{
  "study_id": "NCT07731568",
  "agent": "baseline",
  "run": 3,
  "p_primary_hypothesis_supported": 0.25,
  "effect_direction": "null",
  "effect_size_90pct_interval": {"low": -1.5, "high": 3.0, "metric": "Between-group difference in GAD-7 change, 0-21 scale, at 4 weeks (positive = larger reduction with NPT)"},
  "moderator_prediction": "Higher-baseline-anxiety students show larger absolute GAD-7 drops in both arms (more room, regression to the mean); any NPT-vs-CBT difference, if present, would appear there, but the trial is far too small to detect an interaction.",
  "claim_ids_used": [],
  "rationale": "Active-comparator (CBT) head-to-head with n=55 (~27/arm), a mild-threshold sample (GAD-7>=4), 4-week online group format. Bona fide psychotherapies against each other typically differ by d~0-0.2; detecting that needs hundreds per arm. Both arms will improve substantially (regression to mean, group support), compressing differences. Upward adjustments: developer allegiance (sponsor created NPT), outcome-assessor-only masking with a self-report outcome, small single-site Chinese trials that often report positives, and possible reliance on group x time or 8-week alternatives. Net probability of a significant NPT advantage around 0.25; central estimate ~0.7 GAD-7 points favouring NPT with wide uncertainty."
}