Important methodological notes
We suggest using the "Distance" version of the scale, as it was derived from a simpler procedure and did not appear to
suffer substantial limitations relative to the "Time" version. This is further explained and we make this same recommendation in the
The tract-level data file (here) contains three fields:
- FIPS: Census tract identifer. This is a commonly used identifier that should be 11 (eleven) characters in length for every record.
Note that if you open
the data file in Excel and save it, Excel is very likely to chop off a leading zero, resulting in some identifiers being only
10 (ten) characters. This could cause trouble when trying to link this data with other data sets. Don't use Excel. Ever. For anything.
- IsoDist: "Distance" version of the geographic isolation scale, as described in the
paper (preprint) (δ = 0.86).
- IsoTime: "Time" version of the geographic isolation scale, as described in the
paper (preprint) (δ = 0.89).
As of 2019-4-3, a zip code version of the measure is available. It was constructed in exactly same way as the tract-level
version with δ = 0.86, using zip shape files (2018 TIGER/Line) instead of tract, and population data from the
ACS (2017 5yr estimates). The zip code data file (here) contains two fields:
- Zip: 5 digit zip code.
- IsoDist: Same as tract level version
The published paper
(preprint) contains quantitative descriptions of the geographic isolation
measure that are not likely to align perfectly with those you generate yourself. The reason for this is that, while we used the
exact same data in the paper, the data set was subsetted to include only Census tract records for which all rural/isolation measures
were not missing. This resulted in the exclusion of some tracts from the analysis completed in the paper. Descriptive statistics
should only be subtly different, however.