Supplementary Material for Artificial Intelligence in Web Accessibility: Towards a Theory of AI-Assisted Remediation for Visual Disabilities

This material supports the transparency, traceability, and reproducibility of the research reported in the article: "Artificial Intelligence in Web Accessibility: Towards a Theory of AI-Assisted Remediation for Visual Disabilities".

The materials included in this repository correspond to the empirical and analytical procedures conducted across Phase 1 (Grounded Theory) and Phase 3 (Thematic Synthesis) of the study. The dataset provides access to qualitative data, coding schemas, analysis outputs, and supporting visualizations generated using ATLAS.ti and complementary tools.

The repository is organized as follows:

1. Phase 1 - Atlas.ti
This folder contains all qualitative analysis artifacts derived from the Grounded Theory (GT) process conducted in Phase 1. It includes:
- The ATLAS.ti project file (.atlasti) containing coded data and category structures.
- The complete codebook (PDF) describing codes and categories.
- Code lists (XLSX) used during initial and focused coding.
- Traceability reports linking codes to documents.
- Visual representations of the analysis, including code-document diagrams.
- Screenshots of the ATLAS.ti environment to support interpretability and transparency of the coding process.

2. Phase 1 - Observations from participants for GT
This folder contains the primary qualitative data used in the Grounded Theory analysis. It includes:
- Participant observation reports (P1–P5) documenting interaction experiences of blind users.
- Barrier Walkthrough (BW) evaluation results.
- Automated accessibility evaluation outputs (WAVE reports).
These materials represent the raw empirical evidence from which initial codes and categories were derived.

3. Phase 3 - Atlas.ti
This folder contains the artifacts produced during the thematic synthesis of empirical studies on LLM-based accessibility remediation. It includes:
- Thematic coding outputs and code lists.
- Codebooks documenting themes and higher-order categories.
- Visualizations such as thematic maps, code co-occurrence diagrams, and code-document relationships.
- ATLAS.ti screenshots documenting the analytical process.
- Reports linking documents, codes, and thematic groupings.
These materials support the refinement of the conceptual theory through elaborative coding.

4. Theory_AI_Web_Accessibility.xlsx
This file provides a structured summary of the conceptual theory, including relationships between concepts, themes, and categories identified across the different phases. It serves as a synthetic representation of the theoretical model derived from the study.

Together, these materials enable full reproducibility of the qualitative analysis process, from raw data to conceptual abstraction. The dataset provides a transparent account of how empirical evidence was transformed into codes, themes, and theoretical constructs, supporting future research in AI-assisted web accessibility and theory building in software engineering.