UnPublished Polls

Israel has parliamentary elections on Tuesday and today is the last day that public opinion polls can be published in that country. The idea, I think, is that polling results can reduce turnout, either of a confident majority or a dispirited minority.

Of course, the First Amendment prevents any such reporting ban in the United States. What we do have is agressive polling and poll reporting for presidential elections up until Election Day and then voluntary constraint among mainstream news outlets, which don't release exit polling for a state until voting there has stopped. Early exit polls, though, are leaked online.

Two interesting things happened within this environment during the 2012 presidential cycle. One is that Nate Slver was able to collect and analyze the profluence of political polling to correctly predict the outcome of the presidential race in all 50 states. So, quite unlike a public ban on poll reporting the weekend before an election, American voters knew the exact outcome if they chose to believe Silver's analysis.

The other interesting thing that happened in 2012 was that many refused to believe Silver's analysis. There was a widespread belief on the right that the polls were just wrong -- had to be -- and, as an example, a website called unskewedpolls.com popped up, offering adjusted poll results that showed Romney with a comfortable lead.

It's hard to know what motivated all this rejection of evidence -- it could have been real optimism or skepticism. It could have been cynicism too, and fear that too-accurate polling before Election Day would reduce turnout among a dispirited minority. With a ban on poll reporting impossible, one way to ally fears about turnout effects is to discredit the polls.