Reflexis: Supporting Reflexivity and Rigor in Collaborative Qualitative Analysis through Design for Deliberation

TL;DR

Reflexis addresses a key gap in qualitative analysis tooling by embedding reflexivity, analytic provenance, and productive disagreement into a shared workflow. The system introduces ReflexiveLens for in-situ reflection, automated code-evolution history with drift alerts, and positionality-aware collaboration to surface interpretive differences as productive dialogue. An evaluation with 12 researchers demonstrates increased granular reflexivity, transparency, and richer collaboration, while revealing design tensions around prompt timing and networked memos. The work articulates a design-for-deliberation framework for human-AI qualitative analysis and discusses epistemic risks of modeling positionality, offering guidance for future longitudinal deployments and configurable AI augmentation. Overall, Reflexis exemplifies how AI can scaffold rigorous, deliberative interpretation without supplanting human agency.

Abstract

Reflexive Thematic Analysis (RTA) is a critical method for generating deep interpretive insights. Yet its core tenets, including researcher reflexivity, tangible analytical evolution, and productive disagreement, are often poorly supported by software tools that prioritize speed and consensus over interpretive depth. To address this gap, we introduce Reflexis, a collaborative workspace that centers these practices. It supports reflexivity by integrating in-situ reflection prompts, makes code evolution transparent and tangible, and scaffolds collaborative interpretation by turning differences into productive, positionality-aware dialogue. Results from our paired-analyst study (N=12) indicate that Reflexis encouraged participants toward more granular reflection and reframed disagreements as productive conversations. The evaluation also surfaced key design tensions, including a desire for higher-level, networked memos and more user control over the timing of proactive alerts. Reflexis contributes a design framework for tools that prioritize rigor and transparency to support deep, collaborative interpretation in an age of automation.

Figures (8)

• Figure 1: Reflexis: Supporting Reflexive Analysis. (1) User can easily engage in a just-in-time guided reflection during analysis, the reflection process is designed to be lightweight and optional, reducing labor and encouraging reflexivity. All users can review reflexive notes directly under the coding workspace, or (2) engage in additional reflection. (3) Users can also access reflexive notes within specific code on the sidebar and can filter by unique user, as well as the ability to create structured summary.
• Figure 2: Reflexis: Making Code Evolution Transparent and Tangible. To make the analytical process visible and consistent, Reflexis first (1) actively engages users with codes as evolving concepts by monitoring for code drift. When a shift in code application is detected, it alerts researchers, prompting a deliberate re-evaluation of the ongoing analysis and encouraging reflective practices. Supporting this reflective process, the system provides a detailed record of the analysis: the (2) code-level analysis history records every analytic event, including code creation, application, merges, and splits. Finally, (3) the project-level analysis history visualizes analytical provenance, tracing how codes are transformed over time.
• Figure 3: Reflexis: Scaffolding Collaborative Interpretation. To ground collaboration in reflexive practice, researchers first complete profiles detailing their background and positionality. When interpretive differences arise during coding, (1)Reflexis surfaces them through a mediated discussion prompt that explicitly references the researchers' stated positionalities, helping to contextualize their unique lenses and scaffold constructive disagreement. (2) To help teams manage and prioritize these conversations, a Discussion Focus icon is displayed next to each code, providing an at-a-glance overview of agreement levels and a detailed breakdown after click-through. Furthermore, (3) researchers can toggle the view to show only data snippets with divergent codes, focusing the team's attention on areas requiring discussion.
• Figure 4: Reflexis: Change in self-reported frequency of in-situ reflexivity. The Sankey diagram visualizes how participants perceived their practice of pausing to write a rationale for a code choice, comparing their recalled frequency from their typical workflows ("Before") with their reported frequency during the study using Reflexis ("After"). This data is presented to illustrate a perceived change in practice, not as a direct measure of behavior.
• Figure 5: User Study Survey Question: What tool do you usually use for qualitative analysis?