{
  "study_id": "NCT07611422",
  "agent": "baseline",
  "run": 5,
  "p_primary_hypothesis_supported": 0.18,
  "effect_direction": "null",
  "effect_size_90pct_interval": {"low": -0.2, "high": 0.45, "metric": "Cohen's d for positive-information vs neutral-information on PSS-10 at 4 weeks (positive = less stress with positive information)"},
  "moderator_prediction": "Expectancy effect slightly larger in the cold-shower arms, where a positive rationale makes an unpleasant task credible and may raise adherence; warm-shower rationale is less plausible. Interaction unlikely to reach significance.",
  "claim_ids_used": [],
  "rationale": "A one-off verbal rationale aimed at a trait-like outcome (PSS-10 covers the past month) four weeks later should give a small effect: the primer shows expectation effects are large for acute self-reported outcomes but small for chronic ones, and non-clinical OLP effects are about 0.29. With about 60 per expectancy level in healthy volunteers, power for d of about 0.2 is roughly 20%. The time interaction adds a little extra chance, as do self-report, participant masking, and flexible analyses in a small academic trial. The base rate for pre-registered psychology experiments is about 44%, but this design is underpowered for the realistic effect."
}