Claim at v1.1.0

restaurants-nta-harshness

6b3fe7469b707a663676daee6d0aa3a849c99d43fc53b81164fcfa3e3a04cec8

Data

{
  "has_outcome": true,
  "kind": "confirmation",
  "lineage": {
    "audit": "docs/lineage_audit.md; data/derived/lineage_audit.json",
    "class": "none",
    "class_meaning": "the audit found no city-data source fetched before registration",
    "classes": [],
    "early_sources": []
  },
  "outcome": "restaurants-nta-harshness",
  "prereg_branch": {
    "name": "prereg/restaurants-nta-harshness",
    "sha": "55dabf10dd2516710e6509521f4cfef641e0039f"
  },
  "prior": 0.55,
  "registered_at": "2026-11-13T14:18:37-0500",
  "registrant": "claude (restaurants agent, 13 November 2026)",
  "registration": {
    "alternatives": [
      "Inspector assignment: inspectors work in geographic teams, so an NTA's rate reflects its inspectors' stringency rather than its restaurants.",
      "Unmeasured restaurant mix: size, menu complexity, age of premises and owner experience vary by NTA and are not in the data; cuisine is a weak proxy.",
      "Building stock: older buildings have more pests (the most common critical violations are vermin), which is real hygiene, not harshness.",
      "Clientele and complaints: complaint-driven attention may differ by NTA.",
      "Adjudication: the snapshot holds violations before some hearings; dismissal rates may differ by NTA."
    ],
    "graph_evidence": [
      {
        "association_id": "CCatfE2F4k (variable) chain restaurants -- violations per inspection",
        "direction": "-",
        "n_findings": 1,
        "n_studies": 1,
        "pubmed_ids": [
          "28060568"
        ],
        "summary": "Philadelphia 2013-2014 (Leinwand 2017): chains 6.5 against 9.6 violations per inspection; used here as a covariate."
      }
    ],
    "id": "restaurants-nta-harshness",
    "kind": "confirmation",
    "literature": [
      {
        "citation": "Jones TF et al. Emerg Infect Dis 2004;10:688-692 (PMID 15200861)",
        "doi": "10.3201/eid1004.030343",
        "finding": "Tennessee: mean scores of individual inspectors ranged from 69 to 92 on a 100-point scale, so who inspects changes the result."
      },
      {
        "citation": "Leinwand SE et al. Public Health Rep 2017;132:180-187 (PMID 28060568)",
        "doi": "10.1177/0033354916687741",
        "finding": "Philadelphia: violations per inspection differ by chain status (6.5 against 9.6) and by block-group racial composition (0.6 fewer risk-factor and 1.0 more retail-practice violations), and fall with inspection frequency in non-chains."
      },
      {
        "citation": "Wong MR et al. Am J Public Health 2015;105:e81-e87 (PMID 25602861)",
        "doi": "10.2105/AJPH.2014.302404",
        "finding": "NYC models of inspection outcomes adjust for season and chain status; grading changed scores citywide."
      }
    ],
    "plan": "Script analyses/restaurants/evenness_test.py at commit 48147d6. Fixed source: Underlay nyc/restaurant-inspections v1.0.0 (hash ulv2:07a7f08a0d5c85649920f8adc6946d27c17f5e2d75e9ee07c0ad4ae0fcc6739c), Manhattan inspections 2024-01-02 to 2026-06-09, chosen because the live table drops closed restaurants and revises recent violations downward (docs/underlay_drift.md). NTA by point in polygon (9nt8-h7nd); NTAs with < 20 restaurants dropped. Unit: Cycle Inspection / Initial Inspection; outcome: critical violation records. Null Poisson GLM as in the script docstring; statistic S = sum over NTAs of (sum(y - mu))^2 / sum(mu); null by permuting NTA labels across restaurants. tau from an NTA fixed-effects model, SE clustered by restaurant, method of moments. A companion frequency test (cycle initial inspections per restaurant-year) is computed by the same script; its result was seen by accident during a synthetic-outcome check of the script before this registration, so it is reported as exploratory and is not registered.",
    "prediction": {
      "direction": "+",
      "magnitude": "Between-NTA SD of the adjusted log rate ratio (tau) 0.05 to 0.20; permutation p <= 0.01.",
      "other": "p <= 0.01 with tau < 0.05 means statistically detectable but practically even and is unsettled under rule 2. Per rule 9 the p condition is re-checked with 100,000 draws and other seeds; if it is met under some and not others, it is neither met nor unmet.",
      "refute_if": "Permutation p > 0.10 (the NTAs are as even as random assignment of restaurants to NTAs).",
      "support_if": "Permutation p <= 0.01 (10,000 draws, seed 20261113) AND tau >= 0.05."
    },
    "prior": 0.55,
    "registered_at": "2026-11-13T14:18:37-0500",
    "registrant": "claude (restaurants agent, 13 November 2026)",
    "script_commit": "48147d6",
    "sha256": "6f6e2eac57f3253c217e8cf0115e3b40ee9999caf31641c9e2e48a3ecaa53bc7",
    "statement": "Critical violations detected per routine initial inspection are not even across Manhattan's 2020 NTAs: after cuisine, chain status, quarter, day of week, daily inspection workload, new-restaurant status and the NTA's restaurant count, the between-NTA variation exceeds a permutation null, with a between-NTA standard deviation of 5 to 20 percent."
  },
  "registration_commit": {
    "commit": "55dabf10dd2516710e6509521f4cfef641e0039f",
    "committed_at": "2026-11-13T14:18:37-05:00"
  },
  "registration_file_sha256": "f4f49ad439361d489415c0d23533a8104f91803418d3744de4bb19e27ea935e5",
  "registration_path": "claims/registered/restaurants-nta-harshness.json",
  "registration_sha256": "6f6e2eac57f3253c217e8cf0115e3b40ee9999caf31641c9e2e48a3ecaa53bc7",
  "registration_text": "{\n \"id\": \"restaurants-nta-harshness\",\n \"kind\": \"confirmation\",\n \"statement\": \"Critical violations detected per routine initial inspection are not even across Manhattan's 2020 NTAs: after cuisine, chain status, quarter, day of week, daily inspection workload, new-restaurant status and the NTA's restaurant count, the between-NTA variation exceeds a permutation null, with a between-NTA standard deviation of 5 to 20 percent.\",\n \"graph_evidence\": [\n  {\n   \"association_id\": \"CCatfE2F4k (variable) chain restaurants -- violations per inspection\",\n   \"pubmed_ids\": [\n    \"28060568\"\n   ],\n   \"direction\": \"-\",\n   \"n_findings\": 1,\n   \"n_studies\": 1,\n   \"summary\": \"Philadelphia 2013-2014 (Leinwand 2017): chains 6.5 against 9.6 violations per inspection; used here as a covariate.\"\n  }\n ],\n \"literature\": [\n  {\n   \"citation\": \"Jones TF et al. Emerg Infect Dis 2004;10:688-692 (PMID 15200861)\",\n   \"doi\": \"10.3201/eid1004.030343\",\n   \"finding\": \"Tennessee: mean scores of individual inspectors ranged from 69 to 92 on a 100-point scale, so who inspects changes the result.\"\n  },\n  {\n   \"citation\": \"Leinwand SE et al. Public Health Rep 2017;132:180-187 (PMID 28060568)\",\n   \"doi\": \"10.1177/0033354916687741\",\n   \"finding\": \"Philadelphia: violations per inspection differ by chain status (6.5 against 9.6) and by block-group racial composition (0.6 fewer risk-factor and 1.0 more retail-practice violations), and fall with inspection frequency in non-chains.\"\n  },\n  {\n   \"citation\": \"Wong MR et al. Am J Public Health 2015;105:e81-e87 (PMID 25602861)\",\n   \"doi\": \"10.2105/AJPH.2014.302404\",\n   \"finding\": \"NYC models of inspection outcomes adjust for season and chain status; grading changed scores citywide.\"\n  }\n ],\n \"prediction\": {\n  \"direction\": \"+\",\n  \"magnitude\": \"Between-NTA SD of the adjusted log rate ratio (tau) 0.05 to 0.20; permutation p <= 0.01.\",\n  \"support_if\": \"Permutation p <= 0.01 (10,000 draws, seed 20261113) AND tau >= 0.05.\",\n  \"refute_if\": \"Permutation p > 0.10 (the NTAs are as even as random assignment of restaurants to NTAs).\",\n  \"other\": \"p <= 0.01 with tau < 0.05 means statistically detectable but practically even and is unsettled under rule 2. Per rule 9 the p condition is re-checked with 100,000 draws and other seeds; if it is met under some and not others, it is neither met nor unmet.\"\n },\n \"plan\": \"Script analyses/restaurants/evenness_test.py at commit 48147d6. Fixed source: Underlay nyc/restaurant-inspections v1.0.0 (hash ulv2:07a7f08a0d5c85649920f8adc6946d27c17f5e2d75e9ee07c0ad4ae0fcc6739c), Manhattan inspections 2024-01-02 to 2026-06-09, chosen because the live table drops closed restaurants and revises recent violations downward (docs/underlay_drift.md). NTA by point in polygon (9nt8-h7nd); NTAs with < 20 restaurants dropped. Unit: Cycle Inspection / Initial Inspection; outcome: critical violation records. Null Poisson GLM as in the script docstring; statistic S = sum over NTAs of (sum(y - mu))^2 / sum(mu); null by permuting NTA labels across restaurants. tau from an NTA fixed-effects model, SE clustered by restaurant, method of moments. A companion frequency test (cycle initial inspections per restaurant-year) is computed by the same script; its result was seen by accident during a synthetic-outcome check of the script before this registration, so it is reported as exploratory and is not registered.\",\n \"alternatives\": [\n  \"Inspector assignment: inspectors work in geographic teams, so an NTA's rate reflects its inspectors' stringency rather than its restaurants.\",\n  \"Unmeasured restaurant mix: size, menu complexity, age of premises and owner experience vary by NTA and are not in the data; cuisine is a weak proxy.\",\n  \"Building stock: older buildings have more pests (the most common critical violations are vermin), which is real hygiene, not harshness.\",\n  \"Clientele and complaints: complaint-driven attention may differ by NTA.\",\n  \"Adjudication: the snapshot holds violations before some hearings; dismissal rates may differ by NTA.\"\n ],\n \"prior\": 0.55,\n \"registrant\": \"claude (restaurants agent, 13 November 2026)\",\n \"script_commit\": \"48147d6\",\n \"registered_at\": \"2026-11-13T14:18:37-0500\",\n \"sha256\": \"6f6e2eac57f3253c217e8cf0115e3b40ee9999caf31641c9e2e48a3ecaa53bc7\"\n}",
  "replication": {
    "built_at": "2026-11-13T23:35:50-0500",
    "checked_at": "2026-11-13T23:36:06-0500",
    "decides": "support_if: permutation p <= 0.01 and tau >= 0.05; refute_if: permutation p > 0.10",
    "deciding_number": "city_effect: between-NTA SD (tau) of the adjusted log rate ratio of critical violations per routine initial inspection, with the permutation p of the NTA score statistic (10,000 draws, seed 20261113)",
    "inputs_copied": 2,
    "manifest": "replication/restaurants-nta-harshness/manifest.json",
    "manifest_sha256": "e4193e3444e9c8fcd1540471bf1af850b459c036982192e507a3f1c0573c4d35",
    "package": "replication/restaurants-nta-harshness/",
    "recorded": {
      "estimate": 0,
      "n": 12228,
      "n_ntas": 31,
      "n_restaurants": 8157,
      "permutation_p": 0.3067
    },
    "status": "MATCH",
    "status_from": "replication/restaurants-nta-harshness/verify_result.json",
    "tables_not_rebuilt": 0,
    "tolerance": {
      "estimate": 0.0001,
      "n": 0,
      "n_ntas": 0,
      "n_restaurants": 0,
      "permutation_p": 0.0001
    }
  },
  "statement": "Critical violations detected per routine initial inspection are not even across Manhattan's 2020 NTAs: after cuisine, chain status, quarter, day of week, daily inspection workload, new-restaurant status and the NTA's restaurant count, the between-NTA variation exceeds a permutation null, with a between-NTA standard deviation of 5 to 20 percent."
}

History in this collection

  1. addedv1.0.0Nov 14, 2026this version