Rural Dataverse

American Community Survey (ACS)

Annual demographic, economic, social, and housing survey of ~3.5 million U.S. households. A primary federal source for community-level characteristics not captured by administrative data — educational attainment, income, housing costs, work-from-home rates, self-employment, and more.

DatasetDemographicsLabor MarketHousing

Mapping & Data Analytics Team 2026-09-02

Overview

The American Community Survey is the Census Bureau’s annual nationwide survey of U.S. households, reaching roughly 3.5 million addresses each year with a mandatory questionnaire covering demographics, housing, economics, and social characteristics. The ACS contributes a people-and-households layer to rural analysis that administrative data cannot. Where datasets like QCEW and BDS describe jobs, establishments, employment, and wages, ACS adds the characteristics of the people and households behind those measures: education, income, occupation, housing, commuting patterns, internet access, and work-from-home rates. For any analysis that asks not just “how many jobs” but “who holds them, at what pay, and under what conditions,” ACS is where the answer lives.

The survey produces two main products:

  • 1-year estimates cover geographies with populations of 65,000 or more and are released each fall for the prior calendar year — current, but limited in geographic reach.

  • 5-year estimates aggregate five years of survey data and are available for all geographies including small counties, census tracts, block groups, and places — making them the essential product for rural analysis, where most counties fall below the 1-year threshold. The trade-off is currency: 5-year estimates represent a rolling average, not a snapshot of a single year.

What this data tells us about rural America

The ACS provides the broadest cross-section of socioeconomic conditions at the county level — covering educational attainment, income, employment by occupation, housing costs, work-from-home rates, self-employment, and race and ethnicity. It is the primary federal source for the demographic and household-level context that labor market and business data alone cannot supply.

For rural analysis, ACS is particularly valuable for measures that reflect the quality and accessibility of economic opportunity, not just its quantity. For instance, educational attainment shapes the workforce available to attract and grow businesses; self-employment helps describe the state of rural entrepreneurship; and, the share of remote workers in tradable services describes the nuance of rural knowledge work. These are dimensions of economic conditions that QCEW and BDS cannot see.

The ACS can be accessed through tidycensus, an easy-to-use R package developed and maintained by Kyle Walker that provides a clean interface to the Census Data API. Because an excellent tool already exists, CORI does not maintain a separate ACS package. The Data Verse surfaces ACS as a key rural data resource because understanding what it measures and how to use it well in a rural context is part of what this resource is for.

Coverage

  • Geography:
    • Nation, State, County, Census Tract, Block Group, Place, ZCTA, and more
  • Time range:
    • 1-year estimates: 2005–present
    • 5-year estimates: 2009–present
  • Update frequency:
    • 1-year estimates typically released in September
    • 5-year estimates typically in December
  • Universe: Varies by table; most tables cover the U.S. civilian noninstitutionalized population. Group quarters population is included for many subjects.
NOTE

1-year vs. 5-year for rural analysis: 1-year estimates are only available for geographies with 65,000+ residents, which excludes most rural counties. For county-level rural analysis, 5-year estimates are the correct product. They cover all counties and geographies but represent a rolling 5-year average rather than a single year, so year-over-year changes should be interpreted carefully given overlapping sample periods.

NOTE

Mind the margins of error. ACS estimates are based on a sample survey, so every estimate carries a margin of error. Uncertainty is generally greater for smaller geographies and smaller population groups. When comparing rural communities or measuring change over time, consider the margin of error rather than treating the point estimate as an exact count. Kyle Walker’s margins of error vignette is an excellent primer on working with ACS uncertainty in R.

Key variables

The ACS covers hundreds of variables. The most analytically valuable for rural economic analysis fall into four groups:

Demographics

  • Age and sex distribution
  • Race, Hispanic origin, and ethnicity
  • Nativity and place of birth; year of entry
  • Disability status
  • Language spoken at home

Education

  • Educational attainment (less than high school through graduate degree)
  • School enrollment

Employment, income, and economic characteristics

  • Employment status and labor force participation
  • Occupation and industry of employment
  • Self-employment (incorporated and unincorporated)
  • Remote work / work-from-home status
  • Median household income; per capita income; poverty rate
  • Earnings by sex and occupation

Housing

  • Housing tenure (owner vs. renter)
  • Median home value; median gross rent
  • Housing cost burden (share of income spent on housing)
  • Vacancy rate and vacancy type
  • Internet access and subscription type
  • Units in structure; year built

Accessing the data

tidycensus is the go-to R package for ACS data — and one of the best data access tools in the R ecosystem. Constructed by Kyle Walker, it wraps the Census Data API with a clean, consistent interface that returns tidy data frames with geography attached, handles variable lookups, and supports both 1-year and 5-year estimates across all available geographies. If you are doing ACS analysis in R, start here.

# install.packages("tidycensus")
library(tidycensus)

# Load ACS variable list for a given year
vars <- load_variables(2023, "acs5", cache = TRUE)

# Pull median household income for all counties (5-year estimates)
income <- get_acs(
  geography = "county",
  variables = "B19013_001",   # median household income
  year      = 2023,
  survey    = "acs5"
)

# Pull multiple variables with geometry attached
edu <- get_acs(
  geography = "county",
  variables = c(
    ba_plus  = "B15003_022",  # bachelor's degree
    total    = "B15003_001"   # total population 25+
  ),
  year      = 2023,
  survey    = "acs5",
  geometry  = TRUE            # returns sf object with county boundaries
)

The Census Data API is also accessible directly for users who prefer Python, SQL, or other environments. See the ACS data portal for bulk downloads and the ACS methodology page for full documentation of survey design, sampling, and accuracy.

  • Dataset: Census Population Estimates Program — annual population totals that provide the denominator for ACS-derived per-capita rates
  • Dataset: FCC Broadband Service Data — infrastructure availability that pairs with ACS internet adoption and remote work measures
  • Dataset: QCEW Employment & Wages — administrative employment data that complements ACS’s household-level labor force picture
  • Package: ruraldefinitions — rural/nonrural classifications for filtering and comparing ACS estimates across county types