The next several posts will look at voting patterns in the US Presidential elections. I’ve got data covering the past six elections (1996-2016), broken down by state and county.
Let’s jump right in. Well, before jumping in, I need to make some terminology clear, with an example of how I represent the ‘lean’ of an election. In 2004, George W. Bush beat John Kerry, getting 50.73% of the vote to Kerry’s 48.27% (the rest were third party and write-in votes). Bush won by 2.46%, so in shorthand I write it as R+2.46. So when you see R+X or D+Y, that just means that the Republican or Democratic candidate won the popular vote by X or Y percent, respectively.
Today’s question: which areas reflect America’s voting? Specifically, which states (and counties) come closest to matching how the nation voted as a whole. We can calculate this by measuring how far the state varies from the national vote in each election. As an example, here’s New Hampshire:
| 1996 | 2000 | 2004 | 2008 | 2012 | 2016 | Ave | |
| US | D+8.52 | D+0.52 | R+2.46 | D+7.27 | D+3.86 | D+2.09 | |
| NH | D+9.95 | R+1.27 | D+1.37 | D+9.61 | D+5.58 | D+0.37 | |
| Difference | 1.43 | 1.79 | 3.83 | 2.33 | 1.72 | 1.72 | 2.14 |
To understand this table, let’s look at the 1996 column. Nationally, Bill Clinton won by 8.52%. In New Hampshire, he won by 9.95%. So the Granite State was off by 1.43%. Tally up the differences of the other elections, and the average is 2.14%. Thus, New Hampshire is typically within a couple of points of the national margin.
That actually turns out to be the lowest difference of all the states. This doesn’t mean that voters in that state always pick the “winner” (either popular vote or electoral college), it’s just that they are closest to the national result. Here are the states that were closest to the national vote over the past six elections:
| State | Difference from national vote |
| New Hampshire | 2.14 |
| Florida | 2.77 |
| Pennsylvania | 2.78 |
| Wisconsin | 2.92 |
| Iowa | 3.26 |
It’s not surprising that you see some key ‘battleground’ states in this list (Florida, Pennsylvania, Wisconsin): largish states that reflect the country as a whole. As they tip, so might tip the election.
Next, here’s the data at the county level – the counties that average closest to the national vote, 1996-2016:
| County | State | Difference from national vote |
| Granville | NC | 1.54% |
| Montgomery | OH | 1.74% |
| Marengo | AL | 1.81% |
| Sandoval | NM | 1.81% |
| Sumter | GA | 1.87% |
| Dakota | MN | 1.95% |
| Hillsborough | NH | 1.96% |
| Pinellas | FL | 1.99% |
| Baldwin | GA | 2.05% |
| Radford | VA | 2.09% |
As Granville goes, so goes America? These counties are literally all over the map, but they do have a couple of things in common: they aren’t too rural (with a heavy R population), and they aren’t too urban (heavy D). I might dive into what these counties have (or don’t have) in common in a later post.
What about the other end of the spectrum, states and counties that least reflect how America votes? Since presidential elections hover around 50-50, the least representative states are those that go hard for one party or another. And that’s what we see: Washington D.C. (massively Democratic) and several solid Republican states top the list. (Note that D.C. counts as a ‘state’ because it has electoral votes, and also as a ‘county’ because it isn’t broken up into counties. It’s like a state with a single county.)
| State | Difference from national vote | Lean |
| District of Columbia | 78.07% | D |
| Wyoming | 38.66% | R |
| Utah | 36.83% | R |
| Idaho | 34.18% | R |
| Oklahoma | 30.31% | R |
As far to the Republicans that Wyoming leans, D.C. leans twice as hard to the Democrats.
And here are the counties that are furthest from the norm:
| County | State | Average Difference | Lean |
| Glasscock | TX | 83.89% | R |
| Ochiltree | TX | 82.48% | R |
| Roberts | TX | 81.08% | R |
| Hansford | TX | 79.37% | R |
| Garfield | MT | 79.15% | R |
| King | TX | 78.70% | R |
| District of Columbia | DC | 78.07% | D |
| Arthur | NE | 77.78% | R |
| Wallace | KS | 77.52% | R |
| Hayes | NE | 77.42% | R |
Today’s Lesson: if you want to talk to an average American voter, do not go to the Texas panhandle. Four of the top six counties in this list are in that area (and Glasscock isn’t even one of them; it’s in West Texas). To be fair, the ‘county’ of Washington D.C. is way out of whack too, but on the D side of things. There’s a simple reason why this list has so many heavily-Republican counties: strong R counties tend to be rural and have a smaller population, so there are lots of them. Strong D counties are urban and have a much larger population, and so there are fewer of them. If you broke Washington D.C. into lots of small counties, there’d be a ton of them on this list too.
Those Texas counties are remarkable – reasonably large geographically, very low in population, and extremely Republican. For example, King County is 913 square miles, and 159 people voted in 2016. Hillary Clinton received five of those votes. If you stood at a random place in this county, and put a weight on a 5 mile length of rope, and swung that rope in a circle, the odds are you still wouldn’t hit a Democrat.
And because I know you’re wondering, the corresponding rope in Washington D.C. would only need to be 31 feet long.
I like your five-mile rope…