In this project, we set out to determine the impact of prison gerrymandering in the state of Texas, particularly how it would affect electoral maps. To accomplish this goal we obtained data from the Texas Department of Criminal Justice on every inmate in the state and joined it with Census and election data. We conducted a (to our knowledge) first-of-its-kind analysis, by seeing what would happen if prisoners were counted not in the county they were incarcerated in, but where they were charged (the best proxy we had for their home address) . We also assessed the impact on the
Our analysis shows that if Texas were to not count prisoners at all in the areas where they are incarcerated, this would throw nearly three dozen House districts out of population boundaries, making them subject to court challenge. We also found that nearly 1 in 5 counties would lose population to more urban, liberal areas if prisoners were counted in their county of charge. These findings may prove influential in several ongoing lawsuits against the state alleging that they illegally gerrymandered districts, including one from a state prisoner seeking to be counted where he is originally from. Our analysis was praised by Texas’ state demographer and by independent researchers examining the same issue and was featured by the Washington Post and in the Local Matters newsletter.
To conduct this analysis, we began with three data files: one from the Texas Department of Criminal Justice with each incarcerated person’s name, county of charge and the facility where they were housed on March 31, 2020; another from TDCJ showing the location of each prison; and a third from the Census Bureau containing the population of every county in 2020. These three files were joined together for analysis.
To conduct the reallocation analysis, we wrote a script in Python to loop through every row of the joined data, adding one person to the county where the prisoner was charged and subtracting one person from the county where they were incarcerated, to generate final adjusted populations for every county in Texas.
We then joined a fourth data table in Python showing county election results for 2020, obtained from the Texas Secretary of State’s office, to the combined data. The preprocessing for this data was done using Pivot Tables in Excel.
To create our visualizations for the story, we obtained a file containing the total incarcerated population for every Census Tract in Texas from the Census Bureau and joined it to a shapefile of Census Tracts in QGIS. After joining, we exported this file and put it into Tableau for styling. The other visualizations were created using our in-house charting tool Chartwerk.
To calculate what would happen to the electoral maps if prisoners were not counted at all, we subtracted the total number of inmates from the state’s population and then divided that by 150 (the number of House districts in Texas), to obtain a new “ideal” district size. We then calculated in Excel which districts would be more than 5% greater or less than 5% smaller than the new ideal district size.
What was the hardest part of this project?
The hardest part of this project was working around the limitations of the data that was available. Unfortunately, we weren’t able to get the home addresses for the inmates in our dataset, so we had to use the closest proxy — the county where they were charged. This also made it impossible to assess what would happen in the state and legislative districts if prisoners were moved back to their home addresses. Towards the end of the project, we received data on how many prisoners were from each ZIP Code, but this too had problems. The first issue was that we only had the aggregations not ZIP Code for each of our nearly 140,000 prisoners. The second problem was that these ZIP codes were self-reported by the inmates themselves, meaning that there were hundreds of values missing.
Despite these limitations, the jury should select this project because it allows fellow journalists, advocates and policymakers at the state and federal level to make data-driven decisions about how to address the issue of prison gerrymandering in their state.
What can others learn from this project?
We hope that fellow journalists use this project as a playbook to examine prison gerrymandering in their state. More of this reporting may pressure lawmakers and government officials to release more of this data, allowing the public to have greater insight into the issue. We also hope this project encourages more reporters to pursue stories which straddle the line between social science research and journalism, particularly in the realm of criminal justice and voting rights.