US elections
How to read US election polls correctly
The flood of opinion polls for the US presidential election is often more confusing than informative. Therefore, the DW asked two experts for help – both have a simple tip.
DW: every Day new polls of the US presidential election can lead to citizens overwhelmed instead of informed feel. How can you quickly decide whether a survey is relevant or not?
Ken Goldstein: The answer is not to try it in the first place. Even for a “Nerd” like me, it is sometimes difficult to assess a single survey. Therefore, the theory is to focus on the average values from surveys, and not to be too excited about a single poll, too happy, too sad, or calm. There are three large providers, which publish average values in the case of election polls, namely
Pollster,
Real Clear Politics and
Fivethirtyeight. They all use a similar, but slightly different methodology, and summarize all of the surveys in the Federal States, such as country-wide surveys.
Sam Wang: The easiest way is to avoid individual surveys. Instead, you should focus on three surveys with the same question, which has been roughly in the same period collected. Then one forms the mean value thereof, the so-called Median. For example, if the three polls show that Trump leads with two percentage points of Clinton with three percent, and Clinton leads with seven percent, then the most accurate estimate is that Hillary Clinton is up three percentage points in the lead.
There are different types of surveys: for example, surveys on the state or national level, telephone and Internet surveys, as well as such among likely or registered voters. What are the important categories to consider in surveys?
Ken Goldstein: I have very specific preferences for certain methods and survey Institute. But I think it is unrealistic to expect that someone who is pursuing this just for interest’s sake, will deal with it. Therefore, you should values on the average and the total image focus. While the Republicans, Clinton generally support Trump and the Democrats in General, it is important to know that the whole of the volatility seen in the polls, comes from the group in the middle: the undecided voters. Therefore, I pay particular attention to – and this should also do all that the American policy want to keep track on the behavior of these voters in the middle, where both candidates are very miss.
Sam Wang: country-wide surveys, as long as one has enough to form a mean value, the last moments of the race to reflect. Even better combined surveys from the individual States. When
Princeton Election Consortium, we show a simple Overview of the last States surveys, and what this means for the decisive Electoral College (Electoral College). We measure so easily the current temperature of the race. It is the most transparent snapshot,.
A few people (and polls) have said the result of the British Brexit vote and Trump as a Republican presidential candidate before. How reliable polls for the American presidential election? Or to put it differently, we should be in for a Surprise in November, ready?
Ken Goldstein: There was actually some survey disaster. The Scottish Referendum, the election in Israel, the Greek Referendum, all of these were cases in which the average values of the surveys were in the wrong. But the polls to Trump in pre-election were not wrong. The Problem wasn’t the polls, the Problem was that the people have not believed the polls.
The irony is: In a time in which it is so difficult to make surveys as never before, are there so many polls like never before. Where the polls can be wrong is if you incorrectly judge, who goes to the election. If the polls are wrong, this is so usually the fact that they assess the electorate overall, wrong. The big question is whether the young people and the black Americans go to the polls to vote against Donald Trump, because you do not choose to support Hillary Clinton.
Sam Wang: your question is completely wrong. Surveys prior to the Brexit vote showed a head-to-head race. People who believe that the polls would have said something different before, are wrong. The surveys were not to blame. The surveys have also said the nomination of Donald Trump before. On the basis of the present evidence, I assume that the polls will be in November, as properly.
Ken Goldstein is a Professor of political science at the University of San Francisco. He is also as an election consultant for ABC News and has accompanied every U.S. national election since 1988.
Sam Wang is Professor of molecular biology and neuroscience at Princeton University. He also runs the Princeton Election Consortium, the Meta-analysis of choice surveys, a specialized Blog.
The Interviews conducted by Michael Knigge.