AI-R04 — AI for Qualitative Research

Course Code: AI-R04  |  School: School of Artificial Intelligence  |  Cluster: Level 2 — Research & Academic Practice

Level: Intermediate  |  Duration: 6 weeks · 20–26 learning hours  |  Language: English  |  Certificate: Professional Certificate (non-degree)  |  Format: Self-paced with AI support under human supervision

Overview

Qualitative analysis is interpretive work. Its validity rests on the analyst’s sustained engagement with the material, and that is precisely what machine assistance threatens to remove. This course takes the threat seriously rather than dismissing it, and asks a narrower question: which parts of qualitative work can be assisted without hollowing out the interpretation?

The answer developed across the course is that machine assistance is defensible for transcription, for organising and retrieving coded material, for suggesting candidate codes that the analyst then accepts or rejects with reasons, and for surfacing negative cases the analyst may have overlooked. It is not defensible for producing themes, for deciding what a participant meant, or for writing the interpretation.

Learners work with their own data or a supplied anonymised corpus, produce a full codebook and thematic analysis, and maintain an audit trail showing where machine suggestions were accepted, rejected and why. Reflexivity is assessed, not merely encouraged.

Learning outcomes

On completion, a successful learner will be able to:

1. Prepare qualitative data for analysis with appropriate anonymisation, transcription conventions and consent compliance.
2. Develop a codebook through iterative coding, with definitions, inclusion criteria and worked examples for each code.
3. Use machine assistance for code suggestion and retrieval while retaining and documenting analytic authority.
4. Conduct thematic analysis that distinguishes semantic from latent content and reports negative cases.
5. Demonstrate trustworthiness through an audit trail, reflexive account and, where appropriate, participant checking.
6. State the limits of interpretation, including the influence of the analyst’s position on the reading produced.

Who this course is for

Qualitative researchers, doctoral candidates, anthropologists, sociologists, policy researchers conducting interview or document analysis, and evaluators working with open-ended data.

Prerequisites

Prior training or experience in qualitative methods. AI-F06 recommended. Learners must have either their own consented data or use the anonymised corpus supplied in-course; unconsented material may not be uploaded to any external tool.

Syllabus

Module 1 — Data preparation, consent and what may leave the room

Focus. Anonymisation and pseudonymisation, transcription conventions and what they discard, and the hard constraint that consent for research participation is not consent for third-party processing.

Lessons. 1.1 Anonymisation in practice. 1.2 Transcription conventions and analytic loss. 1.3 Consent scope and third-party processing. 1.4 Local versus vendor processing decisions.

Core reading. Matthew B. Miles, A. Michael Huberman & Johnny Saldaña, Qualitative Data Analysis: A Methods Sourcebook, 4th edition (Thousand Oaks: SAGE, 2020), chapter 2. Undang-Undang No. 27 Tahun 2022 tentang Pelindungan Data Pribadi (Indonesia), or the applicable instrument.

Deliverable. Data handling statement specifying what may and may not be processed externally, with reasons.

Module 2 — First-cycle coding

Focus. Descriptive, process and in-vivo coding. Building code definitions with inclusion criteria and examples, and the discipline of coding enough material by hand before any assistance is introduced.

Lessons. 2.1 First-cycle coding methods. 2.2 Writing a code definition. 2.3 Coding by hand: the required baseline. 2.4 Memoing as analysis.

Core reading. Johnny Saldaña, The Coding Manual for Qualitative Researchers, 4th edition (London: SAGE, 2021), chapters 1–3.

Deliverable. Hand-coded sample of at least twenty per cent of the corpus, with analytic memos.

Module 3 — Machine assistance under analytic authority

Focus. Using a model to propose codes on unseen material, comparing its proposals with the hand-developed codebook, and recording acceptance and rejection with reasons. The specific risk of code drift toward the model’s prior.

Lessons. 3.1 Prompting for code suggestion. 3.2 Comparing machine and hand codebooks. 3.3 Acceptance, rejection and the reasons register. 3.4 Code drift and how to detect it.

Core reading. Saldaña, The Coding Manual, chapter 4. Ziwei Ji et al., “Survey of Hallucination in Natural Language Generation”, ACM Computing Surveys 55, no. 12 (2023).

Deliverable. Comparison report with the reasons register for every accepted and rejected suggestion.

Module 4 — Second-cycle coding and theme development

Focus. Moving from codes to categories to themes. The distinction between a theme and a topic summary, semantic and latent readings, and why theme generation remains the analyst’s work.

Lessons. 4.1 Categories and pattern coding. 4.2 Theme versus topic summary. 4.3 Semantic and latent levels. 4.4 Why themes cannot be delegated.

Core reading. Virginia Braun & Victoria Clarke, Thematic Analysis: A Practical Guide (London: SAGE, 2022), chapters 4–6. Virginia Braun & Victoria Clarke, “Using Thematic Analysis in Psychology”, Qualitative Research in Psychology 3, no. 2 (2006): 77–101.

Deliverable. Theme structure with defining and boundary examples for each theme.

Module 5 — Negative cases and trustworthiness

Focus. Actively seeking disconfirming material, and the established criteria for trustworthiness: credibility, transferability, dependability and confirmability, with the audit trail that evidences them.

Lessons. 5.1 Negative case analysis. 5.2 Credibility and confirmability. 5.3 Building the audit trail. 5.4 Participant checking: when it helps and when it distorts.

Core reading. Yvonna S. Lincoln & Egon G. Guba, Naturalistic Inquiry (Beverly Hills: SAGE, 1985), chapter 11. Miles, Huberman & Saldaña, Qualitative Data Analysis, chapter 11.

Deliverable. Negative case analysis and audit trail.

Module 6 — Reflexivity and writing the interpretation

Focus. Positionality as analytic information rather than confession, representing participants’ words responsibly, and writing an interpretation that shows its own reasoning.

Lessons. 6.1 Positionality and its analytic use. 6.2 Quotation, context and responsibility. 6.3 Writing that exposes reasoning. 6.4 Disclosing machine assistance in qualitative work.

Core reading. Braun & Clarke, Thematic Analysis: A Practical Guide, chapters 8–9.

Deliverable. Final submission: codebook, thematic analysis, reflexive account, audit trail and AI-use disclosure.

Assessment

Pass mark 70 per cent. All assessed components must be attempted. Every mark in this course is issued by a human assessor; no assessment outcome is generated automatically.

Rubric criteria

Each assessed artefact is marked against four criteria at four levels (distinction, pass with merit, pass, fail).

1. Analytic depth: does the interpretation go beyond restating what participants said?
2. Retained authority: is it demonstrable that the analyst, not the model, produced the themes?
3. Traceability: can any claim in the analysis be followed back to coded data through the audit trail?
4. Reflexive honesty: does the account identify how the analyst’s position shaped the reading, specifically rather than ritually?

Reading list

Core. Johnny Saldaña, The Coding Manual for Qualitative Researchers, 4th edition (London: SAGE, 2021). Virginia Braun & Victoria Clarke, Thematic Analysis: A Practical Guide (London: SAGE, 2022).

Reference. Matthew B. Miles, A. Michael Huberman & Johnny Saldaña, Qualitative Data Analysis: A Methods Sourcebook, 4th edition (Thousand Oaks: SAGE, 2020). Yvonna S. Lincoln & Egon G. Guba, Naturalistic Inquiry (Beverly Hills: SAGE, 1985).

Peer-reviewed. Virginia Braun & Victoria Clarke, “Using Thematic Analysis in Psychology”, Qualitative Research in Psychology 3, no. 2 (2006). Ziwei Ji et al., “Survey of Hallucination in Natural Language Generation”, ACM Computing Surveys 55, no. 12 (2023).

All items are published works identifiable by author, title and publisher. Learners obtain them through an institutional library or the publisher. The Academy does not distribute copyrighted texts.

Academic integrity and use of AI

Generative tools may be used in producing assessed work under three conditions. Use must be disclosed in a short statement appended to each submission, naming the tool and the task it performed. Any factual or technical claim originating from a generative tool must be verified against a citable source before it enters assessed work, and the verification must be evidenced. The analytical judgement in each artefact must be the learner’s own and must be defensible in a short follow-up. Participant data must not be processed by any external service beyond the scope of the consent obtained. Breach of this condition ends participation in the course.

Instructor: pending owner confirmation. Pricing: pending owner approval. Reference list verified against publisher records; any later addition is marked for verification before publication.