{
  "study": "brand-priors",
  "registered": "2026-10-28, committed before the final run (seeds 11, 12, 13) was started",
  "spec_digest": "a53a37515e54ba4f",
  "pilot": "seed 1, 6 buyers, 6 arms, 276 model calls, used only to check that prompts parse (0 invalid answers of 936) and that the run completes; its data is not analysed and its journal is not kept.",
  "primary": {
    "outcome": "focal_choice: the buyer chooses the EG-400 in the side-by-side comparison at the base price, $1,490",
    "contrast": "arm kuvra minus arm neutral",
    "estimator": "mean over buyers of the within-buyer difference (each buyer meets both arms), pooled across the three final seeds",
    "interval": "95% percentile bootstrap, resampling buyers, 2,000 replicates, stream seed 0",
    "direction": "two-sided; no direction is predicted",
    "decision": "reported as an effect if the interval excludes zero, otherwise as no detected difference at this sample size"
  },
  "negative_control": {
    "contrast": "neutral-placebo minus neutral: two descriptive names with identical copy",
    "expectation": "interval includes zero; if it does not, the name-swap alone moves outcomes and every identity contrast is read against that"
  },
  "secondary": [
    "orume minus neutral, same outcome and estimator",
    "decomposition: kuvra-name minus neutral (name alone), kuvra minus kuvra-name (positioning copy given the name); likewise for orume",
    "focal_choice at $1,190 and $1,890 by arm (price response)",
    "listing_discovery and engagement_request by arm",
    "LLM-elicited latents (measurement track) by arm: competence, credibility, vendor risk, viability, technical depth, maturity, expected quality, failure rate, maintenance burden, procurement friction, leadership, willingness to pay, overall impression, name associations",
    "expectation gap: choice-track satisfaction after the identical experience report minus measurement-track expected quality, by arm",
    "the primary contrast within each buyer segment (technical, budget, procurement)",
    "factor ablation: leave-one-identity-out prediction of focal_choice from a baseline set (price ratio, overall impression, segment) against the baseline plus the elicited latents"
  ],
  "exclusions": "an answer that does not parse into the item's stated form is missing for that item and the pair is dropped from that contrast; nothing is imputed. Counts of missing answers are reported.",
  "multiplicity": "only the primary contrast is confirmatory. Secondary contrasts are many and are reported as exploratory with unadjusted intervals."
}