What’s the best way to map election results? I’m still trying to figure that out, but since election results have been finalized by Washington State it’s a good time to try and figure it out.
When mapped by the media, election returns are typically illustrated at the county level. Counties are often shaded by the percentage of the vote received by a particular candidate within a given county. The November 8th election is no different. More detailed results dig down to the precinct level and use a similar methodology to show the strength of support for a candidate in percentage terms. We don’t see a lot of precinct-level maps that span counties (Seattle Times did accomplish this for initial Sound Transit 3 ballot returns), partly because of the decentralized nature of the election process — there is no database of GIS files containing precinct shapes for all counties in Washington that is kept up to date…
In any case, spatially larger and more rural districts appear to have a stronger impact on the results than they actually do with the typical way that election results are presented. This is clear at the national and local level.
To solve this problem, folks often use cartograms, which reshape districts, counties, precincts, etc., to show their size in population or voters rather than their geography. KUOW recently did this for 2016 presidential election results in a bluntly titled piece, “This election map is a lie. So we made new ones.” Take a look at their cartogram:
This tells you a little bit more about Washington — where the most voters are located and which counties voted strongest for Clinton and Trump. But the counties lose their shape and you can’t really tell what is where. An alternative approach involves shading counties and precincts based on their voter density.
In mapping the presidential election results, I’ve tried to add detail across the region with precinct results while taking into account the density of precincts. As it turns out, there is no perfect way to map election results and to fully understand what’s happening, we need to view multiple maps, side by side.
These maps are a work in progress, as standardizing across different shapefiles and tabulated precinct results has proved difficult and time consuming. Not all precincts are perfectly matched yet and there are some gaps. Let’s have a look.
First, a traditional map with precinct level results that spans nearly all of the counties surrounding Puget Sound (partly because it was easier to find up-to-date shape files in these areas). This is the first election results map with this level of detail I have seen for our region:What do we see? It looks like Trump voters are everywhere! Large precincts in more rural and suburban areas skew the visual so that it looks like many areas had large numbers and moderate (here I mean only more than 50%) support for Trump. In Seattle, you see high percentages of each precinct voting for Clinton (Surprising or no? No). Viewing the map this way is interesting, but may actually be more problematic than looking at just county results, as it looks more red than blue in and area like Pierce County, even though the majority of voters their voted for Clinton. Let’s have another go at this.
In this second approach, I’ve taken the same precincts, shaded them red or blue depending on whether there were more votes for Trump or Clinton, and added a transparency layer for each precinct that varies based on the density of voters within that precinct.
Here, we get a very different picture. We see where the concentrations of voters are, in the dark red and dark blue areas, and areas with lower density are faded so as not to skew the map. As opposed to the first map, this shows where larger concentrations of Trump voters are — in Pierce County, near Puyallup. An improvement for sure. But what’s missing? Whether or not these precincts voted overwhelmingly for Trump. Comparing the two maps, we can so that this is not the case; while precincts south of Puyallup voted for Trump, they did not do so en masse. It would be great to see such analysis elsewhere, and for the entire state!
Three dimensional maps are a useful fix for the issues outlined above, making it possible to show the concentration of voters and what percentage voted for a given candidate, while keeping the general shape of the precinct, county, or other geography.
Some technical notes. In working to create these two maps I have run up again a few major roadblocks that others should address:
- Many counties do not offer downloads of their GIS shape files for election precincts. Ideally, this could be centralized at the state level. In fact, it is, but only circa 2010 as the result of the 2010 census & redistriciting process. The precincts have changed in some areas, and names do not align, making it difficult to link tables together.
- Not all counties provide precinct-level results. Typically these are more rural, lower population counties, and likely have fewer resources. Overtime I’m sure this will improve and I was surprised how many counties are providing this detail. Still, it is difficult to find the GIS files of the actual precincts.