Supplementary Research Corpus and Qualitative Audit Matrices: Student Reflexivity and Epistemic Accountability Mediated by Generative AI

Abstract

Research Data Repository and Methodological Documentation This repository contains the curated and de-identified qualitative dataset, documentary literature inventory, and methodological documentation supporting the manuscript entitled “Student Reflexivity and Epistemic Accountability in the Platform University: A Critical Realist Analysis of Academic Deliberation Mediated by Generative AI”, submitted for consideration to the Journal of Critical Realism (Routledge / Taylor & Francis). The repository is intended to support research transparency, methodological traceability, and the critical examination of the empirical and documentary evidence underpinning the study. Repository Contents 1. Qualitative_Audit_Matrix_52_Disclosures A systematically organised and de-identified qualitative audit matrix comprising 52 student disclosures collected from online higher education confession platforms and discussion communities. The dataset includes 38 records from Indonesian university confession (menfess) platforms and 14 records from international Reddit discussions. Personally identifiable information has been removed or obscured to minimise the risk of participant identification, with ethical considerations informed by the Association of Internet Researchers (AoIR) Internet Research: Ethical Guidelines 3.0. 2. Documentary_Literature_Inventory_107_Studies A structured inventory of 107 empirical and conceptual publications identified through five thematic research areas: cognitive offloading, metacognitive monitoring, oral examinations, anxiety surrounding false-positive AI detection, and algorithmic problem-solving. The inventory documents the scholarly evidence used to contextualise the empirical findings and inform the study's critical realist interpretation. 3. README_and_Codebook Methodological documentation comprising the dataset description, operational definitions, analytical codebook, and retroductive coding framework informed by the DREIC explanatory model. These materials document the analytical procedures and support the traceability of the study's interpretations, from empirical observations to proposed underlying generative mechanisms. Research Integrity and Data Use The materials are provided to facilitate scholarly scrutiny and methodological transparency. De-identification reduces, but does not entirely eliminate, the possibility of re-identification, particularly where publicly accessible online disclosures are involved. Reuse, quotation, or redistribution of the empirical records should therefore take account of contextual privacy, platform-specific conditions, and applicable research ethics requirement