Here we are at the 2020 Election in America and already data visualization maps are being peddled across social media purporting that America is more red than blue. While these posts (sometimes) acknowledge the fact that the popular vote in the 2016 election did not match the Electoral College, they go on to point to the map as an indicator that the majority of America is in fact conservative (Republican or Red State) which is why the Electoral College should be preserved (to reflect the will of the majority of people as opposed to big cities only).
This is emphasized by the large swath of the country represented as Red State (from an Electoral College standpoint.)
Democrats will quickly point out that this map does not reflect the true nature of the country as it only reflects the ‘winner take all’ approach of the Electoral College (in all but two States).
This is not a post about the validity or not of the Electoral College but rather of the maps that are used during and after Elections.
Done improperly, data visualization can be incredibly misleading.
The map in Figure 1 looks like most of America is Red (Conservative) but this is a wildly inaccurate representation of the population.
In response to this claim, Republicans will then send out a map by county. (Figure 2)
This map, they contend, is even more proof that the country is far more Conservative (Red/Republican) than the popular vote maintains simply because big cities shouldn’t speak for the rest of the country.
This map, however, is also misleading. All of those little shapes representing counties have vastly different amounts and ideological people living within them. Just like the state map, this map assigns a ‘winner takes all’ attitude to each County. Additionally, it does not account for counties that have very few people living in them, thus the often used claim from opponents of this map of “Land doesn’t vote. People do.”
It should be noted that these maps (Figure 1 and Figure 2) are often provided as evidence and with the stipulation that the Constitution determines how the Electoral College works and thus this represents our Election Results. This post does not dispute that these maps accurately reflect how the Electoral College works and thus are a good data visualization of those results.
This post however, is not about the Electoral College and its results, but rather the notion that these maps accurately reflect the makeup of America and its voters (they do not).
Moving forward, a more accurate representation, from a data visualization standpoint, of how people voted (in 2016), taking into account the population of each county is something called a ‘Dasymetric dot density map’. (Figure 3)
The term “dasymetric” refers to a map that accounts for population density in a given area. Instead of the ‘winner takes all’ approach of an entire state or county being red or blue, this map uses red and blue dots to represent every vote that was cast.
Further improving on the Dasymetric dot density map, another map emerged which further colored the country. (Figure 4)
This map rather than distributing the dots evenly around a county, the dots are distributed proportionally according to where people actually live, based on the US government’s National Land Cover Database.
The challenge with this map is that the blue dots, representing large populace areas tend to drown out the red ones and thus what once was a map that looked more red now looks more blue. While this more accurately reflects the popular vote (actual vote of people) it is also somewhat misleading as it doesn’t reflect county proportionality.
In an attempt to further bring the map back to something that reflects every county, comes the the Value-by-Alpha map. (Figure 5)
The Value-by-alpha map is similar to the Dasymetric dot density map but in a simpler form. It doesn’t account for where votes were most likely cast within a county, it uses color to indicate the party’s vote share in each county, and opacity to indicate the population in a given area of the county.
With this map, there are certainly Red and Blue and Pink and Light Blue areas. What is also evident in this map that does not appear as readily in any of the others is the abundance of purple.
This then, is perhaps the truest of maps as it relates to data visualization. Acknowledging that no map is totally comprehensive, this map comes as close as we’ve seen to being a true representation of America as we are no longer a country of simply Red or Blue, but one in which various shades of each and sometimes an amalgamation of both reflects who we are.