Why did the pollsters miss Beppe Grillo?
Over the weekend, I predicted that Beppe Grillo and his Movimento 5 Stelle (M5S - Five Star Movement) would outperform the polls by four to five percentage points in the Italian general election. I didn't, however, expect him to secure one in every four votes.
What is going on? In general, polling organisations are getting more accurate, and on the whole, they get it right. Nate Silver correctly predicted all 50 states in the 2012 US presidential election. But an average of polls calculated before the polling blackout (Italy does not publish polls two weeks before an election) puts the M5S at somewhere between 15 and 16 per cent, a massive 10 percentage points below the final result.
Grillo causes a big problem for polling companies. While mainstream candidates tend to end up close to predictions, new, populist, and radical parties often confound them. Their vote is fickle - many voters change their minds on election day. Far-right parties like Marine Le Pen and her radical right Front National often get a curtain bonus, where people are embarrassed to admit their preference to a polling company, but will cross the box in the safety of the booth. New upstarts often do well in the polls, only to see the electorate opt for the devil they know.
Grillo is different: he is a comedian, who refuses to speak to the Italian media. He couldn't stand himself, because his own movement's rules don't allow him to because he has a criminal record. The reason he has done so well is his ability to get his vote out. With turnout low across the board - bad weather probably contributed to that - his share of the vote increased.
The reason his vote turned up was social media. Grillo has, by an enormous margin, the largest social media following in Italy - in fact, in Europe. He has over one million Facebook friends, and a similar number of Twitter followers (Bersani has about a quarter of that). And he uses this huge social media profile to make things happen offline.
Grillo has constantly encouraged his supporters to discuss the issues he raises on his blog as they relate to local questions in their cities and towns in 'meet up groups' - and there are hundreds of them. According to our report, based on a survey of his supporters, he has around 250,000 people who consider themselves members of the party: an army of volunteers and door knockers that used to take years to recruit.
This election was a litmus test that many politicians are asking: can social media campaigning and support translate into actual votes? The answer is a big, five star, yes.