Episode 22: European Polling Disasters, French Presidential Election & Online vs. Telephone Polls

In this episode, we discuss recent European election polling misses and how their polling challenges are different than the polling challenges we face in United States elections.

Segment 1: Recent Misses for European Election Polls

  • Following the first presidential debate in France’s upcoming Presidential election and with the first round of voting only a month away, U.S. interest in the French election is only going to increase. HuffPost Pollster is now tracking polls for the first round of voting, showing what is likely to be a run-off between Emmanuel Macron and Marine Le Pen which Macron is projected to win easily. While the outcome of the election will certainly be worth watching to determine the true extent Europe’s populist wave, we want to take this opportunity to discuss some of the recent misses in European polling, and what we can learn from them.
  • If the aftermath of last June’s Brexit vote (51.9% leave/48.1% Remain), a lot of British news coverage focused on whether the polls had been wrong again, suggesting that it wasn’t the first big miss for British pollsters. The first example that is generally cited is the Scottish independence referendum of 2014, which polls consistently showed to be closer than it actually was. The final vote was 55.3% opposing Scottish independence, 44.7% supporting it. Nate Silver has a good breakdown of the Scottish independence polling here. This miss was followed the next year by the 2015 U.K. General Election, in which polling suggested that the two parties would finish neck-and-neck yet Conservatives won an outright majority, or 36.9% to 30.5% of Parliament’s 650 seats.
  • The miss in the 2015 General Election spurred an extensive postmortem by the British Polling Council, which pointed to several key groups who were underrepresented in the polling: voters over age 70, young non-voters, and busy voters, and explained that there is “no single, straightforward fix for not having the right sort of people in your sample.”
  • Clearly, these mistakes were not fully corrected in time for the Brexit vote, though what is interesting about the Brexit polling is the clear difference between online and telephone polls. As detailed here by YouGov’s Editor-in-Chief Freddie Sayers, a large majority of online polls predicted the eventual vote to Leave, whereas an even larger majority of phone polls had Remain leading. Sayers states unequivocally that the debate has now been settled: Online polls are more accurate than phone polls.

Segment 2: Online vs Telephone Polling

  • What about here in the United States? Should there be a substantial move toward more online polling at the expense of telephone polls? Not so fast. The most prominent online polling companies, Survey Monkey and YouGov, each predicted easy Electoral College victories for Hillary Clinton, including some pretty bad misses. That said there are of course some advantages to online polling. It’s just our opinion that they fall short in their attempts to interview a completely representative sample of the population.
  • This is one of the first places to look when analyzing where any poll went wrong: Were they interviewing a representative sample of the population? If they were, the next question to ask is: Were they accurate in their projection of what the electorate would look like? For an accurate poll, both of those criteria need to be met.
By | 2017-05-24T03:25:47+00:00 March 24th, 2017|