art photo



I am interested in creating a contextual file for counties, places, and MCDs form 1970 - 2000. What is the best source for this? I want my items to be measured the same over time.

I cannot really determine the feasibility of this with Geolytics' Neighborhood Change database.


This will be a long answer.

First, I share your frustration with the Neighborhood Change database. While it does a great job of normalizing census tract boundaries, which makes for comparisons from 1970 - 2000 possible, it does a poor job of harmonizing the census indicators over time. It is as though they had 4 different research assistants on this project - each in charge of a separate decade.

Of course, the neighborhood change database won't work for you, because of the unit of analyses you are interested in.

I think the best resource for you is Social Explorer. They have tabular data from 1790+. They are using NHGIS as the source of their underlying data which in turn is based on summary files from the Census Bureau.

One really nice feature of their interface is the “pre-made” table choice. These are broad areas:

Essential Report; Comprehensive; Age; Sex; Income, Family Structure; Education; Industry, Occupation; Poverty, Employment, Housing; Means of Transportation; and Foreign Born, Place of Birth

Take a look to see how much comparability there is across these items for counties in Michigan from 1970 – 2000. I think you will be quite satisfied.

On occasion, the item you want will not be available across all years. For instance, for “Race”, % non-Hispanic white will not be available in 1970 because the Hispanic population was not large enough back then to make a difference to the white population. In fact, even in 1980, the percent Hispanic was only around 7%. So, lack of comparability in a Race/Hispanic item over the entire time period may not be very important.

If an item that you are interested in is not included in one of these pre-made reports, you can go back to the underlying data. For instance, the poverty report shows families in poverty. If for some reason, you want persons in poverty, you can find that from the “tables” choice.

There are some complications (of course).

First, not all the geographies you are interested in are available via Social Explorer. Only 2000 and 1990 have counties, places, and MCDs. [Also note that MCDs are not available for all states, even in 1990 and 2000. Only certain states are MCD states.]

For 1980 and 1970, you can make a fake query for the United States for the items of interest via Social Explorer. Social Explorer does an excellent job of documenting how each item is created (by clicking on the heading of the item). Copy the documentation, e.g., algorithm, including table numbers.

Then go to NHGIS and pull off the underlying items for places and MCDs and create the comparable measures. Note that even in 1980 and 1970, MCDs and places will not show up automatically in the NHGIS interface. The user has to click on “show more geographic units.

UM logins have premium access to Social Exploer. NHGIS requires registration, but access is given automatically, once you register.

Another source is from Surveillance Epidemiology and End Results - SEERS

They have attribute files for counties, but only for 1990 and 2000. However, the types of variables they have created might spark some ideas for you. The drawback is that the formulas they are using are based on Census Bureau Table definitions, not NHGIS or Social Explorer.

Annotated Resources:

Direct Links:

Related Question Groups: