AP gives exit polls the boot

Exit polls, the notoriously inaccurate in-person survey of voters at polling places, could be vanishing under a plan announced today by the Associated Press.

The AP says it will replace its exit polling system after 2016 exit polls appeared to favor Hillary Clinton.

The announcement is significant because the Associated Press pools resources with ABC, CBS, NBC, CNN and Fox News for exit polling.

The results were so inconsistent that the news organization’s Washington bureau chief threw them out on election night and ordered reporters and writers to use only actual election returns.

Instead, the AP use a combination of online polling and telephone calls to sample voters, it said.

The poll’s methodology allows for results from every state holding a statewide election, Scott said, as well as details about the opinions of registered voters who elect not to cast a ballot. AP’s approach will deliver to customers more reliable information on what drives the choices of different segments of the electorate than is available from traditional exit polls, Scott said.

Unlike the exit poll, VoteCast won’t use people with clipboards seeking to buttonhole voters after they leave polling places, an approach AP argues is no longer appropriate in an era when 40 percent of the electorate votes early, absentee or by mail. That percentage is growing in every election, Buzbee said.

There’s also concern that in-person exit polls, in a polarized political climate, fail to capture the opinion of all voters. In the roughest years for the accuracy of exit polls, 2004 and 2016, the surveys showed a stronger vote for the Democratic presidential candidates than actually took place.

AP said it successfully tested the approach that would become AP VoteCast in three statewide elections last year. Among them, a special election for U.S. Senator in Alabama in which the poll predicted Democrat Doug Jones would beat Republican Roy Moore 50 to 47 percent. The actual tally was 50 to 48 percent.

FoxNews and the Washington Post are joining in the effort.

ABC, CBS, CNN and NBC will continue to use the traditional exit polling methods.

  • Jack

    I would prefer to just get actual election results.

    • What exit polling CAN do is explain a WHY. Maybe if we had a good “WHY”, we wouldn’t STILL be wasting good reporter time almost two years after the election, sending people off to wherever to ask that question.

      I notice 1-A is going to fill another hour tomorrow with “Trump voters, tell us about what you were trying to say in 2016”. Yawn.

      • Rob

        Yes! Rebecca Solnit, a writer and contributing editor at Harper’s, posted an essay at Literary Hub recently regarding the troubling set of media assumptions about who matters and whose story gets told. No surprise, it’s Protestant white dudes. She contends it’s not the media’s job to understand or obsess about Trump voters. And as a sustaining member of MPR, I am way past tired of programming that focuses on getting into the heads and psyches of Trumpists.

        • Al

          For the record, this white Protestant is kinda bored of hearing about other white Protestants.

          • Rob

            : )

      • Jack

        I honestly don’t remember – did the journalists dissect the Ventura win much? If not, was it because Jesse was not too radical at the beginning (pre-media Jackal times)?

        Also – to clarify my original post – tired of having exit polling affect voting in late poll-closing states. Everyone should be able to cast their vote thinking that the candidate of his/her choice still has a chance.

  • >>The AP says it will replace its exit polling system after 2016 exit polls appeared to favor Hillary Clinton.<<

    To be fair, she DID win the popular vote by several million votes…

    • Well, that’s a problem I have with polling in general, of course. Polls are presented in a fashion that isn’t the way we elect people. It doesn’t make sense.

      Aside from that, though, the exit polling is micro.

      • wjc

        Exactly. We have 51 presidential elections, not 1.

      • Postal Customer

        538 practically had a year-long series on why Trump won. There was a nice, satisfying heaping dose of disdain for polling, and more importantly, the completely dead-wrong ways the media interprets and reports on that polling.

  • lindblomeagles

    The AP is trying to duck media critics, but they technically got the 2016 Election Correct. Hillary Clinton won the POPULAR vote by 2.9 million people. That’s a pretty good margin of victory for a source that conducted scientific polling predicting a solid win for Hillary Clinton. The problem is we elect Presidents via the Electoral College, and that Electoral College is skewed to some degree towards larger states. Trump won most small states and a few large states, and those states Trump did win honestly cannot be ascertained through an Exit Poll. But Trump’s win IS NOT unusual. Our reaction, including the AP’s, is hostile because Trump is hostile to the media, to loses, to imperfection. In 2000, AL GORE DEFEATED GEORGE BUSH by a smaller margin of 543,000 votes. But Bush won Florida’s electoral votes, sending Bush, not Gore, to the White House. John Quincy Adams, Rutherford B. Hayes, and Benjamin Harrison, Presidents who served the U.S. in the 1800s, also LOST the popular vote, but became President because they won the ELECTORAL college. We probably are not seeing a long term trend where every election will be a narrow victory. Instead of reacting to the master of reactors (one Donald Trump), continue the science behind polling and educate the public about the Electoral College in the Presidential Selection Process.

    • We don’t elect presidents by popular vote.

      This is like evaluating a baseball pitcher with a 4.50 ERA and saying, “well if we threw out that game that he gave up 10 runs in, his ERA would be among the league leaders.”

      The purpose of the poll is to analyze a presidential election in the way the election is held, not the way we might want it to be held.

      • lindblomeagles

        I’m not sure, but I think you misread my post. My fourth sentence above says we elect Presidents through the Electoral College, a point I later emphasized after providing lengthy Presidential Trivia. So, I’m not sure where (or how) you thought I was thinking something else. It is true that some polling is good, and that there is a science (math as well) to it. For example, governors, senators, and house of representatives are NOT elected by an electoral college. Polling for dismantling the “Affordable Care Act,” or another federal policy for example may have been valid. It is a fun and useful tool to teach or get students excited about math and social science. But, yeah, Trump is only the 5th President to win the White House without the Popular Vote, and overall, AP Polling results going back some 40 years have been fairly accurate. Exit polls are not the most accurate predictor of who wins the White House, as I clearly stated above, but we and the AP are over-reacting to this one UN-isolated event. 5 out of 45 Presidents failed to win the popular vote. 1 in 9. That’s not a bad average Bob. The AP was wrong. Electoral College determines the President. And we can reasonably assume, for the most part, 1 out of every 9 Presidents won’t win the popular vote.

        • Polls are never a good predictor and the AP doesn’t use them that way. They are an explainer of the present, not a predictor of the future.

          • lindblomeagles

            You did it again Bob – you didn’t read my post. 6th line from the bottom, I said, “Exit polls are not the most accurate predictor of who wins the White House.” None of us has a crystal ball to the future, including the new system you want implemented in place of exit polls. The point is Americans didn’t have a problem with exit polling until Trump won in 2016. We would not have this conversation at all if Hillary Clinton won. He’s unpopular. He came out of nowhere to win. He broke a lot of molds. But his win is no more a predictor of what is going to come than the Exit Polls we’re trying to get rid of. And just for trivia sake, one Trump like lightning rod LOST the Oval Office even though he was popular, and that guy was none other than Andrew Jackson, who, for his day, was seen as breaking the mold. Still not convinced we’re over-reacting. Rutherford B. Hayes and Grover Cleveland lost the popular vote but won the White House in 1876 and 1888 when the South was trying hard to defeat Reconstruction and industry was abusing its power over blue collar workers. The next unpopular President arrived 112 years later (George W. Bush). 16 years later, another unpopular President (Trump) won. These things are cyclical. Let’s just relax about the polling and figure a way to beat Trump in 2020.