US Census Data¶

The US Census Bureau is the single most-used data source for US-based GIS work. It's free, authoritative, and updated regularly.

What's available¶

For most GIS analysis you'll combine TIGER/Line geographies with ACS attributes.

Geographies (smallest → largest)¶

Block → Block Group → Tract → County subdivision → County → State → Nation

How to get the data¶

Option A — data.census.gov (web UI)¶

1. Go to https://data.census.gov.
2. Search a topic (e.g., "median household income").
3. Filter by Geography (e.g., "All census tracts in Texas").
4. Pick a Survey (ACS 5-year, latest).
5. Click Download → CSV.

This is the easiest path if you only need 1–3 variables.

Option B — Census API (programmatic)¶

Get an API key: https://api.census.gov/data/key_signup.html

import requests
import pandas as pd

API_KEY = "YOUR_KEY"
year = 2022
ds = "acs/acs5"
get_vars = "NAME,B19013_001E"           # name + median household income
geo = "for=tract:*&in=state:48"          # all tracts in Texas (FIPS 48)

url = f"https://api.census.gov/data/{year}/{ds}?get={get_vars}&{geo}&key={API_KEY}"
data = requests.get(url).json()
df = pd.DataFrame(data[1:], columns=data[0])
df["GEOID"] = df["state"] + df["county"] + df["tract"]
df["B19013_001E"] = pd.to_numeric(df["B19013_001E"], errors="coerce")

The variable codes (e.g., B19013_001E) are documented at: https://api.census.gov/data/2022/acs/acs5/variables.html

Option C — pygris / tidycensus (Python / R)¶

For a friendlier path:

• Python: pip install pygris (geometries) + pip install census (variables)
• R: install.packages("tidycensus") — extremely popular among researchers

tidycensus example:

library(tidycensus)
library(sf)
income_tx <- get_acs(geography = "tract",
                     variables = "B19013_001",
                     state = "TX",
                     geometry = TRUE,
                     year = 2022)
plot(income_tx["estimate"])

Common ACS variables¶

Full list: https://api.census.gov/data/2022/acs/acs5/variables.html

Joining ACS to TIGER/Line¶

The GEOID is your join key:

Always treat GEOID as TEXT, not a number — leading zeros matter ('06037' not 6037).

In ArcGIS Pro¶

1. Add the TIGER tract shapefile.
2. Open the ACS CSV — make sure GEOID is TEXT.
3. Right-click tract layer → Joins and Relates → Add Join.
4. Join field: GEOID (left) ↔ GEOID (right).
5. Validate — count of matches should equal the number of tracts.

→ See How to Clean Data Before Joining for traps.

Margins of error (MOE)¶

ACS variables include a margin of error (e.g., B19013_001M). For small geographies (block groups, small tracts), MOEs are often huge. Always:

• Include MOE next to estimate when reporting.
• Aggregate up if MOE > ~50% of estimate.
• Use 5-year ACS for stability over 1-year for less-populated areas.

"Recipes" you'll run all the time¶

Median income choropleth (county level)¶

get_vars = "NAME,B19013_001E"
geo = "for=county:*"

Join to county TIGER → choropleth in ArcGIS Pro. → See Population Choropleth project.

% no-vehicle households (block group)¶

total = "B25044_001E"
no_veh_owners  = "B25044_003E"
no_veh_renters = "B25044_010E"

pct_no_veh = (no_veh_owners + no_veh_renters) / total * 100

→ See Transit Desert project.

→ Next: OpenStreetMap.