{
  "release_cohort_contract": {
    "contract_version": 2,
    "research_provenance": {
      "research_consent_field": "student_consented_to_research",
      "research_consent_required_value": true,
      "main_cohort_membership_basis": "physical_placement_in_1_main_leaderboard_data",
      "public_profile_administrative_flags": [],
      "source_collections": [
        {
          "collection_id": "quant_run1_ai_human",
          "instrument": "gre_quant",
          "arms": [
            "ai",
            "human"
          ],
          "experiment_run_ids": [
            1
          ],
          "ai_model_preset_ids": [
            "gemini-3.1-pro-high",
            "gemini-3.5-flash-low",
            "gemma-4-31b-high",
            "gpt-5.4-mini-none",
            "gpt-5.5-high",
            "gpt-5.5-pro-med",
            "kimi-k2.6",
            "opus-4.8-off",
            "opus-4.8-xhigh",
            "sonnet-4.6-low"
          ]
        },
        {
          "collection_id": "quant_run8_corrected_ai_and_control",
          "instrument": "gre_quant",
          "arms": [
            "ai",
            "control"
          ],
          "experiment_run_ids": [
            8
          ],
          "ai_model_preset_ids": [
            "gemini-3.6-flash-low",
            "opus-5-high"
          ]
        },
        {
          "collection_id": "quant_run11_gemini37",
          "instrument": "gre_quant",
          "arms": [
            "ai"
          ],
          "experiment_run_ids": [
            11
          ],
          "ai_model_preset_ids": [
            "gemini-3.7-flash-medium"
          ]
        },
        {
          "collection_id": "verbal_run4_ai_human",
          "instrument": "gre_verbal",
          "arms": [
            "ai",
            "human"
          ],
          "experiment_run_ids": [
            4
          ],
          "ai_model_preset_ids": [
            "gemini-3.1-pro-high",
            "gemini-3.5-flash-low",
            "gemma-4-31b-high",
            "gpt-5.4-mini-none",
            "gpt-5.5-high",
            "gpt-5.5-pro-med",
            "kimi-k2.6",
            "opus-4.8-off",
            "opus-4.8-xhigh",
            "sonnet-5-low"
          ]
        },
        {
          "collection_id": "verbal_run6_corrected_ai",
          "instrument": "gre_verbal",
          "arms": [
            "ai"
          ],
          "experiment_run_ids": [
            6
          ],
          "ai_model_preset_ids": [
            "gemini-3.6-flash-low",
            "opus-5-high"
          ]
        },
        {
          "collection_id": "verbal_run9_control",
          "instrument": "gre_verbal",
          "arms": [
            "control"
          ],
          "experiment_run_ids": [
            9
          ],
          "ai_model_preset_ids": []
        },
        {
          "collection_id": "verbal_run12_gemini37",
          "instrument": "gre_verbal",
          "arms": [
            "ai"
          ],
          "experiment_run_ids": [
            12
          ],
          "ai_model_preset_ids": [
            "gemini-3.7-flash-medium"
          ]
        }
      ]
    },
    "instrument_eligibility": {
      "gre_quant": {
        "assessment_bank_version_id": "gre_quant:assembly_main:27q-section47m:2c31aa3e",
        "pretest_score_minimum_inclusive": 5,
        "pretest_score_maximum_inclusive": 24
      },
      "gre_verbal": {
        "assessment_bank_version_id": "gre_verbal:27q-section41m:037ffb6b",
        "pretest_score_minimum_inclusive": 4,
        "pretest_score_maximum_inclusive": 24
      }
    },
    "assessment_effort": {
      "minimum_item_time_seconds_each_test_inclusive": 600,
      "maximum_tab_outs_inclusive": 10,
      "scope": "pretest_and_posttest"
    },
    "attempt_adjudication": {
      "rule": "first_same_instrument_attempt_per_account",
      "within_run_duplicate_rule": "completed_then_earliest",
      "restricted_identity_verification_required_for_complete_check": true
    },
    "treatment_integrity": {
      "cross_model_lesson_plan_fallback_allowed": false,
      "substantive_additional_treatment_allowed": false,
      "provenance_pilot_arms_allowed_in_main": false,
      "incidental_exposure_policy": "Retain only when PI adjudication documents exposure as non-substantive."
    },
    "control_pretest_review": {
      "minimum_seconds_inclusive": 300,
      "scope": "control_tasks_only",
      "required_condition": "reviewed_pretest"
    },
    "outcome_fields_forbidden_for_selection": [
      "posttest_score_points",
      "score_lift_points",
      "headroom_normalized_score_lift",
      "posttest_irt_theta",
      "irt_theta_lift",
      "posttest_correctness"
    ]
  },
  "analysis_cohort_eligibility": {
    "apply_observed_outcome_filter": false,
    "applies_to": [
      "gre_quant",
      "gre_verbal"
    ],
    "assessment_question_count": 27,
    "assessment_completion_boolean": "pre_and_post",
    "assessment_completion_scope": "all_main_leaderboard_modalities",
    "minimum_attempted_questions_each_test_inclusive": 14,
    "physically_pruned_main_leaderboard": true,
    "scope": "physically_pruned_main_leaderboard_data"
  },
  "human_tutor_cost_sensitivity": {
    "fixed_session_hours": 1.0,
    "hourly_wage_usd": 75.0,
    "interpretation": "Author-specified cost sensitivity scenario; not an observed human-tutor charge in the released study data."
  },
  "resampling": {
    "cost_per_gain_paired_participant_bootstrap": {
      "draws": 10000,
      "minimum_valid_draw_fraction": 0.95,
      "seed": 20260814
    },
    "difficulty_calibration_participant_bootstrap": {
      "draws": 4000,
      "seed": 20260814
    },
    "early_mastery_cross_validation": {
      "folds": 10,
      "seed": 20260817
    },
    "engagement_participant_bootstrap": {
      "draws": 4000,
      "seed": 20260814
    },
    "gemini36_random_deletion_sensitivity": {
      "iterations": 5000,
      "seed_root": 20260814
    },
    "human_ai_section_permutation": {
      "chunk_size": 1000,
      "draws": 20000,
      "seed": 20260807
    },
    "problem_flag_participant_bootstrap": {
      "draws": 4000,
      "seed": 20260814
    }
  },
  "protocol_adherence": {
    "ai_tutoring_activity": {
      "all_three_required": true,
      "distinct_problem_definition": "A practice problem is counted when answer_submission_count is at least 1; released data require this to agree with was_answered.",
      "minimum_distinct_practice_problems_attempted_inclusive": 3,
      "minimum_student_authored_chat_messages_inclusive": 3,
      "minimum_total_practice_answer_attempts_inclusive": 5,
      "scope": "ai_tutored_tasks_only",
      "student_message_source": "student_ai_tutor_interaction_trajectory.csv rows with event_type=student_message and actor=student",
      "total_attempt_source": "sum of answer_submission_count in student_practice_problem_attempts.csv"
    },
    "arm_identity_exclusions": [],
    "assessment_timing": {
      "cutoff_inclusive": true,
      "deduplication": "For each phase and question_id, use the earliest answer_submit event by server_elapsed_seconds.",
      "event_timestamp": "server_elapsed_seconds",
      "final_seconds_ignored": 300,
      "insufficient_support_action": "retain_unless_another_gate_fails",
      "instrument_windows": {
        "gre_quant": {
          "assessment_budget_seconds": 2820,
          "cutoff_minutes_inclusive": 42,
          "cutoff_seconds_inclusive": 2520
        },
        "gre_verbal": {
          "assessment_budget_seconds": 2460,
          "cutoff_minutes_inclusive": 36,
          "cutoff_seconds_inclusive": 2160
        }
      },
      "interval_definition": "Chronological differences between consecutive retained first submissions within a phase.",
      "minimum_intervals_inclusive": 13,
      "percentile_method": "linear",
      "phase_boolean": "pre_or_post",
      "phase_start_timestamp": "Use payload_json.<phase>_started_elapsed_seconds when present and valid; otherwise use server_elapsed_seconds of the phase_started event.",
      "primary_rule": {
        "label": "p90_before_final_five_minutes",
        "percentile": 90,
        "rapid_interval_threshold_seconds_exclusive": 10.0
      },
      "scope": "all_main_leaderboard_modalities",
      "submission_event": "answer_submit",
      "sensitivity_rule": {
        "label": "p80_before_final_five_minutes",
        "percentile": 80,
        "rapid_interval_threshold_seconds_exclusive": 10.0
      }
    },
    "task_level_boolean": "assessment_completion_and_assessment_timing_and_applicable_ai_tutoring_activity"
  },
  "schema_version": 8,
  "statistical_inference": {
    "benjamini_hochberg_fdr_alpha": 0.05,
    "confidence_level": 0.95,
    "equivalence_margins_pooled_sd": {
      "primary": 0.25,
      "sensitivity": 0.2
    },
    "equivalence_tost_alpha": 0.05,
    "holm_familywise_alpha": 0.05,
    "nominal_two_sided_alpha": 0.05
  }
}