LA Crime Data (2020 to Present) vs National Crime Victimization Survey (NCVS)

Geography at the block. Each of the 1,004,894 rows carries latitude and longitude, a street address rounded to the nearest hundred block with its cross street, one of 21 area designations and a four-digit reporting district. The survey publishes no point locations whatsoever, which makes this file the default instrument for anything spatial in one metro.

Evidence texture on every row. Offense codes ride alongside modus operandi codes, premise type, weapon description, victim age, sex and descent, Part 1 or Part 2 severity, and case status. Two sampled incidents show the grain: DR_NO 211507896, a theft of identity logged at 0845 in North Hollywood's reporting district 1502, victim age 31, premises a single family dwelling; DR_NO 201516622, an assault with a deadly weapon on a sidewalk at 1845, eight MO codes attached, the weapon recorded as a knife with a blade of six inches or less.

Severity and outcome in one file. The Part 1/Part 2 split separates serious offenses from the remainder, and the status field carries each case toward an outcome - texture a victimization rate structurally cannot reproduce.

It counts what the map cannot see. Interviews capture victimization regardless of reporting, and a dedicated variable records whether police were told. Every row in the Los Angeles file exists because an officer wrote it up; the survey measures the incidents that never produced a row. For household security work it is the only national source quantifying burglary victimization rates and the protective measures households actually take - see burglary offense category.

Half a century of trend. The series runs from 1973 through the latest data year, 2024, with trend estimates typically computed from the 1993-forward files; the Los Angeles window spans 2020 through 2024 and closes there. Five dense years of point events against five decades of household interviews - see the National Crime Victimization Survey entry for the program's anatomy.

A sample built for projection. Roughly 240,000 persons age 12 or older in about 150,000 households, interviewed twice a year across a seven-interview panel - about 109,000 personal interviews completed in 2024 - with person, household and incident weights turning responses into national numbers.

Instrument evolution on the record. In 2024 the program ran a split-sample design pairing the legacy questionnaire with a redesigned instrument whose estimates arrive separately - methodological care a police log, by construction, does not attempt.