{
  "generated": "2026-07-02",
  "prereg": "preregistration/2026-07-02_study2-phaseA-contact.md",
  "n_items": 122,
  "bootstrap": {
    "resamples": 10000,
    "seed": 20260702
  },
  "arms": {
    "NT_r0": {
      "n": 122,
      "accuracy": 0.6475,
      "mean_confidence": 0.6265,
      "overconfidence": -0.0211,
      "ece": 0.1569,
      "mce": 0.4464,
      "brier": 0.1676,
      "auc": 0.7929
    },
    "NT_post": {
      "n": 122,
      "accuracy": 0.6475,
      "mean_confidence": 0.8056,
      "overconfidence": 0.1581,
      "ece": 0.1581,
      "mce": 0.2857,
      "brier": 0.1621,
      "auc": 0.8272
    },
    "T_r0": {
      "n": 122,
      "accuracy": 0.6475,
      "mean_confidence": 0.6862,
      "overconfidence": 0.0386,
      "ece": 0.123,
      "mce": 0.2143,
      "brier": 0.1276,
      "auc": 0.8699
    },
    "T_post": {
      "n": 122,
      "accuracy": 0.6557,
      "mean_confidence": 0.7892,
      "overconfidence": 0.1335,
      "ece": 0.1335,
      "mce": 0.2857,
      "brier": 0.1556,
      "auc": 0.8176
    }
  },
  "P_A_primary": {
    "delta_NT": 0.1792,
    "delta_T": 0.0949,
    "delta_NT_minus_T_ci": {
      "mean": 0.0852,
      "lo": 0.0234,
      "hi": 0.1452,
      "defined_in": 10000
    },
    "verdict": "CONFIRMED"
  },
  "P_B_mechanism": {
    "turn_uptake": 0.9859,
    "grounded_items": 121,
    "ungrounded_items": 1,
    "verdict": "UNDERPOWERED_BY_CONTRAST",
    "note": "near-universal grounding leaves too few ungrounded items to contrast"
  },
  "P_C_hard_accuracy": {
    "n_hard": 43,
    "accuracy_T": 0.2093,
    "accuracy_NT": 0.186,
    "delta_ci": {
      "mean": 0.0238,
      "lo": -0.0784,
      "hi": 0.1277,
      "defined_in": 10000
    }
  },
  "descriptives": {
    "round0_unanimity": {
      "NT": {
        "unanimous": 30,
        "n": 122,
        "share": 0.2459
      },
      "T": {
        "unanimous": 45,
        "n": 122,
        "share": 0.3689
      }
    },
    "diversity_by_round": {
      "NT": [
        3.2705,
        2.1639
      ],
      "T": [
        2.8361,
        2.2705
      ]
    },
    "tool_turn_uptake": 0.9859,
    "item_grounding_rate": 0.9918,
    "tokens": {
      "NT": 905530,
      "T": 2454628
    },
    "by_stratum": {
      "easy": {
        "NT": {
          "n": 56,
          "accuracy": 1.0,
          "mean_confidence": 0.9668,
          "overconfidence": -0.0332,
          "ece": 0.0332,
          "mce": 0.4286,
          "brier": 0.0084,
          "auc": null
        },
        "T": {
          "n": 56,
          "accuracy": 1.0,
          "mean_confidence": 0.9617,
          "overconfidence": -0.0383,
          "ece": 0.0383,
          "mce": 0.4286,
          "brier": 0.0098,
          "auc": null
        }
      },
      "medium": {
        "NT": {
          "n": 11,
          "accuracy": 0.8182,
          "mean_confidence": 0.8182,
          "overconfidence": 0.0,
          "ece": 0.1039,
          "mce": 0.5714,
          "brier": 0.1299,
          "auc": 0.7778
        },
        "T": {
          "n": 11,
          "accuracy": 0.8182,
          "mean_confidence": 0.7792,
          "overconfidence": -0.039,
          "ece": 0.1169,
          "mce": 0.4286,
          "brier": 0.0909,
          "auc": 0.9444
        }
      },
      "hard": {
        "NT": {
          "n": 43,
          "accuracy": 0.186,
          "mean_confidence": 0.6113,
          "overconfidence": 0.4252,
          "ece": 0.4252,
          "mce": 0.7692,
          "brier": 0.373,
          "auc": 0.6821
        },
        "T": {
          "n": 43,
          "accuracy": 0.2093,
          "mean_confidence": 0.6246,
          "overconfidence": 0.4153,
          "ece": 0.4286,
          "mce": 0.7692,
          "brier": 0.3868,
          "auc": 0.6373
        }
      }
    },
    "by_source": {
      "gsm8k": {
        "NT": {
          "n": 42,
          "accuracy": 0.9286,
          "mean_confidence": 0.9728,
          "overconfidence": 0.0442,
          "ece": 0.0986,
          "mce": 0.2857,
          "brier": 0.0773,
          "auc": 0.4231
        },
        "T": {
          "n": 42,
          "accuracy": 0.9524,
          "mean_confidence": 0.966,
          "overconfidence": 0.0136,
          "ece": 0.0816,
          "mce": 0.5714,
          "brier": 0.0622,
          "auc": 0.45
        }
      },
      "mmlu": {
        "NT": {
          "n": 32,
          "accuracy": 0.6562,
          "mean_confidence": 0.9241,
          "overconfidence": 0.2679,
          "ece": 0.2946,
          "mce": 0.4643,
          "brier": 0.287,
          "auc": 0.5844
        },
        "T": {
          "n": 32,
          "accuracy": 0.6875,
          "mean_confidence": 0.9241,
          "overconfidence": 0.2366,
          "ece": 0.2545,
          "mce": 0.7143,
          "brier": 0.2787,
          "auc": 0.4773
        }
      },
      "triviaqa": {
        "NT": {
          "n": 36,
          "accuracy": 0.3611,
          "mean_confidence": 0.5278,
          "overconfidence": 0.1667,
          "ece": 0.1667,
          "mce": 0.8571,
          "brier": 0.1531,
          "auc": 0.8963
        },
        "T": {
          "n": 36,
          "accuracy": 0.3333,
          "mean_confidence": 0.5317,
          "overconfidence": 0.1984,
          "ece": 0.1984,
          "mce": 0.3571,
          "brier": 0.1848,
          "auc": 0.8542
        }
      }
    },
    "locked_controls": {
      "NT": {
        "n": 12,
        "accuracy": 0.5,
        "mean_confidence": 0.7381,
        "overconfidence": 0.2381,
        "ece": 0.2381,
        "mce": 0.5714,
        "brier": 0.1531,
        "auc": 0.9861
      },
      "T": {
        "n": 12,
        "accuracy": 0.5,
        "mean_confidence": 0.5833,
        "overconfidence": 0.0833,
        "ece": 0.1071,
        "mce": 0.2857,
        "brier": 0.0663,
        "auc": 0.9861
      }
    }
  }
}