Perception bias in assessing rodent welfare during euthanasia and anaesthesia

Perception bias in assessing rodent welfare during euthanasia and anaesthesia This repository contains anonymized raw and processed data, preprocessing scripts, and statistical analysis code for an experimental eye-tracking and survey study investigating perception bias during the evaluation of rodent welfare in anaesthesia and euthanasia contexts. Study overview Participants experienced in rodent experimentation observed short videos depicting mice undergoing either: direct CO₂ euthanasia, or isoflurane anaesthesia followed by CO₂ euthanasia. Videos were presented with either correct or incorrect contextual framing labels (“Anaesthesia” or “Euthanasia”) to investigate whether contextual framing influences perceived distress of rodents ratings, visual attention, and welfare perception. During video observation, participants: wore eye-tracking glasses, completed concurrent and retrospective think-aloud procedures, and rated perceived rodent distress on ordinal Likert scales after each video. Raw eye-tracking data Raw export contents The raw eye-tracking exports include: continuous pupil diameter recordings (right eye, mm), gaze coordinates, fixation and saccade information, trial timing variables, participant pseudonyms, video and trial identifiers, AOI-related variables, and tracking quality information. The files are original export files generated from BeGaze eye-tracking software. Anonymization All participant identifiers were pseudonymized prior to release. No directly identifying information is included in the shared exports. Purpose of raw data sharing The raw eye-tracking exports are shared to support: reproducibility of preprocessing procedures, replication of pupil preprocessing and baseline correction, inspection of filtering and aggregation procedures, and alternative preprocessing or modelling approaches. Relation to processed datasets The processed long-format pupil datasets included elsewhere in this repository were derived from these raw exports using the provided R preprocessing scripts. Processing steps included: baseline pupil calculation, baseline correction, aggregation into 5-second time bins, filtering of low-sample bins, and creation of repeated-measures long-format analysis tables. Raw survey and rating data Raw survey contents The raw survey exports include: participant pseudonyms, demographic variables, professional background variables, exposure and experience variables, framing and procedure conditions, and ordinal Likert-scale distress ratings collected after each video observation. Anonymization Survey datasets were anonymized prior to release. No directly identifying information is included in the shared files. Purpose of raw data sharing The raw survey data are shared to support: reproducibility of questionnaire preprocessing, replication of rating analyses, reconstruction of repeated-measures long-format datasets, and transparency of demographic recoding procedures. Relation to processed datasets The processed long-format rating datasets were generated from the raw survey exports using the provided R preprocessing scripts. Processing steps included: recoding demographic variables, harmonization of framing and procedure variables, generation of repeated-measures long-format tables, and merging of survey and eye-tracking datasets for statistical analyses.