{
  "generated": "2026-07-03",
  "prereg": "preregistration/2026-07-03_study2-independence.md",
  "n_items": 90,
  "bootstrap": {
    "resamples": 10000,
    "seed": 20260703
  },
  "cells": {
    "HOMO": "7x claude-haiku-4-5-20251001",
    "HETERO": "4x claude-haiku-4-5 + 3x gpt-4o-mini-2024-07-18"
  },
  "arms": {
    "HOMO_r0": {
      "n": 90,
      "accuracy": 0.7778,
      "mean_confidence": 0.8667,
      "overconfidence": 0.0889,
      "ece": 0.1079,
      "mce": 0.2857,
      "brier": 0.1365,
      "auc": 0.8014
    },
    "HOMO_post": {
      "n": 90,
      "accuracy": 0.7667,
      "mean_confidence": 0.9381,
      "overconfidence": 0.1714,
      "ece": 0.1714,
      "mce": 0.2857,
      "brier": 0.1859,
      "auc": 0.6563
    },
    "HETERO_r0": {
      "n": 90,
      "accuracy": 0.7444,
      "mean_confidence": 0.7873,
      "overconfidence": 0.0429,
      "ece": 0.1444,
      "mce": 0.2857,
      "brier": 0.158,
      "auc": 0.7437
    },
    "HETERO_post": {
      "n": 90,
      "accuracy": 0.7667,
      "mean_confidence": 0.9016,
      "overconfidence": 0.1349,
      "ece": 0.1508,
      "mce": 0.4286,
      "brier": 0.1698,
      "auc": 0.7188
    }
  },
  "P_I1_primary_independence_lowers_overconf": {
    "delta_ci": {
      "mean": 0.0366,
      "lo": -0.0079,
      "hi": 0.081
    },
    "verdict": "NOT CONFIRMED"
  },
  "P_I2_independence_blunts_debate_inflation": {
    "delta_ci": {
      "mean": -0.0095,
      "lo": -0.0714,
      "hi": 0.054
    },
    "verdict": "NOT CONFIRMED"
  },
  "P_I3_manipulation_check_hetero_more_diverse": {
    "delta_ci": {
      "mean": 0.3115,
      "lo": 0.1111,
      "hi": 0.5222
    },
    "verdict": "CONFIRMED",
    "note": "if NOT CONFIRMED the independence manipulation did not produce disagreement; P-I1 is uninterpretable"
  },
  "P_I4_accuracy_estimation_confounded": {
    "delta_ci": {
      "mean": -0.0001,
      "lo": -0.0333,
      "hi": 0.0333
    },
    "note": "estimation only, not a significance test; confounded by any capability gap between the two families"
  },
  "descriptives": {
    "round0_unanimity": {
      "HOMO": 0.6111,
      "HETERO": 0.4889
    },
    "diversity_by_round": {
      "HOMO": [
        1.7222,
        1.2778
      ],
      "HETERO": [
        2.0333,
        1.4444
      ]
    },
    "tokens": {
      "HOMO": 521796,
      "HETERO": 455311
    }
  }
}