{
  "preRegistered": false,
  "unit": "conversation",
  "seed": 20261008,
  "permutations": 10000,
  "resamples": 2000,
  "comparisons": {
    "break vs control": {
      "perModel": {
        "amazon/nova-2-lite-v1": {
          "difference": -0.003,
          "low": -0.0075,
          "high": 0.0
        },
        "anthropic/claude-haiku-4.5": {
          "difference": 0.035867346938775524,
          "low": 0.00849999999999998,
          "high": 0.06150000000000003
        },
        "google/gemini-3-flash-preview": {
          "difference": 0.005998995493742868,
          "low": -0.00048572859288933436,
          "high": 0.012012012012012012
        },
        "meta-llama/llama-4-maverick": {
          "difference": 0.0055,
          "low": 0.0,
          "high": 0.0145
        },
        "openai/gpt-5.4-mini": {
          "difference": 0.005500000000000005,
          "low": -0.031000000000000028,
          "high": 0.041000000000000036
        },
        "x-ai/grok-4.20": {
          "difference": -0.007000000000000006,
          "low": -0.03900000000000001,
          "high": 0.025999999999999995
        }
      },
      "excessErrors": 42.49954637695445,
      "p": 0.1533846615338466
    },
    "pause vs control": {
      "perModel": {
        "amazon/nova-2-lite-v1": {
          "difference": 0.2035,
          "low": 0.16899999999999998,
          "high": 0.2375
        },
        "anthropic/claude-haiku-4.5": {
          "difference": -0.0011326530612244812,
          "low": -0.025999999999999995,
          "high": 0.022999999999999993
        },
        "google/gemini-3-flash-preview": {
          "difference": 0.001996994493242617,
          "low": -0.0040115283145158625,
          "high": 0.0085
        },
        "meta-llama/llama-4-maverick": {
          "difference": 0.0095,
          "low": 0.0,
          "high": 0.026
        },
        "openai/gpt-5.4-mini": {
          "difference": 0.01100000000000001,
          "low": -0.026000000000000023,
          "high": 0.04749999999999999
        },
        "x-ai/grok-4.20": {
          "difference": 0.015000000000000013,
          "low": -0.017499999999999988,
          "high": 0.04500000000000001
        }
      },
      "excessErrors": 239.87428450125472,
      "p": 9.999000099990002e-05
    }
  }
}