Supplemental Material for "You Shall Not Pass! Where and Why Developers Draw The Line on AI Autonomy"

# Contents

1.Participant_demographics.xlsx: Summary statistics for the N = 448 software developers who participated in the study, including SE experience, AI experience, SE responsibilities by SDLC area, and median splits for technophilia and risk tolerance.

2.Survey_constructs&items.xlsx: Survey constructs and exact item wordings. Covers the per-task appraisals (Task Value, Identity, Accountability, Demand), the open-ended Want AI / Don't Want AI questions asked per category block, and developer trait and experience measures (SE experience, AI experience, risk tolerance, technophilia, demographics).

3.Hyponyms&marker_tables.xlsx: Classification algorithm reference tables, the category-specific hyponym table mapping tasks to example phrases (Table S1), and the marker table of class cue phrases used by the classifier (Table S2).

4.Extra_stats.xlsx: Supplementary statistical results. Contains the full 9-level autonomy distribution by SDLC category, the linear mixed-effects model on autonomy level (1–5), the binary (≤L3 vs >L3) logistic model for robustness checks, and variance inflation factors (VIF) for the model predictors.