A member of Maine Voices Network contacted the Help Desk to inquire about our precinct boundary and election results data. They were looking for shapefiles of the state’s US House results for 2018 and 2020, which we did not have on the website at the time. They wanted to join those results to the 2018 and 2020 precinct boundaries that they obtained from the website. The goal was to share this data with other groups, who could then understand the partisan implications of proposed plans, including those drawn by the state legislature’s advisory commission (the Maine Reapportionment Commission).
Maine Voices Network had already obtained these election results from the Secretary of State’s website, so the Help Desk offered to join them with our existing precinct boundary files. But there were a couple of complicating factors:
These maps show the percent of the two-party vote for the Republican or Democratic congressional candidates. The darker the shade of red, the more votes for the Republican congressional candidates; the darker the shade of blue, the more votes for the Democratic congressional candidates. Pale or white areas are towns in which both candidates received roughly equal numbers of votes, and gray areas are towns without any votes.
An important aspect of working with redistricting data is understanding the rules in place in a state when it comes to election administration, reporting, and redistricting. That is because, as a general rule, states are constitutionally charged with administering elections, and each state has developed their own unique rules over time. This means that each state runs elections, reports election results, and redistricts legislative boundaries differently.
For example, Maine adopted Ranked Choice Voting (RCV) in 2016. In order to work with the 2018 and 2020 congressional elections, the results had to be filtered in order to produce a file with one number of votes cast for each candidate. In this instance, the results of the last round of RCV in each district were used for analysis.
Maine also has their own standards around reporting election results. Unlike most states, in which the county or county-equivalent perform most of the day to day administration of running elections and reporting results, towns (also known as municipalities) administer elections in Maine. Some towns create and report the data by wards (which is another name for precincts). In other towns, the election results are reported in aggregate at the town-level. When creating the precinct boundary and election results files, our data partners report the results at whatever level it is provided at, whether that be towns or wards.
As it so happens, the Reapportionment Commission stated they were unwilling to split towns in the congressional plan. This meant that town-level election results would suffice for the purposes of analyzing proposed congressional plans in Maine. But additional processing was necessary in order to merge the tabular election results with the precinct boundary shapefiles.
Maine Voices Network had acquired town-level election results from the state. Election results at the ward level are available directly from the towns who use them, but the reporting of the results are not standardized across towns. In addition, towns with wards report their absentee ballots at the town-level. The data partner who provides Maine’s election results joined with precinct boundaries allocate absentee votes down to wards based on the proportion of each candidate’s vote in the precinct. Using town-level data from the state, especially since the Commission was unwilling to split towns, would require significantly less time and effort than collecting and processing the data by contacting each town.
In order to prepare the data for joining, Maine Voices Network utilized the precinct boundaries and election results files we had on the website to match the precinct names in our 2020 file with the towns in their 2018 and 2020 files. This enabled us to join the election results by name. Because some precincts have the same names, we joined the data by matching both precinct and county.
While joining, we also noted that some precincts did not have any recorded votes associated with them. This can happen if a state has uninhabited areas, in which they do not create precincts. When creating precinct boundary and election results files, our data partners create state-wide data, which means that “null” precincts are created in order to cover these geographic areas. It can also occur in a sparsely populated area of a state, if no one from that area happens to vote in that election or particular race.
These datasets were necessary in order to join 2018 and 2020 US congressional election results in tabular form at the town level with shapefiles of 2020 precinct boundaries:
Shapefiles of the precinct boundaries and election results for the 2020 elections in Maine.
This dataset is available through the Redistricting Data Hub by creating a free account filtering by type of data: precinct boundaries and election results.
These data were provided by Maine Voices Network at the town level for the 2018 and 2020 US congressional races in xlsx format.
Joining multiple election results is a common task in redistricting. But the reporting of election results comes with many idiosyncrasies, and it is important to understand the rules in your state.
These datasets can also be used to examine the partisan fairness of a proposed plan, or identify partisan gerrymandering.
The code we used is available on Github. It is free to use, and if you need assistance or have any questions about using or understanding the code, please reach out to the Help Desk.
With the Help Desk’s assistance, Spotlight PA, a nonpartisan investigative journalism publication, created a tool that enables Pennsylvania residents to easily look up information about their old and new districts.
A member of a civil rights organization contacted the Redistricting Data Hub Help Desk with a question about our data. They had local election results at the district level that they wanted to merge with citizen voting age population (CVAP) data by race.