Geo/Election Data QA: mggg-states

Data quality assurance on mggg-states. The QA notebook does the following:

• Finds naming discrepancies in the mggg-states datasets
• Compares mggg-states vote counts with MEDSL vote counts at the State-level
• Compares mggg-states vote counts with Wikipedia vote counts at the State-level
• Returns State-level summations of the mggg-states, MEDSL, and Wikipedia datasets for spot checking and column datatype analysis
• Checks the mggg-states shapefiles for missing or empty geometries
• Provides next steps recommendations for further QA

The QA is isolated to United States federal elections between 2016 and 2018 (inclusive).

Wikipedia Election Dataset

The Wikipedia Election Dataset was created through wrangling data scraped from Wikipedia pages. The scraping and wrangling details can be found in the jupyter notebook src/wikipedia-election-data-mining.ipynb. The dataset contains the Democrat, Green, Libertarian, and Republican vote counts for the following elections:

• 2016 United States presidential election
• 2016 United States Senate elections
• 2016 United States House of Representatives elections
• 2017 United States Senate elections (special elections)
• 2018 United States Senate elections
• 2018 United States House of Representatives elections

The data collection is timestamped at 10:00 AM ET, 14 Aug. 2020.

The columns are named according to the MGGG naming convention, which could be found in src/naming_convention.json.

Limitations:

Write-in candidate votes are not counted towards to overall party vote count. In the cases where elections have competing candidates of the same party, only the winning candidate's vote is counted.