Census Blockgroup data and Connecticut

Hi there. I’m trying to work with the Census Blockgroup data and I’m running into an issue regarding Connecticut.

In the Safegraph Open Census data, I see 2716 unique Blockgroups for CT. When I tried to align the Safegraph data with my own data, I’m able to match all of my records except for Connecticut.

I thought that was odd, so I dug around a bit on the Census site and they have 2717 Blockgroups listed in CT, and not one of them is in the files provided by Safegraph.

Looking a bit deeper, all of the Blockgroups in the Safegraph data start with (090).
Looking at the data straight from the Census, all of their Blockgroups start with (091)

An example can be found in this image:

Since this data is broken, it makes me somewhat suspect for usage of all of the data.

I’m pulling my CT blockgroups by looking at the data from this link:


Specifically the CT file: tl_2020_09_bg.zip

Just thought folks should be aware.

Hi @jreynolds, thanks for sharing this issue! I found this article on recent changes to Connecticut’s FIPS, which are detailed here.

I haven’t dug into the details yet, but I strongly suspect these changes—which appear to have begun rolling out in late 2022—are the culprit behind the discrepancy you’re seeing.

I will escalate this to ensure Dewey researches, handles, and communicates the transition appropriately.


Thanks for the info. That appears to be the problem. Sigh.

I dug a bit deeper than there appears to be a crosswalk table:

I guess I will see if I can figure out how to apply that change to my overall process.

Thanks again.

1 Like

Jeff–i think SafeGraph uses the 2010 map (at least that merge has always worked for me).

If you have to use 2020 data then I think ipums publishes crosswalks.

1 Like