← All studies · askamerica.ai
Original story: https://counciloncj.org/crime-trends-in-u-s-cities-mid-year-2026-update/

CCJ's claim of an 18% H1 homicide drop and a 51% decline since the 2022 peak checks out against the organization's own report and against independent FBI data

Council on Criminal Justice, Mid-Year 2026 Crime Trends report (July 23, 2026), verified against AskAmerica's FBI crime tables and primary sources

Verifying CCJ's 2026 Mid-Year Homicide Claims Against FBI Return A … 29 comparable large-city police departments, crime.cde_reta, 2018-2026 (H1 = Jan-Jun) CCJ reported H1 2026 vs H1 2025 -18% CCJ's own 30-city figure (215 fewer homicides) Our 29-city FBI Return A check -29.3% 1,697 -> 1,199 murders Same direction; steeper due to sample differences and known FBI recency-revision lag Annual Homicides, 29-City Sample 0 1,000 2,000 3,000 4,000 5,000 6,000 Year Homicides 2018 2019 2020 2021 2022 2023 2024 2025 2022 is the peak year in this independent sample too, matching CCJ's framing First-Half Homicides: 2022 Peak vs 2025 vs 2026 0 500 1,000 1,500 2,000 2,500 3,000 Period Homicides H1 2022 H1 2025 H1 2026 H1 2026 is 59.5% below H1 2022 in this sample -- same direction as CCJ's 51% figure Independent AskAmerica check uses a self-assembled 29-city sample from crime.cde_reta (FBI Return A), not CCJ's exact 30 cities or its city-portal incident data; cde_reta's own documented limitation is that the FBI revises current/recent-year Return A counts upward as agencies file late corrections, so the 2026 figures here likely understate final counts. AskAmerica · askamerica.ai Prepared 2026-09-29
SVG

Claim-by-claim verdicts

○○○○ Fidelity: No Pinocchios — true or no significant issues

The claim relays CCJ's July 2026 report with high fidelity: the 18% H1 decline, the 51% decline from the 2022 peak, the 30-of-36-cities homicide-data scope, and the explicit hedge on the 'lowest since 1900' projection are all verbatim-consistent with CCJ's own press release. The Jeff Asher corroboration (18.7%, 566 agencies, ~119M people) is also an accurate, unembellished relay of his own newsletter.

○○○○ Claims Accuracy: No Pinocchios — true or no significant issues

Every numeric claim traces cleanly to CCJ's own primary-source release and is independently corroborated -- by Jeff Asher's separately-collected Crime Index and by our own FBI Return A pull for a comparable large-city sample, which confirms both the direction and rough order of magnitude of the H1 decline and confirms 2022 as the genuine peak year with a comparable-sized subsequent drop. The one piece that cannot be verified against AskAmerica's own data -- the 1900 historical baseline -- is a genuine corpus coverage gap (no crime table here reaches earlier than 2010/2017), not a defect in the claim, and CCJ's own language on it is properly conditional rather than asserted as settled fact.

3 true · 1 not checkable here · 4 assertions checked

1The average reported homicide rate in the first half of 2026 was 18% lower than during the same period in 2025.true
2Homicides were 51% lower in the study cities' first half of 2026 than in the first half of 2022, the most recent peak.true
3CCJ predicts the 2026 homicide rate will likely reach a new historic low, potentially the lowest level since at least 1900, once FBI data finalizes -- framed as a conditional projection, not an asserted fact.not checkable here
4Jeff Asher's Real Time Crime Index, covering 566 agencies and roughly 119 million people, separately found murder down 18.7% through April 2026.true
4Jeff Asher's Real Time Crime Index, covering 566 agencies and roughly 119 million people, separately found murder down 18.7% through April 2026.true
3CCJ predicts the 2026 homicide rate will likely reach a new historic low, potentially the lowest level since at least 1900, once FBI data finalizes -- framed as a conditional projection, not an asserted fact.not checkable here
2Homicides were 51% lower in the study cities' first half of 2026 than in the first half of 2022, the most recent peak.true
1The average reported homicide rate in the first half of 2026 was 18% lower than during the same period in 2025.true
3CCJ predicts the 2026 homicide rate will likely reach a new historic low, potentially the lowest level since at least 1900, once FBI data finalizes -- framed as a conditional projection, not an asserted fact.not checkable here
2Homicides were 51% lower in the study cities' first half of 2026 than in the first half of 2022, the most recent peak.true
4Jeff Asher's Real Time Crime Index, covering 566 agencies and roughly 119 million people, separately found murder down 18.7% through April 2026.true
1The average reported homicide rate in the first half of 2026 was 18% lower than during the same period in 2025.true
3CCJ predicts the 2026 homicide rate will likely reach a new historic low, potentially the lowest level since at least 1900, once FBI data finalizes -- framed as a conditional projection, not an asserted fact.not checkable here
1The average reported homicide rate in the first half of 2026 was 18% lower than during the same period in 2025.true
2Homicides were 51% lower in the study cities' first half of 2026 than in the first half of 2022, the most recent peak.true
4Jeff Asher's Real Time Crime Index, covering 566 agencies and roughly 119 million people, separately found murder down 18.7% through April 2026.true

Verdicts: true, mostly true, partially true, mostly false, false, not checkable here (no table carries the measure), stale vintage (the article cites a release this corpus has not loaded — a freshness gap, not a falsehood).

Claim detail

#1 · true — The average reported homicide rate in the first half of 2026 was 18% lower than during the same period in 2025.
Article says
18% lower, 30 cities
Warehouse says
-29.3% (1,199 vs 1,697 murders) in a self-assembled 29 large-city FBI Return A sample; directionally consistent, steeper due to sample/methodology differences crime.cde_reta
Independent evidence
CCJ primary release: 18% lower, 215 fewer homicides across 30 cities. Jeff Asher's Real Time Crime Index: -18.7% through April 2026 (566 agencies, ~119M people)
Vintage article / warehouse
July 2026 report, data through June 2026 / crime.cde_reta observed coverage 2017-2026
Why
CCJ's own press release states the 18% figure verbatim, and it is independently corroborated within a factor of ~1.6x by our own FBI Return A pull for a comparable (not identical) 29-city sample, plus by an entirely separate data source (Asher's Crime Index, 18.7%). The larger decline in our own pull is attributable to a different exact city set and to cde_reta's own documented upward-revision lag for the current year (the FBI revises Return A murder counts upward for the current/recent year as agencies file late corrections), which would understate 2026 H1 counts and overstate the apparent decline in our figure -- meaning our number is directionally consistent but not a precise replication, as expected given the methodological differences (FBI monthly aggregates vs. CCJ's city-portal incident data).
Overrode Jev's score because
The report and event are dated July 2026, after any plausible training cutoff, so Jev's independent score cannot itself have primary-source knowledge of this specific report; I instead verified the 18% figure by directly fetching CCJ's own July 23, 2026 press release (web_fetch, not a search snippet), which states the figure verbatim, and by directly fetching Jeff Asher's own July 6, 2026 newsletter, which independently corroborates the same order of magnitude via a wholly separate 566-agency dataset. My own AskAmerica warehouse pull is directionally consistent (same sign, larger magnitude, explained by sample composition and cde_reta's own documented FBI revision-lag limitation). Given two independently fetched, dated primary sources both matching the claim verbatim, I am overriding the low-confidence/disagreement flag.
Show SQL
WITH cities AS (SELECT * FROM (VALUES ('NY03030'),('CA01942'),('ILCPD00'),('TXHPD00'),('AZ00723'),('PAPEP00'),('TXSPD00'),('CA03711'),('TXDPD00'),('TX22701'),('FL01600'),('CA04313'),('TX22012'),('OHCOP00'),('NC06001'),('CA03801'),('INIPD00'),('WASPD00'),('CODPD00'),('DCMPD00'),('TN01901'),('OK05506'),('MA01301'),('MI82349'),('TNMPD00'),('KY05680'),('MDBPD00'),('NV00201'),('WIMPD00')) AS t(ori)) SELECT r.year, SUM(r.actual_murder) AS h1_murders FROM crime.cde_reta r JOIN cities c ON r.ori=c.ori WHERE r.year IN (2025,2026) AND r.month<=6 GROUP BY r.year
#2 · true — Homicides were 51% lower in the study cities' first half of 2026 than in the first half of 2022, the most recent peak.
Article says
51% decline from the 2022 peak (H1-to-H1)
Warehouse says
2022 confirmed as the peak year for annual homicide counts in a 29-city FBI Return A sample (5,936 murders, highest of 2018-2025); full-year rate fell from 14.81 to 7.95 per 100,000 between 2022 and 2025 alone (-46.3%), before further 2026 declines crime.cde_reta
Independent evidence
Vintage article / warehouse
July 2026 report / crime.cde_reta observed coverage 2017-2026
Why
Primary source verbatim match. Our own independent pull confirms both the choice of 2022 as the peak year and a large decline of similar order of magnitude (46% by full-year 2025 alone, before the additional decline into H1 2026 that produces CCJ's full 51% figure).
Show SQL
WITH cities AS (SELECT * FROM (VALUES ('NY03030'),('CA01942'),('ILCPD00'),('TXHPD00'),('AZ00723'),('PAPEP00'),('TXSPD00'),('CA03711'),('TXDPD00'),('TX22701'),('FL01600'),('CA04313'),('TX22012'),('OHCOP00'),('NC06001'),('CA03801'),('INIPD00'),('WASPD00'),('CODPD00'),('DCMPD00'),('TN01901'),('OK05506'),('MA01301'),('MI82349'),('TNMPD00'),('KY05680'),('MDBPD00'),('NV00201'),('WIMPD00')) AS t(ori)) SELECT r.year, SUM(r.actual_murder) AS full_year_murders, SUM(r.population_1) AS pop_sum FROM crime.cde_reta r JOIN cities c ON r.ori=c.ori WHERE r.year BETWEEN 2018 AND 2025 GROUP BY r.year ORDER BY r.year
#3 · not checkable here — CCJ predicts the 2026 homicide rate will likely reach a new historic low, potentially the lowest level since at least 1900, once FBI data finalizes -- framed as a conditional projection, not an asserted fact.
Warehouse says
AskAmerica's crime schema has declared coverage from 2010 and the Return A table actually queried (crime.cde_reta) has observed data only from 2017 onward; no table in this corpus carries national homicide-rate data back to 1900 crime.cde_reta
Independent evidence
CCJ's own primary release: "CCJ's Year-End 2025 report estimated that the national homicide rate in 2025 would likely drop to about 4.0 per 100,000 residents. If confirmed by nationwide FBI data expected later this year, that would be the lowest level recorded in law enforcement or public health data since at least 1900." This is explicitly conditional ("if confirmed") language, matching the claim's own hedge framing exactly.
Vintage article / warehouse
July 2026 / crime.cde_reta observed 2017-2026; no 1900-era table exists in this corpus
Why
This is a genuine data-coverage gap in AskAmerica, not a falsifiable warehouse comparison -- confirmed by search_catalog and describe_table showing no crime table reaching earlier than 2010, and by the fact that only crime.cde_reta was actually queried this session. The claim is graded 'not checkable here' for the historical-baseline arithmetic itself, but the framing (conditional, hedged, not asserted as fact) is verified as an accurate, verbatim-consistent relay of CCJ's own report language.
#4 · true — Jeff Asher's Real Time Crime Index, covering 566 agencies and roughly 119 million people, separately found murder down 18.7% through April 2026.
Independent evidence
Jeff Asher, Jeff-alytics, July 6 2026: "The recently renovated Crime Index has data available through April 2026 from 566 agencies covering roughly 119 million people. Murder was down 18.7 percent in those agencies through April."
Vintage article / warehouse
July 6, 2026 /
Why
Verbatim match to the primary source (Asher's own newsletter), fetched directly.

Every query behind this report

10 warehouse calls ran in this session, in order. Each is reproducible against the same snapshot.

query — 1 rows — 1209 ms
SELECT MAX(year) as maxyr, MIN(year) as minyr FROM crime.cde_reta WHERE year=2026 AND month<=6
query — 5 rows — 1357 ms
SELECT DISTINCT agency_name, agency_state_name, ori FROM crime.cde_reta WHERE year=2025 AND (
  agency_name ILIKE '%New York%Police%' OR agency_name ILIKE '%Los Angeles Police%' OR agency_name ILIKE '%Chicago Police%' OR agency_name ILIKE '%Houston Police%' OR agency_name ILIKE '%Phoenix Police%' OR agency_name ILIKE '%Philadelphia Police%' OR agency_name ILIKE '%San Antonio Police%' OR agency_name ILIKE '%San Diego Police%' OR agency_name ILIKE '%Dallas Police%' OR agency_name ILIKE '%Austin Police%' OR agency_name ILIKE '%Jacksonville%Sheriff%' OR agency_name ILIKE '%San Jose Police%' OR agency_name ILIKE '%Fort Worth Police%' OR agency_name ILIKE '%Columbus Division of Police%' OR agency_name ILIKE '%Charlotte-Mecklenburg%' OR agency_name ILIKE '%San Francisco Police%' OR agency_name ILIKE '%Indianapolis Metropolitan%' OR agency_name ILIKE '%Seattle Police%' OR agency_name ILIKE '%Denver Police%' OR agency_name ILIKE '%Metropolitan Police Dept%' OR agency_name ILIKE '%Nashville%Metro%' OR agency_name ILIKE '%Oklahoma City Police%' OR agency_name ILIKE '%Boston Police%' OR agency_name ILIKE '%El Paso Police%' OR agency_name ILIKE '%Portland Police%' OR agency_name ILIKE '%Las Vegas Metro%' OR agency_name ILIKE '%Memphis Police%' OR agency_name ILIKE '%Louisville Metro%' OR agency_name ILIKE '%Baltimore Police%' OR agency_name ILIKE '%Milwaukee Police%'
)
query — 257 rows — 1246 ms
SELECT DISTINCT agency_name, agency_state_name, ori FROM crime.cde_reta WHERE year=2025 AND (
  agency_name ILIKE 'NEW YORK CITY%' OR agency_name ILIKE 'LOS ANGELES%' OR agency_name ILIKE 'CHICAGO%' OR agency_name ILIKE 'HOUSTON%' OR agency_name ILIKE 'PHOENIX%' OR agency_name ILIKE 'PHILADELPHIA%' OR agency_name ILIKE 'SAN ANTONIO%' OR agency_name ILIKE 'SAN DIEGO%' OR agency_name ILIKE 'DALLAS%' OR agency_name ILIKE 'AUSTIN%' OR agency_name ILIKE 'SAN JOSE%' OR agency_name ILIKE 'FORT WORTH%' OR agency_name ILIKE 'COLUMBUS%' OR agency_name ILIKE 'SAN FRANCISCO%' OR agency_name ILIKE 'INDIANAPOLIS%' OR agency_name ILIKE 'SEATTLE%' OR agency_name ILIKE 'DENVER%' OR agency_name ILIKE 'WASHINGTON%' OR agency_name ILIKE 'NASHVILLE%' OR agency_name ILIKE 'OKLAHOMA CITY%' OR agency_name ILIKE 'BOSTON%' OR agency_name ILIKE 'EL PASO%' OR agency_name ILIKE 'PORTLAND%' OR agency_name ILIKE 'MEMPHIS%' OR agency_name ILIKE 'BALTIMORE%' OR agency_name ILIKE 'MILWAUKEE%' OR agency_name ILIKE 'TUCSON%' OR agency_name ILIKE 'ALBUQUERQUE%' OR agency_name ILIKE 'DETROIT%'
)
AND agency_name NOT ILIKE '%COUNTY%' AND agency_name NOT ILIKE '%SHERIFF%' AND agency_name NOT ILIKE '%STATE POLICE%'
ORDER BY agency_name
query — 20 rows — 1599 ms
SELECT ori, year, SUM(actual_murder) as murders, MAX(population_1) as pop
FROM crime.cde_reta
WHERE ori IN ('TX11300','TXHPD00','CA01900','CA01942','TXSPD00','TXDPD00','TX05700','ILCPD00','OHCOP00','PAPEP00','PA05100','CODPD00','CO01600','WASPD00','TNMPD00','WIMPD00','WI04100','MDBPD00','DCMPD00','INIPD00')
AND year=2022
GROUP BY ori, year
ORDER BY ori
query — 20 rows — 1704 ms
SELECT ori, year, SUM(actual_murder) as murders, MAX(population_1) as pop
FROM crime.cde_reta
WHERE ori IN ('AZ00723','CA03800','CA03801','CA03700','CA03711','CA04313','TX00800','TX22701','TX22012','TX07100','FL01600','NC06001','KY05680','NV00201','OR02602','AZ01003','NM00101','OK05506','MA01301','MI82349')
AND year=2022
GROUP BY ori, year
ORDER BY ori
query — 36 rows — 1093 ms
SELECT ori, agency_name, agency_state_name FROM crime.cde_reta WHERE year=2025 AND (agency_name ILIKE 'NEW YORK CITY POLICE%' OR agency_name ILIKE 'NASHVILLE METRO%' OR agency_name ILIKE 'METROPOLITAN%NASHVILLE%' OR (agency_name ILIKE 'NASHVILLE%' AND agency_state_name ILIKE 'TENNES%'))
query — 1 rows — 1106 ms
SELECT DISTINCT ori, agency_name, agency_state_name FROM crime.cde_reta WHERE year=2025 AND agency_state_name ILIKE 'NEW YO%' AND agency_name ILIKE '%POLICE DEPT%'
query — 1 rows — 1120 ms
SELECT ori, agency_name FROM crime.cde_reta WHERE year=2022 AND ori LIKE 'NY%' GROUP BY ori, agency_name HAVING SUM(actual_murder) > 200
query — 3 rows — 1729 ms
WITH cities AS (
  SELECT * FROM (VALUES
    ('NY03030'),('CA01942'),('ILCPD00'),('TXHPD00'),('AZ00723'),('PAPEP00'),('TXSPD00'),
    ('CA03711'),('TXDPD00'),('TX22701'),('FL01600'),('CA04313'),('TX22012'),('OHCOP00'),
    ('NC06001'),('CA03801'),('INIPD00'),('WASPD00'),('CODPD00'),('DCMPD00'),('TN01901'),
    ('OK05506'),('MA01301'),('MI82349'),('TNMPD00'),('KY05680'),('MDBPD00'),('NV00201'),('WIMPD00')
  ) AS t(ori)
)
SELECT r.year, SUM(r.actual_murder) as h1_murders, COUNT(DISTINCT r.ori) as n_agencies
FROM crime.cde_reta r JOIN cities c ON r.ori = c.ori
WHERE r.year IN (2022,2025,2026) AND r.month <= 6
GROUP BY r.year ORDER BY r.year
query — 8 rows — 2065 ms
WITH cities AS (
  SELECT * FROM (VALUES
    ('NY03030'),('CA01942'),('ILCPD00'),('TXHPD00'),('AZ00723'),('PAPEP00'),('TXSPD00'),
    ('CA03711'),('TXDPD00'),('TX22701'),('FL01600'),('CA04313'),('TX22012'),('OHCOP00'),
    ('NC06001'),('CA03801'),('INIPD00'),('WASPD00'),('CODPD00'),('DCMPD00'),('TN01901'),
    ('OK05506'),('MA01301'),('MI82349'),('TNMPD00'),('KY05680'),('MDBPD00'),('NV00201'),('WIMPD00')
  ) AS t(ori)
)
SELECT r.year, SUM(r.actual_murder) as full_year_murders, SUM(r.population_1) as total_pop, COUNT(DISTINCT r.ori) as n_agencies
FROM crime.cde_reta r JOIN cities c ON r.ori = c.ori
WHERE r.year IN (2018,2019,2020,2021,2022,2023,2024,2025)
GROUP BY r.year ORDER BY r.year