Finding data normalized to 2000 tract boundaries

NHGIS
1

Hello,

I have a set of census tract numbers for participants in a study. Participants’ addresses were collected in 1980 and 1990, but these addresses were converted to tract numbers using the 2000 census boundaries. I would like to match these tract numbers to demographic data from the 1980 and 1990 censuses.

If I had access to participants’ addresses, I could use them to recalculate census tract numbers from 1980 and 1990 and then access the original tables from those years. Unfortunately, I don’t. It seems that census tract relationship files are available that provide interpolation weights quantifying changes in population. As a novice, however, I’m not sure how involved this conversion process would be given that the population is distributed across the entire country.

I understand that NHGIS only provides geographically standardized tables for 2010 census units, so I was wondering if a member of the community knew whether geographically standardized tables for 2000 census units were available elsewhere.

Thanks for any help you can provide.

2

A couple options:

  1. Get data “normalized to 2000 boundaries” from GeoLytics. According to their website, there are separate product bundles for 1980 data and 1990 data normalized to 2000 boundaries, costing $695 and $449 respectively. If you’re at a educational or research institution, you might check to see if it has already purchased these datasets or the 2000 version of GeoLytics’ “Neighborhood Change Database” (NCDB), which provided 1970-2000 data for 2000 tracts.

    (They’ve updated the NCDB using 2010 tracts and now 2020 tracts. I don’t see an option on their site to get the original 2000 version now, which is unfortunate not only because you need data for 2000 tracts but also because the 2000 version was more accurate, using a sound, well documented remapping method. They haven’t documented the methods they used for the 2010 or 2020 updates, and the data for 2010 boundaries is significantly less accurate than other options.)

  2. You could yourself interpolate 1980 & 1990 data for 2000 tracts. As we explain on our crosswalks page, it’s best to start with data for the smallest possible units. If the subjects you’re interested in are covered in short-form census data (from 1980 and 1990 Summary Tape File 1), you could start with 1980 and 1990 block data. You’d need to determine relationships between 1980 and 1990 blocks and 2000 census tracts, which you could do by overlaying NHGIS boundary data for 1980 & 1990 blocks with 2000 census tracts. (NHGIS provides all of this data through the NHGIS Data Finder, except for the 1980 block data, which are available through this page.)

    If your subjects of interest are only covered in long-form census data, then you could interpolate data from 1980 and 1990 census tracts (instead of blocks). I’d recommend using the techniques I defined in this article and in Chapter 3 of my dissertation.

I hope one of these options work for you. Unfortunately, I don’t know of any other simpler/cheaper options at this time!