Should I compare data year-over-year?

The US Census Bureau recommends against comparing data across the Census’ American Community Survey 5 year datasets (2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, and 2020) with overlapping years.

The data estimates are from the Census Bureau’s American Community Survey. To produce these estimates, the ACS combines data across time periods. The ACS’s ongoing survey collects data all year round to produce 1-year estimates, and over 5 years to create the 5-year estimates. These estimates represent the characteristics over an entire data collection period.

The difference between the 2019 ACS and the 2020 ACS datasets is largely driven by the differences between dropping off 2015 and adding 2020.

  • For the 2020 dataset, surveys were collected in 20202019, 2018, 2017, and 2016. Then, the later year survey responses were projected forward (e.g. 2016 income was projected into 2020 dollars). 
  • The 2019 dataset is made up of surveys that were collected in 2019, 2018, 2017, 2016, and 2015.

Comparing overlapping data does not measure what you may be seeking. The difference between overlapping estimates is essentially measuring the difference between the nonoverlapping portions.
For example, comparing the 2019 5-year ACS to the 2020 5-year ACS does not measure the change between 2019 and 2020, which is what you may be interested in knowing. Since the 5-year data represent a longer timeline, they are not likely to capture abrupt changes.

There are 4 years of overlapping survey results (2019, 2018, 2017 & 2016) between the 2019 dataset and the 2020 dataset. If you are doing historical comparison work, you'll instead want to compare the 2020 data with the 2015 data, because there won't be any overlapping years between these 2 datasets.  We tend to pull 2020 (most current), 2015 (5 year without overlapping collection years) and 2000 data for these types of projects.

The good news is that there won't be a lot of change between 2019 data and 2020 data, because the estimates are based on 4 of the same survey years and only have 1 unique survey year. Most of our clients wait 2 to 3 years before buying a more current dataset. The only clients who get year over year updates are those clients, who for marketing purposes, have to be able to say "we are using the most current US Census data." So marketing drives the data update decision rather than changes in statistics for these clients.


Bureau, US Census. “Period Estimates in the American Community Survey.” The United States Census Bureau, Accessed 30 Mar. 2022.

Did this answer your question? Thanks for the feedback There was a problem submitting your feedback. Please try again later.