Why Some Pollsters Get it Wrong

In an uncertain election cycle like 2016, political pollsters are putting forth their best efforts to accurately predict the outcome of elections across the country, including a volatile and unpredictable Presidential race. However, not all political polls will be correct for a myriad of reasons.

Why exactly are some polls better at predicting outcomes than others?

Many pollsters fail to predict Republican primary elections due to a flawed approach that limits their sample, and does not allow all eligible voters to participate in their survey. For example, in 2014 pollsters predicted Virginia Congressman Eric Cantor would win his primary by a landslide, only to be upset by opponent Dave Brat 56% to 44% on Election Day. While little was disclosed regarding the methodology for the polls that predicted Cantor’s win, they did disclose that the surveys only polled likely Republican primary voters from previous elections. We believe this is why they missed the mark and under-measured Dave Brat’s strength.

The bottom line is if your sample is only drawn from a a population of past primary voters and new registrants, you could be limiting the survey’s ability to measure voter opinion accurately. We believe this sampling method is flawed because it will only survey a portion of the electorate, and a portion that could be dramatically different that another portion that makes up the total voting population.

Furthermore, there could be a grassroots or populist movement transpiring that motivates a different portion of voters to participate in the election that have not in the past few cycles. Looking at past Republican polls where the pollsters have missed the mark, the people surveyed were not representative of the entire population, which subsequently leads to failed poll predictions.

At Magellan Strategies, we follow the approach of drawing our sample from a voter file a sample across the whole population of eligible voters from a voter file. We believe this approach allows for more voters participating in the survey, which will lead to a more accurate prediction of an election.

By | 2017-05-24T03:25:53+00:00 July 7th, 2016|