This post begins a series of posts intended to lay out what may and may not be concluded from the recent election, based on hard data. As much as possible, this series will attempt to be entirely non-partisan, simply laying out as accurate an explanation of the data as possible. As the vigorous debates about polling indicate, voters as a whole, and Christians in particular, struggle to examine data apart from partisan passions, but doing so is a tremendous boon to understanding the world as it is, and that itself is key to understanding how the world must change to become what it should be.
The popular vote is still being tallied. Clinton leads by a million votes — just under one percentage point — and seems likely to retain that lead when the final tallies come in. That fact does nothing to help her, however, because Trump has already locked up (not accounting for possible defections, which will be rare) at least 290 electoral votes, and will likely win 306 when Michigan’s final results come in at the end of November. Does this, as some of his supporters claim, constitute an electoral college “landslide”?
It depends how you define a “landslide win,” but I suggest that an examination of historical data strongly suggests otherwise. Of course, it doesn’t take much digging to recognize that Trump’s Electoral College victory was far more convincing than George W. Bush’s 2000 271-267 margin or his 2004 286-251 win. On the other hand, Trump’s win falls well short of Obama’s 2008 365-173 win, which more than doubles Trump’s margin of victory, as well as Obama’s 2012 332-206 win. Of those, the last four presidential elections, the winning candidate took an average of just over 311 electoral votes, or 57.9% of the Electoral College, thus of the last five presidential elections Trump’s win in the Electoral College is the median, and slightly below the mean.
You could keep walking that four election window back in time, to Clinton in 1996 (379-159, 70.4%) and 1992 (370-168, 68.8%), George H.W. Bush in 1988 (426-111, 79.2%), or Reagan in 1984 (525-13, 97.6%) and 1980 (489-49, 90.9%). By this time, the winner’s average share of the electoral vote has risen to 70% (with a standard deviation for the sample, which measures how close together the results are, of 16 percentage points). You might be tempted to argue that Clinton, George H.W. Bush, and Reagan were all among the most decisive victories in US history, thus that average isn’t an accurate representation of what margin should be considered ordinary. Something interesting starts to happen when you back farther, though. If you go back to 1964, the first election in which the Electoral College had 538 members, neither the overall average nor the standard deviation change much, both rising by one point to 71% and 17%, respectively (the average and deviation for that period alone, from 1964 to 1980, are 75% and 22%). You can go even farther back, to the turn of the last century, and the overall average remains quietly at 73% while the deviation drifts down to 15% (75% and 13% for the period 1900-1964 alone). In fact, if you go all the way back to the founding, the average winner in US presidential elections takes 71% of the electoral college with a standard deviation of 16%.
Now, it’s debatable how much the election of 1796, for example, should inform our expectations for 2016, but it is certainly interesting to see how constant the average victory has been through time. The last five elections have actually been singularly close in the Electoral College compared to what had been the norm in US presidential elections. One way of seeing this is to plot the percentage of the electoral vote earned by the winner against the relative cumulative frequency (or percentile) — that is, the percentage of the results less than the one being considered (this is the same way test scores are often reported — someone in the 50th percentile, for example, has a score higher than 50% of those who took the test).
The winning candidate’s share of the Electoral College, plotted by percentile.
Here all 57 elections decided in the Electoral College are plotted, with the share of the electoral vote taken by the winner on the x-axis and the percentage of US presidential elections in which the winner took a lower share of the electoral vote on the y-axis. (The graph doesn’t start at zero on the y-axis because 1824, which was not decided by the Electoral College, is not plotted because 1824 was a mess and we Americans do our best to forget it.)
The median result, then, was Clinton’s 1992 370-168 victory. The overall average plots just above that, at 71%, and 67% of all US presidential elections lie between 55% (one standard deviation below the mean) and 87% (one standard deviation above the mean); elections between these two boundaries are most common, considered roughly average wins. Trump’s victory with 56.9% of the Electoral College is therefore on the lower end of an average electoral vote outcome, and sits at the 22nd percentile — that is, it is more decisive than 22% of presidential election and narrower than 77%.
You could argue that by opening the window so wide — we’ve considered more than two centuries of elections, and those reflect widely varying conditions — we’ve reached a flawed conclusion. That argument is not without merit, but consider the outcomes annotated on the graph: the ten most recent elections. For these, the average and standard deviation are, as noted above, again the same, again placing Trump’s win on the lower end of average. Further, although Trump’s percentage of the Electoral College creeps upwards, it’s only into the 30th percentile. Further still, as we’ve already shown, the window that presents Trump’s victory most favorably, 2000-2016, which includes two of the closest elections in US history, still leaves Trump’s 2016 percentage the median, and just below the average.
In short, the conclusion that Trump’s victory was on the low-end of average is independent of the size of window used. Does that mean you can’t call it a landslide? Well, that’s where we get to what a “landslide” is — the term is not precise, so there’s some latitude in assigning definitions. If you’re willing to concede that more than three-quarters of presidential elections have been landslides in the Electoral College, than yes, you can still call Trump’s win a landslide. Be aware, however, that you’ve rendered the term meaningless with that definition.
(As an aside, if we’re going to be strictly precise in our definitions, then a landslide should be an abnormally large share, most simply defined as more than one standard deviation from the mean. That would place the “landslide line” at 87%, falling neatly into the gap between Eisenhower in 1956, who took 86% of the electoral vote, and Roosevelt in 1932, who took 89% of the electoral vote. That may exclude some results that would generally be considered landslides, however, so your mileage may vary).
The key takeaway here is twofold, and neither one is really whether or not you should call Trump’s win a landslide (you could call it landslide, or you could call it a spandopeashire, and as long as you invented a definition that fit, you’d be correct). First, much of the political media is prone to hyperbole, and nothing can replace carefully thinking through claims when they are presented to you. Truth matters, and it should matter to us. Second, and more immediately applicable, when drawing conclusions about the state of the electorate from the election results, we should bear in mind what kind of result we’re looking at. That will become more clear once the final popular vote tallies come in (and once that happens, we’ll be here to examine it), but in the results where we already have clarity, we can conclude that the 2016 election is not a particularly decisive outcome.
What does that mean? Well, mostly it concerns what we cannot reasonably conclude from the election, but we’ll have to get to that next week.