In honor of Labor Day 2021, and to adhere to our mission of democratizing data, the Cornell ILR Buffalo Co-Lab’s High Road Policy blog is pleased to share a county-level geography of labor union membership for the United States…from a non-conventional data source: the MRI-Simmons LOCAL consumer survey. The MRI-Simmons survey collects and reports data on “the robust geographic nuances” of American “consumer attitudes and lifestyle”. One cluster of questions that features in the survey instrument concerns organizational memberships. Among a host of organizational types, adult survey respondents are asked if they are active labor union members. Upon collecting responses, MRI-Simmons uses sophisticated (proprietary) geobehavioral modeling techniques to produce union membership estimates for various geographies in the United States.

Below, we map these estimates for all U.S. counties for which a minimum of 1,000 adults in the MRI-Simmons LOCAL dataset were associated with active or former employment. The choice of 1,000 here was arbitrary, but made to exclude small sample cases with potentially unstable estimates. Data were obtained via SimplyAnalytics for both the most current (2019) and earliest (2013) versions of the survey available to researchers at Cornell University. The first visualization shows what might be thought of as a crude measure of union density: the number of self-reported labor union members divided by the sum of self-reported full-time and part-time employees, plus retired persons, in a county. Retired persons were included to account for the possibility that former union members remain active, and/or retain their identity, with their labor unions. The reason the measure is not a true proxy for union density is that the denominator is an approximation of a county’s workforce. As such, the percentages shown below are likely to under-represent the percentage of a county’s labor force that is unionized. Counties that did not meet the population threshold described above, and counties for which no data were available, are shown in grey.

Next, the second visualization shows the percentage point change in self-reported union membership over the six-year period from 2013 to 2019. The start and end points were chosen for reasons of data availability. As stated above, 2013 was the earliest version of the dataset available to our team, and 2019 is the current version. Counties that did not meet the population threshold described above, and counties for which no data were available, are shown in grey.

Finally, due to constraints with the mapping application we used to visualize the data, mousing over a county will show you the value being mapped plus the given county's unique geographic identifier (called its "FIPS-Code"). The following lookup table can be used to associate a FIPS-Code with a county and state name. Simply press CTRL+F on your keyboard and type the FIPS-Code that you wish to look up.