Nate Silver made a name for himself by accurately predicting the winners of Presidential primaries using stats back in 2008 and 2012. Fast forward to 2016 and we once again try and predict winners, but has polling lost its touch? Just how confident are we in polling after Nate Silver’s catastrophic collapse in the UK Election in 2015?
Nate Silver fared terribly in the UK election last May: In his pre-election forecast, he gave 278 seats to Conservatives and 267 to Labour. Shortly after midnight, he was forecasting 272 seats for Conservatives and 271 for Labour. But when the sun rose in London on Friday, Conservatives had an expected 329 seats, against Labour's 233.
The fault, Silver claimed, was with the polling: "It’s becoming increasingly clear that pre-election polls underestimated how well Conservatives would do and overestimated Labour’s result," the statistician guru wrote in the wee hours of the morning. (He also overestimated the Liberal Democrats' result by roughly 20 seats).
But the problem went beyond the UK. "The World May Have A Polling Problem," Silver asserted. "In fact, it’s become harder to find an election in which the polls did all that well." Silver went on to cite four examples where the polls had failed to provide an accurate forecast of the election outcome: the Scottish independence referendum, the 2014 U.S. midterms, the Israeli legislative elections, and even the 2012 U.S. presidential election, where "Obama beat the final polling averages by about 3 points nationwide."
"There are lots of reasons to worry about the state of the polling industry," Silver concluded, citing a range of factors. "There may be more difficult times ahead for the polling industry."
So the statistician who gained national fame for correctly anticipating the outcome of the 2008, 2010 and 2012 U.S. elections was admitting failure.
Why are they failing?
Silver’s success was largely based on understanding how to read the polls, and by knowing which polls were worth reading, but now it seems they can’t be trusted. Opinion polls can be saying one thing on Monday, but it could have all changed by Tuesday.
Is the answer online? Social media is operating at a much quicker speed than polls done by phone but unfortunately, we have yet to fully harness what we can get out of this platform.
Using Facebook may help indicate what a particular group of people are thinking and which candidate they are leaning towards, but there is still a lot of detective work to be done before we arrive at that hypothesis.
Looking at candidates Twitter pages can show you the buzz being generated there in real-time. Each tweet displays the number of retweets and likes. And searching for the candidates’ names (in various combinations of a first and last name) will give you a sense of the frequency and tone of tweets about them.
USA Today has teamed up with Facebook to provide metrics on the amount of weekly buzz each candidate is generating on Facebook (likes, comments, shares and mentions). This data is not always as current as today, but a quick look at the candidate’s Facebook pages can give you a sense of their activity per post at the moment.
SocialMention.com provides real-time searches of publicly accessible social media posts. Click “microblog” to get a sense of buzz for any candidate search on Twitter (and a few smaller, but similar social networks). Similarly, you can click “blogs” to see the buzz there.
This is so relevant today because back in 2008 while the polls over the weekend before the vote suggested Barack Obama would win the New Hampshire primary, search activity for the candidates on Yahoo indicated a surge for Hillary Clinton and she ended up winning there that year. The bottom line is that social media and search engines provide voters with more tools than ever before for tracking the rise and fall of candidates and this is the space we should be watching and not the polls!