US Elections 2016 - Live Blogs - Al Jazeera English


US Elections 2016

  • What do polls actually tell us?

    By Tom Ackerman
    Six weeks before the election, American media are spending as much time tracking poll numbers as they are reporting the candidates’  latest pronouncements.
    As the race tightens in the homestretch, they can be forgiven for paying such close attention to the daily dynamics of the numbers.
    But which polls?
    Without much qualification, they cite a bewildering variety of surveys, some more meaningful than others.
    Take polls of registered voters. In 2012, registred voters totalled 146 million – but only 126 million actually cast ballots.
    Polls of "likely" voters should be more reliable, but the polling companies use varying methods to figure out who fits that definition.
    Then there are the tracking polls. They check back with the same group of people to gauge how they are changing their opinions over time.
    Push polls are conducted by campaigns or their allies whose purpose is to persuade voters by using loaded or manipulative questions. 
    The worst are online or snap polls, which are open to anyone and don't prevent multiple voting by participants trying to run up the totals for their favourite.
    (When one of Fox News' primetime hosts hailed Donald Trump's online poll performance just after his first debate with Hillary Clinton, a Fox executive warned that such "polls do not meet our editorial standards".)
    What the public is less likely to hear about are the campaign observers who take the long view – leaving aside the latest survey results to concentrate on the fundamentals of presidential elections.
    Yale University's Ray Fair uses an economic model to forecast a winner based on growth per capita in the four years before the election.
    Alan Lichtman of American University has devised a checklist of 13 "keys" to the presidency, combining economic, social, foreign and military indicators with the perceived charisma of the
    candidates and how long their parties have been sitting in or out of the White House.
    Alan Abramowitz, a political scientist at Emory University, has a "Time for Change" model which employs some of the same factors as well as the incumbent president's approval rating. (Obama is currently at 51 percent, slightly higher than the historical average for an elected president in the final months of his second term.)
    And then there's the Polly Vote, which combines a number of inputs including expert judgment, prediction betting markets and citizen forecasts.
    The Polly poll shows Clinton ahead on every count except for its econometric model, which gives Trump an advantage of less than one point.
    But the other prognosticators mentioned above now give the edge to Trump.
    So which to believe?
    Abramowitz, for one, is hedging his own scientific findings.
    He admitted to one interviewer that his model doesn't capture everything.
    "These forecasting models assume that you have mainstream candidates who will unify each party," he told the Los Angeles Times. "Trump doesn't fit that pattern. He's off the charts. And it's very hard to predict how that's going to play out."
    Abramowitz added that given Trump's sky-high unfavourability ratings (however close to Clinton's own negatives), he can't trust his own forecast fundamentals.
    So much for political science.
View the full stream
Powered by ScribbleLive Content Marketing Software Platform