US Presidential Election Predictions

Frequently asked questions

Actually nobody has asked these questions, but just in case someone was thinking about this, we have provided the answers.

Q: It's obvious that you are biased as every forecast you have shows a democratic candidate. Why do you claim to be non-partisan?

A: The forecasts based on state rankings (Solid to tilts) show, as of March 2020, a higher chance that a democratic candidate will win in November. However, the rankings are not ours. The probability assignment table (PAT) is ours and we believe they are a reasonable interpretation of the ranking categories. In the near future, we hope to allow our visitors to change both rankings and the assignment table, to gauge the impact on the calculated results.

Q: Why is it necessary to calculated expected EV's and probabilities, when they were generally wrong in 2016? I mean, these websites were forecasting around a 70% chance that Hillary would win, yet Trump won the election.

A: In general, the forecasts missed on about 4 to 6 states which were key to Trump's election. Four states were decided by less than 1% of the electorate: MI(16), NH(4), WI(10) and PA(20) for a total of 50 EV's. Polling data certainly suggested that Florida would "lean" Democratic.

Q: Do you feel that the forecasts will improve the closer we are to November?

A: Forecasts rely on data, so will the data be better as we get closer to November? There's always the problem that the people surveyed don't always show up at the polling booth. More data are not necessarily better data. As a general rule, as the election predictions identify a very close race, it becomes more difficult to predict winners.

Q: How do you select the election forecast sites?

A: I have tried hard to include only reputable sites. So, no obviously partisan sites are permitted. I like transparency and full explanations. If someone takes the 270towin map, and my Probability Assignment Table, they can write a simple program and reproduce my numbers. In fact, I will make the simulation program available in the near future.