What the study analysed
The qualitative study, completed at the Faculty of Humanities and Social Sciences and published open access in the faculty repository, examined how medically transitioned transgender women interpret contemporary media representations of their community, and how those representations bear on identity, belonging, and psychological wellbeing. Eight participants aged 21 to 39, from Croatia, neighbouring Balkan countries, and the United States, were recruited through purposive, convenience, and snowball sampling.
Interviews were semi-structured, conducted online in March 2026, and ran roughly 45 minutes each. All were audio-recorded, transcribed verbatim, manually reviewed and corrected by the researcher, and stripped of identifying information before analysis. The study was approved by the University of Split’s Human Research Ethics Committee.
Manual coding first, AI as a consistency check
Coding was conducted manually and inductively through repeated close readings, following a general inductive approach and reflexive theme development. Only then was the initial thematic coding processed through Evidano to support triangulation, with the thesis mentor providing methodological oversight as an outsider to the community studied.
The analysis produced five themes and 14 sub-themes: Authenticity of Identity, Hypervisibility Incongruence, Representational Distancing, Psychological Impact of Media, and Desired Forms of Representation. The thesis went on to propose a new construct, hypervisibility-induced representational distancing.
How Evidano fit the workflow
A reflexive, researcher-led process, with AI used as a supplementary check.
- 1Interviews transcribed and anonymised
Transcripts were manually reviewed and corrected, and identifying information removed.
- 2Manual inductive coding
Codes were refined and organised into thematic categories through repeated close readings and comparison across accounts.
- 3Initial coding processed through Evidano
A supplementary pass to support analytic triangulation across the dataset.
- 4Outputs reviewed and verified by the authors
Final coding decisions, interpretations, and theme development remained researcher led.
“Following transcription, coding was conducted manually and inductively through repeated close readings of the interview material.”
“To support analytic triangulation, initial thematic coding was processed using AIlyze (2023), and all outputs were subsequently reviewed and verified by the authors.”
“All final coding decisions, interpretations, and theme development remained researcher led.”
“AIlyze was used solely as a supplementary analytic tool to enhance the consistency and thoroughness of the coding process rather than to generate or determine the final themes.”
A second pass over sensitive interview data
Solo coding of emotionally rich interviews leaves the researcher’s own positionality hard to test without an independent reading to compare against.
A supplementary analytic pass that tests the consistency and thoroughness of manual coding, without taking the interpretation out of the researcher’s hands.
The thesis discloses exactly what the tool was and was not used for.
Identifying information was removed before any analysis.
Used to check consistency, not to generate or determine final themes.
Approved by the University of Split’s Human Research Ethics Committee, with informed consent from every participant.