Claim at v1.0.0
heat-shootings
ef850c8860cbc9f2656b9de28bf86f2715e3558cba055580a7fa12e47b186bb6
Data
{
"has_outcome": true,
"kind": "confirmation",
"lineage": {
"audit": "docs/lineage_audit.md; data/derived/lineage_audit.json",
"class": "a",
"class_meaning": "exposure or treatment only, not the outcome",
"classes": [
"a"
],
"early_sources": [
{
"class": "a",
"dataset": "GHCN-Daily v3 (NCEI), doi:10.7289/V5D21VHZ",
"evidence": "Daily temperature (GHCN-Daily) is the exposure, not the outcome. It is weather, fetched for earlier heat claims and reused.",
"fetched": "2026-11-12T15:06:44-05:00"
}
]
},
"outcome": "heat-shootings",
"prior": 0.9,
"registered_at": "2026-11-12T15:19:38-0500",
"registrant": "Claude (heat agent), for Jeff Vernon",
"registration": {
"alternatives": [
"Residual seasonality within month (school calendar, summer youth programmes, daylight) correlated with temperature",
"Time outdoors and street encounters as the mechanism (the temperature effect is real but works through routine activity, not physiology); the design cannot separate mechanisms",
"Police summer deployments (e.g. Summer All Out) that respond to expected violence",
"The 2020-2021 rise in shootings coinciding with summers; handled by year terms but not by year-by-month"
],
"graph_evidence": [
{
"concepts": [
"temperature",
"homicide"
],
"direction": "+",
"n_findings": 4,
"n_studies": 2,
"pubmed_ids": [
36858215,
32738768
],
"relationship_id": "4FaXcuNJRt|YZku29igbl",
"summary": "Two case-crossover studies. Rahman et al. 2023 (California 2014-2019): homicide odds +4.9% (1.6 to 8.1) per 5 C Tmax lag 0. Xu et al. 2020 (9 US cities 2007-2017): intentional homicide +9.5% (4.3 to 15.0) per 5 C mean temperature in Chicago and +8.8% (1.5 to 16.6) in New York, lag 0-7. The graph stores Xu's estimates as odds ratios 0.095 and 0.088 and Rahman's as odds ratios 4.9 and 6.2; the abstracts give percent changes, so the graph values are mis-extracted (true ORs about 1.095, 1.088, 1.049, 1.062)."
},
{
"concepts": [
"temperature",
"injury"
],
"direction": "+",
"n_findings": 4,
"n_studies": 3,
"pubmed_ids": [
37448837
],
"relationship_id": "4FaXcuNJRt|rULjCt1BqH",
"summary": "Girma et al. 2023, New York City ED visits 2005-2011 (SPARCS), warm season: intentional injury among 20-25 year olds OR 1.54 (1.28 to 1.85) for daily minimum 77 F vs 48 F, lag 0-5. The only NYC violence-temperature finding in the graph."
},
{
"concepts": [
"gun violence",
"temperature"
],
"direction": "0",
"n_findings": 0,
"n_studies": 0,
"pubmed_ids": [],
"relationship_id": "(concept X7hfhwez9f, 34 relationships)",
"summary": "The gun violence concept has no relationship with temperature, air temperature, heat wave, hot or extreme heat. Violence (2qLKgaxkm_), crime, aggression and violent crime have none either. The graph's heat-violence evidence is homicide and injury only."
},
{
"concepts": [
"temperature",
"suicide"
],
"direction": "+",
"n_findings": 12,
"n_studies": 8,
"pubmed_ids": [
35394808,
34034696,
36858215
],
"relationship_id": "4FaXcuNJRt|TLeQZQ_leT",
"summary": "Self-directed violence rises with temperature in 8 studies (e.g. RR 1.35, 1.26 to 1.45, PMID 35394808; per 1 C IRR 1.008, PMID 34034696). Context for the violence claim."
}
],
"id": "heat-shootings",
"kind": "confirmation",
"literature": [
{
"citation": "Reeping PM, Hemenway D. The association between weather and the number of daily shootings in Chicago (2012-2016). Inj Epidemiol 2020;7:31.",
"doi": "10.1186/s40621-020-00260-3",
"finding": "Negative binomial GEE with month and year fixed effects: +34.0% (28.1 to 40.2) shootings per 10 C daily high on weekdays, +41.6% (34.7 to 48.9) on weekends and holidays; per 5 C that is IRR 1.158 and 1.190."
},
{
"citation": "Lyons VH, Gause EL, Spangler KR, Wellenius GA, Jay J. Analysis of daily ambient temperature and firearm violence in 100 US cities. JAMA Netw Open 2022;5(12):e2247207.",
"doi": "10.1001/jamanetworkopen.2022.47207",
"finding": "Gun Violence Archive 2015-2020, distributed-lag models, lag 0 best: RR 1.17 (1.12 to 1.21) at the 84th percentile of Tmax vs the city median; 6.85% of shootings attributable to days above the median; Northeast attributable fraction 9.54%."
},
{
"citation": "Xu R, Xiong X, Abramson MJ, Li S, Guo Y. Ambient temperature and intentional homicide: a multi-city case-crossover study in the US. Environ Int 2020;143:105992.",
"doi": "10.1016/j.envint.2020.105992",
"finding": "New York 2007-2017: +8.8% (1.5 to 16.6) intentional homicide per 5 C daily mean temperature, lag 0-7; 7.1% attributable to temperatures above the median."
},
{
"citation": "Rahman MM et al. Ambient temperature and air pollution associations with suicide and homicide mortality in California. Sci Total Environ 2023;874:162462.",
"doi": "10.1016/j.scitotenv.2023.162462",
"finding": "+4.9% (1.6 to 8.1) homicide per 5 C Tmax lag 0."
}
],
"plan": "Data: NYPD Shooting Incident Data, 5ucz-vwe8 (Shootings 2006-Present) as primary, count of distinct incident_key per occur_date, 2006-01-01 to the last full month; the archived historic file 833y-fsy8 (to 2024) used to check the counts match. Weather: GHCN-Daily USW00094728 (Central Park) TMAX and PRCP via concord/noaa.py. Primary model: negative binomial GLM (log link, dispersion by ML) of daily incidents on linear TMAX/5, factors for day of week, calendar month, year, an indicator for federal holidays plus 4 July and 31 December, and an indicator PRCP >= 1 mm. Report IRR per 5 C with 95% CI (robust HC0 SE). Curve: same model with a natural cubic spline of TMAX (4 df), RR vs the median TMAX; read off the RR at the 84th percentile for comparison with Lyons et al. Sensitivity, pre-specified: (a) TMAX averaged over lags 0-1; (b) TMIN instead of TMAX; (c) warm season May-September only; (d) 2006-2019 vs 2020-present; (e) murders only (statistical_murder_flag, historic file). Comparison: literature estimates converted to IRR per 5 C where possible (Reeping weekday and weekend; Xu NYC homicide; Rahman CA homicide; corrected graph values).",
"prediction": {
"direction": "+",
"magnitude": "IRR per 5 C of Central Park TMAX (lag 0) between 1.05 and 1.20, with month, year and day-of-week terms. The spline curve rises monotonically from the coldest days to about the 85th-95th percentile; RR at the city's 84th percentile TMAX vs the median between 1.05 and 1.25 (Lyons et al. pooled 1.17).",
"refute_if": "upper 95% limit at or below 1.00 (contested); an estimate above 1.00 with a limit below 1.00 is unsettled",
"support_if": "lower 95% limit of the IRR per 5 C above 1.00 (confirmed); hit if the estimate lies in 1.05-1.20"
},
"prior": 0.9,
"registered_at": "2026-11-12T15:19:38-0500",
"registrant": "Claude (heat agent), for Jeff Vernon",
"sha256": "b2a1bd8ecec550da1e1e07d368d2e5f1b7d108b07ce6f51ef658bbb197ebb762",
"statement": "In New York City, the daily number of shooting incidents rises with the day's maximum temperature at Central Park, within month and year, by about 5 to 20 percent per 5 C."
},
"registration_commit": {
"commit": "67507946fe78b52a39097d7882340ebe335aadad",
"committed_at": "2026-11-12T15:19:38-05:00"
},
"registration_file_sha256": "f3c9ef53efa7dc3d9904edcb9d6d17f300dabf7194a6629cf3bc037b8bebac52",
"registration_path": "claims/registered/heat-shootings.json",
"registration_sha256": "b2a1bd8ecec550da1e1e07d368d2e5f1b7d108b07ce6f51ef658bbb197ebb762",
"registration_text": "{\n \"id\": \"heat-shootings\",\n \"kind\": \"confirmation\",\n \"statement\": \"In New York City, the daily number of shooting incidents rises with the day's maximum temperature at Central Park, within month and year, by about 5 to 20 percent per 5 C.\",\n \"graph_evidence\": [\n {\n \"relationship_id\": \"4FaXcuNJRt|YZku29igbl\",\n \"concepts\": [\n \"temperature\",\n \"homicide\"\n ],\n \"n_findings\": 4,\n \"n_studies\": 2,\n \"direction\": \"+\",\n \"summary\": \"Two case-crossover studies. Rahman et al. 2023 (California 2014-2019): homicide odds +4.9% (1.6 to 8.1) per 5 C Tmax lag 0. Xu et al. 2020 (9 US cities 2007-2017): intentional homicide +9.5% (4.3 to 15.0) per 5 C mean temperature in Chicago and +8.8% (1.5 to 16.6) in New York, lag 0-7. The graph stores Xu's estimates as odds ratios 0.095 and 0.088 and Rahman's as odds ratios 4.9 and 6.2; the abstracts give percent changes, so the graph values are mis-extracted (true ORs about 1.095, 1.088, 1.049, 1.062).\",\n \"pubmed_ids\": [\n 36858215,\n 32738768\n ]\n },\n {\n \"relationship_id\": \"4FaXcuNJRt|rULjCt1BqH\",\n \"concepts\": [\n \"temperature\",\n \"injury\"\n ],\n \"n_findings\": 4,\n \"n_studies\": 3,\n \"direction\": \"+\",\n \"summary\": \"Girma et al. 2023, New York City ED visits 2005-2011 (SPARCS), warm season: intentional injury among 20-25 year olds OR 1.54 (1.28 to 1.85) for daily minimum 77 F vs 48 F, lag 0-5. The only NYC violence-temperature finding in the graph.\",\n \"pubmed_ids\": [\n 37448837\n ]\n },\n {\n \"relationship_id\": \"(concept X7hfhwez9f, 34 relationships)\",\n \"concepts\": [\n \"gun violence\",\n \"temperature\"\n ],\n \"n_findings\": 0,\n \"n_studies\": 0,\n \"direction\": \"0\",\n \"summary\": \"The gun violence concept has no relationship with temperature, air temperature, heat wave, hot or extreme heat. Violence (2qLKgaxkm_), crime, aggression and violent crime have none either. The graph's heat-violence evidence is homicide and injury only.\",\n \"pubmed_ids\": []\n },\n {\n \"relationship_id\": \"4FaXcuNJRt|TLeQZQ_leT\",\n \"concepts\": [\n \"temperature\",\n \"suicide\"\n ],\n \"n_findings\": 12,\n \"n_studies\": 8,\n \"direction\": \"+\",\n \"summary\": \"Self-directed violence rises with temperature in 8 studies (e.g. RR 1.35, 1.26 to 1.45, PMID 35394808; per 1 C IRR 1.008, PMID 34034696). Context for the violence claim.\",\n \"pubmed_ids\": [\n 35394808,\n 34034696,\n 36858215\n ]\n }\n ],\n \"literature\": [\n {\n \"citation\": \"Reeping PM, Hemenway D. The association between weather and the number of daily shootings in Chicago (2012-2016). Inj Epidemiol 2020;7:31.\",\n \"doi\": \"10.1186/s40621-020-00260-3\",\n \"finding\": \"Negative binomial GEE with month and year fixed effects: +34.0% (28.1 to 40.2) shootings per 10 C daily high on weekdays, +41.6% (34.7 to 48.9) on weekends and holidays; per 5 C that is IRR 1.158 and 1.190.\"\n },\n {\n \"citation\": \"Lyons VH, Gause EL, Spangler KR, Wellenius GA, Jay J. Analysis of daily ambient temperature and firearm violence in 100 US cities. JAMA Netw Open 2022;5(12):e2247207.\",\n \"doi\": \"10.1001/jamanetworkopen.2022.47207\",\n \"finding\": \"Gun Violence Archive 2015-2020, distributed-lag models, lag 0 best: RR 1.17 (1.12 to 1.21) at the 84th percentile of Tmax vs the city median; 6.85% of shootings attributable to days above the median; Northeast attributable fraction 9.54%.\"\n },\n {\n \"citation\": \"Xu R, Xiong X, Abramson MJ, Li S, Guo Y. Ambient temperature and intentional homicide: a multi-city case-crossover study in the US. Environ Int 2020;143:105992.\",\n \"doi\": \"10.1016/j.envint.2020.105992\",\n \"finding\": \"New York 2007-2017: +8.8% (1.5 to 16.6) intentional homicide per 5 C daily mean temperature, lag 0-7; 7.1% attributable to temperatures above the median.\"\n },\n {\n \"citation\": \"Rahman MM et al. Ambient temperature and air pollution associations with suicide and homicide mortality in California. Sci Total Environ 2023;874:162462.\",\n \"doi\": \"10.1016/j.scitotenv.2023.162462\",\n \"finding\": \"+4.9% (1.6 to 8.1) homicide per 5 C Tmax lag 0.\"\n }\n ],\n \"prediction\": {\n \"direction\": \"+\",\n \"magnitude\": \"IRR per 5 C of Central Park TMAX (lag 0) between 1.05 and 1.20, with month, year and day-of-week terms. The spline curve rises monotonically from the coldest days to about the 85th-95th percentile; RR at the city's 84th percentile TMAX vs the median between 1.05 and 1.25 (Lyons et al. pooled 1.17).\",\n \"support_if\": \"lower 95% limit of the IRR per 5 C above 1.00 (confirmed); hit if the estimate lies in 1.05-1.20\",\n \"refute_if\": \"upper 95% limit at or below 1.00 (contested); an estimate above 1.00 with a limit below 1.00 is unsettled\"\n },\n \"plan\": \"Data: NYPD Shooting Incident Data, 5ucz-vwe8 (Shootings 2006-Present) as primary, count of distinct incident_key per occur_date, 2006-01-01 to the last full month; the archived historic file 833y-fsy8 (to 2024) used to check the counts match. Weather: GHCN-Daily USW00094728 (Central Park) TMAX and PRCP via concord/noaa.py. Primary model: negative binomial GLM (log link, dispersion by ML) of daily incidents on linear TMAX/5, factors for day of week, calendar month, year, an indicator for federal holidays plus 4 July and 31 December, and an indicator PRCP >= 1 mm. Report IRR per 5 C with 95% CI (robust HC0 SE). Curve: same model with a natural cubic spline of TMAX (4 df), RR vs the median TMAX; read off the RR at the 84th percentile for comparison with Lyons et al. Sensitivity, pre-specified: (a) TMAX averaged over lags 0-1; (b) TMIN instead of TMAX; (c) warm season May-September only; (d) 2006-2019 vs 2020-present; (e) murders only (statistical_murder_flag, historic file). Comparison: literature estimates converted to IRR per 5 C where possible (Reeping weekday and weekend; Xu NYC homicide; Rahman CA homicide; corrected graph values).\",\n \"alternatives\": [\n \"Residual seasonality within month (school calendar, summer youth programmes, daylight) correlated with temperature\",\n \"Time outdoors and street encounters as the mechanism (the temperature effect is real but works through routine activity, not physiology); the design cannot separate mechanisms\",\n \"Police summer deployments (e.g. Summer All Out) that respond to expected violence\",\n \"The 2020-2021 rise in shootings coinciding with summers; handled by year terms but not by year-by-month\"\n ],\n \"prior\": 0.9,\n \"registrant\": \"Claude (heat agent), for Jeff Vernon\",\n \"registered_at\": \"2026-11-12T15:19:38-0500\",\n \"sha256\": \"b2a1bd8ecec550da1e1e07d368d2e5f1b7d108b07ce6f51ef658bbb197ebb762\"\n}",
"replication": {
"built_at": "2026-11-13T22:56:48-0500",
"checked_at": "2026-11-13T22:56:48-0500",
"decides": "support_if: lower 95% limit above 1.00",
"deciding_number": "city_effect: IRR per 5 C of Central Park TMAX (lag 0), negative binomial GLM, robust HC0 95% CI",
"inputs_copied": 19,
"manifest": "replication/heat-shootings/manifest.json",
"manifest_sha256": "a566f0ed212e06867e3b5d351b4bdbef0b8189567fcae4088b0df66d4eff0741",
"package": "replication/heat-shootings/",
"recorded": {
"estimate": 1.1462159704993367,
"hi": 1.162930031787583,
"lo": 1.1297421298925683,
"n": 7548
},
"status": "MATCH",
"status_from": "replication/heat-shootings/verify_result.json",
"tables_not_rebuilt": 0,
"tolerance": {
"estimate": 0.0001,
"hi": 0.0001,
"lo": 0.0001,
"n": 0
}
},
"statement": "In New York City, the daily number of shooting incidents rises with the day's maximum temperature at Central Park, within month and year, by about 5 to 20 percent per 5 C."
}History in this collection
- addedv1.0.0Nov 14, 2026this version