Election Anayltics: David & Goliath of election forecasting

Fivethirtyeight.com has become the gold standard for election forecasting. Since being purchased by ESPN in 2014, they have a staff over two dozen writers, analysts, and graphic artists,, providing statistical analysis across a spectrum of areas, broadly classified as politics, economics, science and health, life, and sports. Their stream of blog commentaries and insights typically based on data analysis, attract an enormous following across a wide swath of the population.

With their launch in 2008, the genesis of their name, Fivethirtyeight.com provided forecasts for the United States Presidential election. Since then, they have expanded to include the United States Senate, the Presidential primaries, and any event for which predictions can be made.

A natural question to ask is, “How accurate are they?”

Two measures can be used to make this assessment. The first, called the Brier Score, measures the squared distance between each forecast (the probability of the correct selection, given between zero and one) and the outcome (given as a one), summed across all the forecasts. The second, called the Entropy Score, measures the negative of the natural logarithm of each forecast (the probability of the correct selection), summed across all the forecasts. A perfect prediction would result in both a Brier Score and an Entropy Score of zero. In general, smaller scores indicate better forecasts.

With Fivethirtyeight.com being the “Goliath” of election forecasting, a lesser known academic site, the “David” of election forecasting, is a group of undergraduate computer science students at the University of Illinois at Urbana‐Champaign. This group oversees Election Analytics, a nonpartisan website that forecasts the Presidential and Senate elections.

Launched in 2008, Election Analytics represents a STEM learning laboratory for these students. The students experiment with new ways to present data, participate in the interface design, and analyze the data for posting. The methodologies used by Election Analytics have been published in peer-reviewed journals, allowing anyone to see how they use polling data to make their forecasts.

The following tables looks at the Brier and Entropy Scores for the 2008 and 2012 Presidential Elections and the 2012 and 2014 Senate Elections. These are the four elections that the two web sites (fivethirtyeight.com and Election Analytics) both provided forecasts.

Election Analytics had a lower Brier Score than Fivethirtyeight.com in three of the four elections. For the Entropy Score, the two sites split. The total Brier scores for the two sites were 3.7402 (Election Analytics) and 4.1298 (Fivethirtyeight.com), and the total Entropy scores for the two sites were 11.993 (Election Analytics) and 14.2357 (Fivethirtyeight.com).

Based purely on accuracy of the final forecasts, Election Analytics has provided a more accurate picture of the outcome of the four recent elections. They have also done this at a fraction of the cost (Total direct expenditures by the Election Analytics group in 2016 was $25, not including computers and support provided by the University of Illinois.)

Fivethirtyeight.com is an entertaining site for the general population to gain insights into the election and provide informative discussion and thoughts. No matter which site is more accurate in predicting the outcome of the 2016 Elections, the students involved in designing and maintaining Election Analytics are the real winners. Election Analytics is an activity that will launch their STEM learning in ways that makes a difference, far beyond this year’s election.

The ideas and methods used in the Election Analytics website originated in a research paper written by Sheldon H. Jacobson and his collaborators, published in 2009.
_____________________