The Census Bureau released the newest 5-year set of American Community Survey data (2015-2019) just last month. This is important! We now have 3-sets of non-overlapping 5-year databases: 2005-2009, 2010-2014, and 2015-2019. I’ve been focusing on the 1-year databases up until now (2005-2019), but there are times when only the 5-year datasets will work (e.g., examining all counties within a particular state, region, or nation.)

 My first attempt was to try a simple analysis of total population, household population, etc., for all places in California. The r-package “tidycensus” works great in extracting the 5-year ACS databases.

 (By the way, the 2005-09 ACS does indeed have estimates for “total population” even though “group quarters” data was not collected in the year 2005. Somehow the Census Bureau used the 2006-09 data on GQ to make it up for 2005?)

 I ran this for 2005-2009, 2010-2014, and 2015-2019. Then stitched them together creating a database of California places with lots of variables. Perfect!

 Well, geographies change.

 I then had an “ah ha!” moment: the 2005-2009 geographies are based on the 2000 Census; and the 2010-2014 and 2015-2019 geographies are based on the 2010 Census. There are new and re-named places (at least in the Bay Area and California) in the 2010 Census relative to the 2000 Census.

 But this is not entirely accurate.

 The geographies in each of the 5-year ACS databases may be different. They reflect the “last year” vintage of each database. I had to look this up on the Census Bureau’s website to clear things up!

 So, for California:

2005-2009 ACS: 1,066 places in California

2010-2014 ACS: 1,543 places in California

2015-2019 ACS: 1,549 places in California

2010 Decennial Census SF1: 1,527 places in California.

 My merged database of the three sets of ACS data plus the decennial yields 1,558 records. So, I still have some work to do to clean this up. (The decennial census is used to append a county code to place records. I still haven’t figure out how I want to handle multi-county places, but I’ll leave that for another time.)

 ###

 Question: What is the vintage of the geography used in the 2015-2019 American Community Survey? 2015? 2019?

 Answer: 2019.

https://www.census.gov/programs-surveys/acs/geography-acs/geography-boundaries-by-year.html

“For ACS 5-year estimates, use the last year of the estimate period to determine the vintage. For example, the following datasets use the same vintages of geographic boundaries:

I thought there was a list of “new places” in the 2019 ACS products. I can’t seem to find it now. Just another buried treasure on the Census Bureau website.

Happy New Years to All, and Stay Safe!

Chuck Purvis

Hayward, California