The Census could be the Decennial Census of Population and Housing, a count of every person in the United States, done every ten years. This is one of the products of the Census Bureau, which also administers and collects the American Community Survey (ACS), the Census of Governments, and the Economic Census. This combination of products provides governments, policymakers, researchers, and ordinary folks a wide array of demographic, social, and economic data on a continuous basis at various geographic scales.
The types of data you can find through census products include:
Statistics surveys such as the American Community Survey (ACS) require tradeoffs between accuracy and currency based on sample size. ACS 1-year estimates are lagged by a year, but the tradeoff for this more recent data is lower accuracy. 5-year estimates are created by aggregating data over a 5-year period to create larger sample sizes. These estimates are more accurate but are further lagged.
More information can be found here.
5-year estimates that include 2020, 2021, and 2022 may be somewhat strange depending on the topic - housing, commuting patterns, income, and employment datasets were affected by temporary COVID-19 pandemic era policies such as lock-downs, expanded remote work, stimulus checks, and eviction moratoriums, among other policies. Choosing 1-year rather than 5-year estimates may be more appropriate in these cases.
The geographic unit is another area of consideration. The most common census geographies range from the national level, to state, metropolitan area, coynty, tract, block group, and block. These particular geographies can be aggregated, that is, census blocks can be aggregated up to block groups, up to tracts, up to counties, etc. There are additional, non-comparable geographies such as county subdivisions, zip codes, urban areas, and tribal areas, among others.
The tradeoff in choosing a geographic area is between geographic size, currency, and accuracy. While the block may be the most granular unit, it contains the smallest number of people and thus is the least accurate unit, particularly for 1-year estimates (if available at all). To avoid losses in accuracy at the block level, you could use the 5-year estimates, but then the data will be less current.
The census tract represents an even balance between granularity, currency, and accuracy.
