Polling of Presidential Elections
In the wake of another exciting presidential race, we originally had the joy of speculating who was most likely to win. Perhaps the most effective way of catching a glimpse of who the next president will be is through polling. My main goals for this research project were to examine the overall accuracy of polls while attempting to determine which demographics tend to be most reliable for polling.
First, I think it’s important to take a look at America’s different voting groups. Using data from the 2012 and 2016 CNN Exit Polls, I’ve put together a graph representing eight significant voting groups with their respective voting percentages. I’ve set up the data to reflect the proportion of which each voting group consisted of during both Presidential Elections.
The proportions of the voting groups, of course, are a very important determinant in deciding who takes office. Even more important, however, is understanding how each group tends to vote. Below is both the polling and actual results of how each demographic group voted in the 2012 elections.
As you can see, with the exception of just a few groups, polling performed relatively well in 2012. Here are the projections and actual results for the 2016 election:
As you can assume, polling isn’t a perfect system. Although the 2012 and 2016 polls saw plenty of accuracy, they were each certainly off-kilter in predicting the voting pattern of a few key demographic groups. Interestingly enough, pollsters had trouble predicting the tendency of the 4 same groups in both elections. These groups include Women, Hispanics, College Graduates, and voters of age 45–65. In fact, each of these voting groups voted more-so Republican than the polls predicted. On average, the polls were off with these groups by about 4–5 percentage points.
Following the 2016 election, many people likely wouldn’t be surprised to hear there were voting groups projected to vote (D) that turned out (R) in the booths. After all, some polling inaccuracies surely must have occurred as we are left with the end result of President-Elect Trump. BUT, the thing I find most interesting about these 4 voting groups is that they turned out more Republican in the 2012 election as well. The main reason that I’m intrigued by this is because Obama won in much wider margins than originally projected. Therefore, I think there’s at least something special going on with the voting groups of Women, Hispanics, College Grads. and voters aged 45–65. Naturally, I think this mishap in polling had something to do with the great surprise that came with the results below:
I think it’s important to evaluate mistakes and try to make sense of why certain demographic groups are hard to predict. Amidst the many interpretations available for why these groups have been tough to predict, my prediction for this occurrence is relatively simple: Both elections have had extraordinary candidates. Whether people are voting to choose the nation’s first black president or voting to keep a ‘maniac’ out of office, both elections have essentially possessed never-seen-before factors. Specifically for the 4 demographic groups previously mentioned, there at least existed some level of “taboo” for them to vote for a particular candidate. For women, there was a potential to elect the first woman president. For Hispanics, they had an opportunity to try to keep Trump and his comments towards them out of office. Regardless, these groups voted as they voted and polling simply fell short of it’s goal to accurately predict the winner of 2016.
In Conclusion, I think polling is certainly important and should continue to be studied and taken seriously for assistance in predicting our future holders of office. Perhaps the wise words from President Obama can apply to this year’s polling dilemma. I will end with this: “The path that this country has taken has never been a straight line. We zig and zag and sometimes we move in ways that some people think is forward and others think is moving back. And that’s okay.”