Aurevia Digital CEO Earnings Update Experiment Dataset: Accountability Attribution, Trust, and Investment Allocation (N=266)

This dataset contains participant-level survey and behavioral-decision data from a randomized, video-based earnings disclosure experiment examining how accountability attribution cues shape perceived credibility and investment decisions. Participants were randomly assigned to one of two conditions: High Accountability vs Low Accountability attribution, while all other stimulus elements (earnings content, duration, and delivery format) were held constant. The study focuses on whether attributing responsibility to an executive speaker versus an AI-supported communication process changes (i) perceived accountability, (ii) trust/credibility, and (iii) investment allocation.

The Excel file includes four sheets: (1) Data (raw responses plus derived variables; N=266), (2) Codebook (variable definitions and coding), (3) Summary_Tables (key descriptives, balance checks, manipulation checks, and main outcomes), and (4) Figures (figure-ready values/outputs).

Key variable groups in the Data sheet include: demographics (age, gender, country), investing background (prior investing, experience 1–5, trading frequency), a financial knowledge screener, watch time (seconds), manipulation checks (Q9 responsibility attribution; Q10 personal accountability; Q11 AI-prepared perception), comprehension checks (Q12–Q14, with item-level correctness and ComprehensionScore_0to3), perceived accountability mediators (Q15–Q16), trust/credibility mediators (Q17–Q22), presence/automation perceptions (Q23–Q24), the primary behavioral outcome (Q25_Alloc_Aurevia_USD, 0–10,000; with the residual risk-free allocation), willingness-to-invest items (Q26–Q28), controls (Q29 risk tolerance; Q30 baseline AI trust), and an attention check (Q31, correct response = 6).

Derived composites are provided for replication and analysis: PerceivedAccountability_Comp (mean of Q15–Q16), TrustCredibility_Comp (mean of Q17–Q22), and Willingness_Comp (mean of Q26–Q28). Exclusion flags are included (Exclude_Attention, Exclude_Watch for watch time < 60s, Exclude_Comprehension for comprehension score < 2) along with Include_Analysis to identify the analysis-ready sample used in the summary tables. Researchers can reproduce the main tests using simple t-tests/OLS or mediation models by filtering on Include_Analysis = 1.