Daily Kos Elections is pleased to announce the completion of its project to calculate the 2020 presidential election results for each of the nation's 435 congressional districts that will be used in the November midterm elections. This data, available in a single chart here, shows how the 2020 election would have gone under the new district lines that are now in place following the most recent round of decennial redistricting. (We analyze some of our key findings in this post.)
To calculate these results, we relied on four types of data:
- Congressional districts broken down by census block
- Voting precincts broken down by census block
- Election results broken down by voting precinct
- Voting-age population broken down by census block
Census blocks are the smallest, most granular geographic unit for which the Census Bureau provides data. These blocks serve as a sort of Rosetta Stone that can transform one type of data into another.
When states adopt new congressional maps, those maps are defined by the census blocks that make up each district. States either publish lists of which blocks are assigned to each district, known as a block assignment file or block equivalency file, or they provide a data file known as a shapefile that is typically used for mapping but can be transformed into a block assignment file. We have collected BAFs or shapefiles (which we then turned into BAFs) from official sources in every state. This gives us the first type of data listed above: congressional districts broken down by census block.
States do not, however, typically provide block assignment files for voting precincts. Fortunately, the dedicated researchers of the Voting and Election Science Team, better known as VEST, have created and shared “crosswalks” that combine census data with precinct-level election data to create files that show which census blocks are assigned to each voting precinct. (To do this, they’ve used the same method described above: They’ve painstakingly collected shapefiles for every voting precinct in the country and transformed them into blocks.) This gives us the second type of data listed above: voting precincts broken down by census block.
Since we now have lists of voting precincts by census block and congressional districts by census block, we can combine these two data sets to tell us which precincts are in which district. This is necessary because, as above, states seldom provide this data themselves.
We have also collected the third type of data listed above—election results broken down by voting precinct—by going state by state, and in many cases, county by county, since many states do not provide precinct-level election results in a centralized repository.
Some voting precincts are split between two or more congressional districts, creating “sub-precincts.” However, election authorities typically do not report presidential election results at the sub-precinct level. We therefore use the fourth type of data, voting-age population broken down by census block, to proportionally allocate precinct-level election results between sub-precincts. (This population data is calculated by the Census Bureau and accessed via IPUMS NHGIS.)
We can then combine our list of congressional districts broken down by precinct and sub-precinct with our collection of election results broken down by precinct and sub-precinct to calculate the 2020 presidential election results by congressional district. This yields the data we provide in our new chart. For maximum transparency, we’ve also published all of our spreadsheets containing our state-by-state calculations, which you can find linked here. You can also find the raw election results we’ve relied on here.
For a more in-depth explanation of our complete methodology, please check out our detailed methodology statement.
We owe a special debt of gratitude to VEST’s Mike Migurski, who patiently answered our many questions as we undertook this project. Without his crosswalks, this effort would not have been possible. We are also incredibly grateful to the singular Jay Fierman of the Redistrict Network, Dave Bradlee and Terry Crowley of the indispensable Dave’s Redistricting App, Kari Chisholm, Nathaniel Rakich, Scott Kennedy, Open Elections, and the many state and local officials who assisted us in gathering data.