Parsed and aggregated data from the Pennsylvania Secretary of State's official election reports for every election between 2006 and 2020.
We used it to determine how often candidates win without a majority. Indeed, whenever there are 3 or more candidates running for a seat in a Primary, the winner is more likely than not to have not reached a majority: instead, they receive a plurality of the votes with less than 50% of the total.
This means that most voters vote against the winner in primaries. We believe this shows that we need a new system, like Ranked Choice Voting, that ensures that majorities elect the winner.
We scraped the PA Secretary of State website for all CSVs for all elections they had available. This data is available in the outputs directory.
download.py: Downloads CSVs from pa.gov using selenium. Generatesoutputs/downloads/*.CSV.parse.py: Extracts statewide data from these CSVs, aggregating their county-level data. Generatesoutputs/results.csv.
The downloader works using a headless Chrome instance to click buttons on pa.gov to download all the data, since the data isn't easily accessible without user interaction.
Pennsylvania provides election data in a consistent format starting in 2006. That data is available here: https://www.electionreturns.pa.gov/ReportCenter/Reports
They also allow you to browse the data on the web is a really pleasing format, e.g.: https://www.electionreturns.pa.gov/General/SummaryResults?ElectionID=72&ElectionType=G&IsActive=0
This repo is only if you want to download ALL the data, rather than browse it. For most users, using electionreturns.pa.gov directly will make more sense than using this repo.
If you do want to use this repo, you probably don't need download.py unless there are new elections that are on pa.gov and not yet in this repo.
Otherwise, just modify parse.py to extract the data you want, or look at outputs/results.csv to see the data we extracted.