social_context_incarceration_county_year
County × year jail and prison incarceration rates (per 100,000 residents aged 15–64), from the Vera Institute Incarceration Trends dataset (sourced from BJS: COJ, ASJ, NCRP, NPS), 2010 onward. Jail population extends to 2024; county-level PRISON population ends at 2019 and is absent entirely for states where NCRP does not report county-of-commitment. jail_coverage and prison_coverage are tracked separately and a missing source value is flagged UNSUPPORTED with a NULL rate — never imputed. total_incarceration_rate (jail + prison) is populated only where both components exist (mostly ≤ 2019). Source 4 of the P1 social-context (SDOH) cluster. Tier C.
overview
Jail rate populated through 2024; county prison rate only through 2019 and only for NCRP-reporting states (~2,600 of ~3,200 counties in covered years). jail_coverage / prison_coverage mark every cell SUPPORTED or UNSUPPORTED; the rate is NULL where UNSUPPORTED. Confidently imputing a prison rate in a non-reporting county is the failure this design prevents.
current vintage — Vera Incarceration Trends (2026-03 release); jail to 2024, county prison to 2019
history — Vera county panel spans 1970–2024 (jail) and 1983–2019 (county prison); exposed here from 2010 onward.
source & licensing
fields
| name | type | definition |
|---|---|---|
| country_idkey | string | ISO alpha-2 country code (always 'US' for domestic tables).Part of primary key. |
| county_idkey | string | 5-character FIPS code identifying the county.Part of primary key. Joins dim.counties on county_id. |
| is_regional_jail | int64 | 1 where the county's jail population is parceled from a shared regional facility by resident-population share (an allocation assumption); 0 otherwise. |
| jail_coverage | string | Per-cell coverage flag (context/guides/coverage-flags.md): SUPPORTED or UNSUPPORTED (jail count unavailable for this county-year). |
| jail_pop | float64 | Average daily jail population (fractional where interpolated). NULL where unavailable. |
| jail_pop_rate | float64 | Jail population per 100,000 residents aged 15–64. NULL where jail_coverage = UNSUPPORTED. |
| pop_15to64 | int64 | County population aged 15–64 (the rate denominator Vera uses). |
| prison_coverage | string | Per-cell coverage flag: SUPPORTED or UNSUPPORTED. UNSUPPORTED for NCRP-non-reporting states and all years after 2019. Do NOT impute. |
| prison_pop | float64 | County prison population (NCRP-derived allocation). NULL after 2019 and in NCRP-non-reporting states. |
| prison_pop_rate | float64 | Prison population per 100,000 residents aged 15–64. NULL where prison_coverage = UNSUPPORTED. |
| state_idkey | string | 2-character FIPS code identifying the state.Part of primary key. Joins dim.states on state_id. |
| total_incarceration_rate | float64 | Combined jail + prison rate per 100,000 aged 15–64. Populated only where both components exist (mostly ≤ 2019). |
| yearkey | int64 | Reference year of the observation.Part of primary key. |
joins
how to use this table
Vera-computed rates per 100,000 residents aged 15–64, carried through as-is from incarceration_trends_county.csv. jail_coverage = UNSUPPORTED where total_jail_pop is null; prison_coverage = UNSUPPORTED where total_prison_pop is null. total_incarceration_rate is Vera's jail+prison combined rate, present only where both components exist.
Current-year prison incarceration (county prison ends 2019). Any county-year where prison_coverage = UNSUPPORTED (NCRP-non-reporting states) — do not impute. Individual inference (this is an area-level rate). Cross-county comparison without accounting for the regional-jail allocation (is_regional_jail) and the ~1/3 jail sampling.
County prison data is not available in every state (NCRP-absent → UNSUPPORTED) and ends at 2019. Jail data leans on the ~1/3 Annual Survey of Jails with interpolation. Regional-jail populations are parceled across member counties by resident-population share. Coverage tracks region and rurality. Vera county_fips use pre-2022 county definitions (Connecticut old counties), which may not all resolve against current dim.geographies planning regions.