and decision instability as diagnostic signals which can reveal ambiguous construct boundaries, underspecified coding criteria, or competing interpretations. This directs expert attention to cases where human judgment is most needed and supporting iterative refinement of codebooks and coding decisions.
In reflexive interpretive inquiry, meaning is emergent and situated, and rigorous analysis values context, positionality, participant meaning, and iterative cycles of engagement with data. Here, AI can help researchers revisit more of the corpus across more iterations, keep more context available, surface multiple candidate interpretations, search for confirming and disconfirming cases, and notice voices, patterns, or tensions they might otherwise miss. AI models never deliver fully-formed findings, but help expand the researcher’s field of noticing while researchers retain interpretive authority.
The goal is not to use AI to simply analyze faster, but to help researchers analyze more deeply and systematically
Human–AI Partnering in Qualitative Data Analysis' Impact
Much of the early conversation about generative AI in qualitative research focused on efficiency: using AI to summarize, code, cluster, or generate themes more quickly. This work reframes the opportunity around better, rather than just faster, analysis.
For structured coding, the opportunity lies in treating disagreement and uncertainty as information rather than simply error. Instead of treating human labels as unquestioned ground truth and measuring success only through agreement with this fallible standard, researchers can examine where and why AI coders disagree, hesitate, or produce unstable decisions. These moments can expose ambiguities in code definitions and tacit assumptions in human coding, focusing expert attention where conceptual judgment is most valuable.
For reflexive inquiry, AI offers a different opportunity: extending researchers’ capacity for sustained and iterative engagement with large and complex datasets. It can make it possible to engage more systematically with full corpora at scale: revisiting more of the data across multiple iterations, examining patterns through multiple analytic lenses, searching extensively for confirming and disconfirming evidence, and surfacing cases or interpretations researchers might otherwise overlook. Used critically and transparently, these capabilities can support credibility, dependability, confirmability, authenticity, and other established dimensions of trustworthy qualitative research
The Team Behind the Work
Alyssa Wise | Vanderbilt University
Fanjie Li | Vanderbilt University
Melissa Gresalfi | Vanderbilt University
Jesse Spencer-Smith | Vanderbilt University
Get Involved
Interested in working with this project or other LIVE research projects? Email us at: LIVE@Vanderbilt.edu