My wife is frequently the only person who hears my outbursts when I read or see the rampant irrelevant, or inaccurate and or incomplete news reports about who is likely to win the presidential election in November.
I hope to point people toward the gold standard of statisticians reporting on the candidates chances of winning in November. As of Saturday, August 6 Clinton’s chances of winning are at least 81.5 percent based on polls predicting the November 8 election. Donald Trump’s chance of winning is 18.5 percent. This forecast is volatile: on Friday, July 29 Clinton’s chance of winning was 53.3 percent and Trump’s chance of winning was 46.7 percent.
My source is statistician Nate Silver and his election forecast website FiveThirtyEight (http://projects.fivethirtyeight.com/2016-election-forecast/). The name represents the number of Electoral College votes available in the nation’s voting districts. Silver’s 538 forecast both President Obama’s victories in 2008 and 2012, missing one in 2008 getting each state correct in 2012.
Fivethirtyeight aggregates results from polls forecast for election day by HuffPost Pollster, RealClearPolitics, polling firms and news reports.
Silver also uses two other forecast versions.
His “polls plus forecast” is based on the polls forecast plus what the economy and historical data tell about the chances of winning on November 8. On August 6 Clinton’s chance of winning was 74.9 percent and Trump’s chance of winning was 25.1 percent. Again forecasts are volatile: Clinton’s chance of winning on that model was 61.7 percent on July 29 and Trump’s chance of winning was 38.2 percent
Silver’s “now-cast” model says who would win an election on August 6. Clinton’s chance of winning was 91.5 percent and Trump’s was 8.5 percent. On July 29, Clinton’s chance of winning was 48.4 percent, coming in second to Trump at 51.6 percent.
He plans to revise his forecasts and plans to publish them every time new data is available. Through November 8.
The models also built in procedures to handle third-party candidates, such as Libertarian Gary Johnson.
The website explains the forecast models in great detail, the electoral college map of the country, the closeness of the race in each state, the predicted results for each state and more. I think it’s a wonderful model for education and following results daily. I highly recommend it as an anecdote to the daily publications of one poll by one organization. Ignore those and the announcers who report them
For example, when Silver forecast Obama’s chance of winning close to 95 percent on election day in 2012, I watched in amazement as columnist George Will erroneously predicted Romney would win, even win an upset in Minnesota. Another columnist on the panel predicted the exact result Silver had made. It’s troubling to see such ignorance on display and it’s encouraging to see at least analysts have good sources.