{
  "study_id": "NCT07513337",
  "agent": "baseline",
  "run": 2,
  "p_primary_hypothesis_supported": 0.25,
  "effect_direction": "null",
  "effect_size_90pct_interval": {"low": -0.08, "high": 0.12, "metric": "Mean difference in one-time-use harm perception, 0-4 scale, labelled/claimed ads vs no-label/no-claim (positive = lower perceived harm)"},
  "moderator_prediction": "Inexperienced users may shift slightly more, since they have weaker prior beliefs about strains and rely more on ad cues; experienced users' harm ratings are anchored by their own use.",
  "claim_ids_used": [],
  "rationale": "Single-exposure ad manipulation in an online survey; strain labels and effect claims are subtle cues that do not mention safety, so one-time-use harm perception (already low among users) likely barely moves. Power is good (about 760 vs 2280 for label, 1010 vs 2020 for claim; SE about 0.04 points), so an effect of about 0.08-0.1 would be detected, but two tests give two chances. Claims like 'sedating' or 'energizing' could even raise perceived risk. The primer's persuasion evidence shows small effects on items the ad does not address. I put the chance that at least one main effect significantly lowers harm perception at about 25%."
}