#CERCAtheVote: Nate Silver’s Election Predictions" width="596">
Nate Silver is the most well-known and well-regarded statistician when it comes to predicting electoral outcomes. He rose to prominence in 2008 when he accurately predicted 49 of 50 states’ election outcomes and was named one of TIME‘s most influential people. In 2010, his blog was incorporated into The New York Times, and in 2012 he correctly predicted all 50 states’ votes in the presidential election. The following year, he left the Times to focus on FiveThirtyEight, a data-driven journalism site with sections dedicated to politics, sports, and culture.
Recently, Silver has been in the news for predicting that Donald Trump has a 1-in-3 chance of winning the general election, despite many other respectable polling sites giving Trump only about a 10-15% chance of winning. This has caused some journalists to accuse Silver of skewing polls to favor Trump or overcompensating for predicting early on that Trump would not become the Republican nominee.
[Related: Does Donald Trump’s claim of a ‘rigged election’ hold up?]
The weekend before the election, Silver took to Twitter to respond to his critics and defend the 1-in-3 odds. What do your students think? Does Silver’s claim that Trump has a shot at wining the presidency hold up?
The Claim
Clinton's rebounded just a tiny bit in our forecast. Current lead over Trump is 3.3 points. https://t.co/WLXtJovjYd pic.twitter.com/WaTVjkD4C2
— Nate Silver (@NateSilver538) November 4, 2016
The Evidence
Clinton's polling is MUCH weaker than Obama's in the swing states. People seem to miss this. https://t.co/EspnywJDcv https://t.co/DEARB8vqvU
— Nate Silver (@NateSilver538) November 7, 2016
Also, Obama had 49% of the vote, whereas Clinton has 45%. About 3x more undecideds.
— Nate Silver (@NateSilver538) November 7, 2016
The Reasoning
But mostly it's the Electoral College. Clinton underperforming Obama's 2012 forecast by *4 points* in the Midwest. That's a major liability. pic.twitter.com/inVXthqhez
— Nate Silver (@NateSilver538) November 7, 2016
People don't debate the premises of 538 model (e.g. state errors correlated, undecideds=uncertainty). They just don't like the conclusions.
— Nate Silver (@NateSilver538) November 7, 2016
The Counterargument
Actually, one exception—there's good, healthy debate about how aggressive a model should be about chasing down swings early in the campaign.
— Nate Silver (@NateSilver538) November 7, 2016
It's possible—I think likely, to be honest—that a model should be fairly conservative on swings early on but aggressive late.
— Nate Silver (@NateSilver538) November 7, 2016
The Audience
Nate Silver’s main audience is his 1.7 million Twitter followers, many of which it can be assumed read his writing and follow his analysis. He uses statistical jargon (“points,” “model,” “correlated,” “undecideds”) appropriate for such an audience.
Was this helpful? Visit our Elections Hub to find more ways to teach the election, or check out these previous #CERCAtheVote prompts:
- Michelle Obama Says Trump Comments Can’t Be Ignored
- The Chicago Tribune States the Case for Gary Johnson
- Clinton and Trump Accuse Each Other of Enabling Terrorism
Mallory Busch is ThinkCERCA's Editor of Content Strategy. A graduate of Northwestern University, Mallory came to ThinkCERCA from stops in audience strategy at TIME magazine and news applications development at Chicago Tribune and The Texas Tribune. She holds degrees in Journalism and International Studies, and was a student fellow at Knight Lab in college.