top of page

Census Data:

Voter Turnout & State Legislative Districts

This project involves two highly related parts:

  • Building measures of voter eligible population at both the county and state legislative district levels, broken down by race, Hispanic status, gender and age. This involves estimating voting age population for these different groups and removing non-citizens and felons who are not eligible to vote.

  • Working with state legislative district level measures of census demographics


Uses specific to these data beyond the general uses already listed include:

  • Targeting eligible voters who are unregistered

  • Estimates of deadwood registrants

  • Analyses of the impact of demographic characteristics on voting behavior

  • Analyses of legislator behavior and responsiveness to the public

This project is more relevant to voter turnout than the other projects, obviously.

The felon data has many uses in the area of criminal justice. Existing datasets and work are over-reliant on federal statistics.


There are numerous challenges with Census data that I can discuss at length. One major problem is that the American Community Survey will not be adjusted by the Census with the benefit of information gleaned from the 2020 Decennial. Estimating non-citizens and felons at a higher level of detail than the Census reports is another challenge.


I am aware of at least six separate instances of analysts applying Census data to the wrong state legislative district because of confusion about redistricting and the timing of the ACS (American Community Survey). I believe such mistakes are common. I have ready-made datasets that will prevent such mistakes in the future and allow seamless merging of census and state legislative election or legislator data.


My log indicates I’ve spent 993 hours working with state legislative district census data, and another 804 hours on county census data, including model building for estimation, reading technical bulletins, and vetting data through analysis.


For building measures of voter eligible population, I’ve also gathered felon data from state-specific sources of much higher level of detail than that provided by the Bureau of Justice Statistics. Much of these data were in image form, requiring optical character recognition and cleaning by workers.


A large amount of the work on county level voter eligible population was done together with Michael McDonald at the University of Florida. His interests in any data we created together must be appropriately taken into account.

bottom of page