A feasibility study of an artificial intelligence based decision support system for personalised housing adaptations and assistive technology.

Qualitative research is essential for studying institutions whose work cannot be reduced to indicators, interfaces, or formal rules. Case study and qualitative inquiry are especially useful for reconstructing how institutional actors interpret mandates, how records move across organizational units, how errors are corrected, and how responsibility is assigned when decisions become contested (; ; ). Yet qualitative studies of complex institutions face a recurring problem: the route from empirical material to explanatory model is often difficult to inspect. Readers may see rich description, thematic categories, or a final diagram, but not the analytical steps through which interviews, documents, observations, and public records become mechanism-based claims (; ; ). The issue is therefore not simply how to collect qualitative evidence, but how to make the movement from evidence to coding, from coding to comparison, and from comparison to model-building transparent enough for review, adaptation, and cumulative research (; ; ). Existing qualitative studies of complex institutions can provide rich descriptions, thematic categories, and final models, but the analytical route from evidence to explanation often remains difficult to inspect (; ; ). The problem is not a lack of qualitative traditions; grounded theory, discourse analysis, institutional ethnography, narrative inquiry, content analysis, and interpretive case analysis already offer established approaches (; ; , ; ; ). The unresolved problem is how researchers can transparently reconstruct institutional mechanisms from interviews, documents, observations, and public records when digital systems operate under legal, organizational, political, and security constraints. Existing evaluations commonly emphasize technical audits, adoption, maturity, usability, service performance, or public trust, while the institutional work of validating records, authorizing corrections, assigning responsibility, documenting decisions, explaining outputs, and learning from failures remains insufficiently traced (; ; ; ). This gap is especially consequential in electoral administration, where digital systems affect voting rights, political competition, logistics, and the credibility of results. Without an inspectable route from corpus construction to coding, comparison, rival-explanation assessment, evidence-strength judgment, and model refinement, qualitative models risk appearing as interpretive assertions rather than reviewable explanations. The purpose of this study is to develop and demonstrate a critical realist qualitative methods protocol for transparent, reviewable, and ethically safe institutional mechanism tracing in high-stakes digital governance. The study is deliberately bounded: it does not offer a comprehensive framework for all qualitative data analysis traditions, nor does it conduct a cybersecurity audit, software evaluation, or population-level assessment of trust. Instead, it treats an electoral management body as the main case and four digital arenas—voter data, party and candidate verification, logistics, and results publication—as embedded units through which institutional work can be reconstructed. The protocol specifies how researchers can build an evidentiary corpus, conduct process-oriented interviews and document analysis, apply abductive and mechanism-oriented coding, compare arenas through matrices, assess rival explanations and evidence strength, preserve audit trails, validate emerging claims, and use tiered transparency to protect sensitive information. Electoral administration is used as a demanding methodological setting because its digital outputs are produced under compressed timelines, legal constraint, partisan scrutiny, and strong expectations of accuracy, auditability, accountability, and public explanation (; ; ; ).