We’re still a long way from knowing who will be the eventual Democratic and Republican candidates for President in 2016. So you might think it’s a bit premature to make a prediction about who will win the election and go on to become the 45th President of the United States.
And that’s probably true.
However, in my last post I showed that by looking at the amount of alcohol people consume over the course of a year, we can actually do a fairly good job of predicting which states will be won by a Democratic candidate and which will be won by a Republican candidate (insert joke here).
An analysis of the five most recent presidential elections suggests that we can predict the winning presidential candidate in about 79% of states (including the District of Columbia) based on nothing more than the following three variables:
- Per capita consumption of beer in each state during the year prior to the general election.
- Per capita consumption of wine in each state during the year prior to the general election.
- Per capita consumption of spirits in each state during the year prior to the general election.
Table 1 shows the number of states each election cycle for which the winning presidential candidate was correctly predicted by a statistical model that included only these three variables as predictors.
Table 1: Number of states (including the District of Columbia) for which the winning presidential candidate was correctly predicted on the basis of per capita consumption of beer, wine, and spirits (1996-2012).
Moreover, the model correctly predicted which candidate would go on to win 270 votes in the Electoral College (and thus the Presidency) in every single election between 1996 and 2012 (see Table 2).
Table 2: Predicted vs. Actual Electoral College (EC) Votes for the Democratic Candidate for President (1996-2012).
And so this brings us to 2016.
Can we predict which party will win the White House based solely on how much people drink in 2015?
Perhaps, but one obvious problem is we don’t yet know how much alcohol people have consumed throughout 2015. After all, at the time of my writing this post we are only three months into 2015.
To somewhat get around this problem, I based my 2016 presidential forecast on projections of what per capita alcohol consumption in each state might be at the end of the year. The projections are based on alcohol consumption rates from 1994 to 2011, as published in the Brewer’s Almanac at BeerInstitute.org.
Before reading further though, an important disclaimer: Obviously, there are problems with using projections rather than real data as a basis for making election predictions (e.g., some of the projections are probably wrong!), so imagine fairly wide error bars around all the numbers reported below and take this all with a grain of salt.
2016 Presidential Election Predictions
Figure 1 below shows per capita alcohol consumption in all 50 states, including the District of Columbia, between 1994 and 2011.
It also shows the probability of the Democratic candidate winning each state in 2016, based on a projection of how much beer, wine, and spirits people in each state might consume in 2015.
Figure 1: Per capita alcohol consumption in each state and the District of Columbia, from 1994 to 2011 (top); and Probability of the Democratic candidate winning each state in 2016 based on projections of per capita alcohol consumption in each state (bottom).
Note: Zoom out to view Alaska and Hawaii.
*Data on per capita alcohol consumption from 1994 to 2011 are from from the Brewer’s Almanac at BeerInstitute.org.
Assuming the Democratic candidate will be victorious in any state with a probability higher than 0.5, the model predicts the following outcome on November 8, 2016:
Table 3: Number of states and Electoral College (EC) Votes predicted to go to the Democratic and Republican presidential candidates in 2016.
“EC Votes” reflects the number of Electoral College votes predicted to go to each candidate, based on projected per capita consumption of beer, wine, and spirits in each state.
“Incorrect” reflects the number of Electoral College votes that I expect the model to award incorrectly to each candidate, based on my previous analysis of presidential elections between 1996 and 2012.
We obviously don’t expect the model to predict every state accurately. In fact, I would only expect it to predict the outcome correctly in about 79% of states (see Table 1). We can get a sense for where the model’s predictions might break down by again looking at Figure 1. For example, the model predicts a Democratic victory in both Alaska and Idaho, something that seems rather unlikely given that neither state has gone to a Democrat since 1964.
“Correct” reflects the number of Electoral College votes that I expect the model to award correctly to each candidate, calculated as “EC Votes” – “Incorrect.”
“Corrected EC Votes” reflects a refined estimate of the number of Electoral College votes predicted to go to each candidate. Given that Electoral College votes incorrectly awarded to the Democrat rightfully belong to the Republican and that Electoral College votes incorrectly awarded to the Republican rightfully belong to the Democrat, “Corrected EC Votes” is calculated as shown below:
Corrected EC Votes (Dem) = Correct Dem (254) + Incorrect Rep (70) = 324.
Corrected EC Votes (Rep) = Correct Rep (167) + Incorrect Dem (47) = 214.
So as you can see by looking at Table 3, the model predicts a fairly decisive win in 2016 for the eventual Democratic candidate, with a victory in 26 states and somewhere between 301 and 324 votes in the Electoral College. Either way, substantially more than the 270 Electoral College votes required to win the Presidency.
So now that we know who will win the White House in 2016, all that remains to be seen is who precisely will be running for President…in case that matters at all.