Too close to call:
Why presidential polls can't predict the winner
(This opinion piece first appeared in Alternet)
By Harry G. Levine and Craig Reinarman
Four years ago, on October 27, 2000, CNN reported that polls
from Time, Gallup, ABC News, USA Today, and the
Washington Post found that Bush was ahead in the popular
vote. Some said Bush had a "solid advantage."
Republican presidential nominee George W. Bush holds a 49-to-43
percent edge over Democratic rival Al Gore in the latest CNN/Time
poll.... The poll of 2,060 adult Americans ... is thus in essential
agreement with a CNN/USA Today/Gallup tracking poll also released
Friday. That poll gives Bush a 52 percent [to] 39 percent edge
over Gore. More important, both polls show the same snapshot
of the current state of the presidential campaign: a solid advantage
for Bush. ABC News and the Washington Post both have daily tracking
polls today putting the race at 48 percent for Bush and 45 percent
Eleven days later Gore won the popular vote by more than 500,000
votes. Four years ago, shortly before the election, these major
polls were wrong.
The 2000 polls could not correctly predict the winner because
they could not accurately predict the effect of the substantial
"get out the vote" efforts, especially by labor unions
and Democratic organizations in important swing states. Bush's
supposed "solid advantage" in the popular vote did
not actually exist. It was a fabrication of the polls.
The 2000 election was essentially a tie. Out of a total of over
a hundred million votes cast, the difference between Gore and
Bush was one half of one percent. No poll of a few thousand
potential voters can accurately predict who will win the popular
vote in such a close election.
Nor can pollsters accurately predict who will win the electoral
vote, because the margin of victory in a decisive swing state
might be a statistically insignificant few thousand or even
a few hundred votes. In 2000, Bush won Florida by 527 votes
and New Hampshire by 7,200--and Gore won Iowa by 4,100 votes,
Wisconsin by 5,700, Oregon by 6,700, and New Mexico by 366 votes.
The combined margin of victory in these six states was about
25,000 votes--less than three hundredths of one percent of the
total presidential vote.
Fast forward to September 18, 2004, when the New York Times
reported that a Gallup poll found Bush leading Kerry 52 to 44
percent among registered voters, and 55 to 42 percent among
likely voters. A New York Times/CBS News poll also had
Bush over Kerry "51 percent to 42 percent among likely
Sound familiar? Since the first presidential debate, some polls
have showed Kerry and Bush as even and suggested the election
is too close to call. This is the only trustworthy prediction
that anyone can make.
Only in "safe" states where leads are large can polls
reliably predict the winner. Yes, Kerry will win Illinois, New
York and other "blue" states, and Bush will win Texas,
Indiana, and other "red" states. But in this election,
and especially in the crucial swing states, it is not the few
undecided voters but the many new and returning lapsed voters
-- the unknown voters -- who will make the difference.
In 2004, Democrats are energized and engaged like never before.
Several independent get-out-the-vote campaigns -- especially
ACT (Americans Coming Together) and MoveOn.org -- have been
working in the swing states to increase the Kerry vote. Some
individuals are even getting out voters on their own. For example,
Frank Phillips, an indefatigable 90-year-old retired businessman
from New Hyde Park, New York, recently organized friends and
neighbors to call voters in Florida. Now he is arranging calls
to 3,000 more in Ohio.
Syndicated columnist Mark Shields recently reported that in
Ohio "one of the smartest Republican professionals I know
... confided that he feared the GOP 'needs a 5-point lead in
the polls heading into Election Day' to counter what he sees
as 'the Democrats' intensity' and organizational commitment."
Because the margins are so small, unpredictable events like
stormy weather on election day could also effect the outcome.
Nader's campaign remains a wild card; a few thousand or even
a few hundred votes that Nader takes from Kerry could tip the
electoral votes of one or more swing states to Bush -- as Nader's
newfound Republican backers understand.
In this election, the only certainty is that turnout on November
2 will matter enormously.
And even after the election, if a poll seems to have predicted
the winner, remember that it had a 50 percent chance of being
right. Which makes even more impressive the mistaken predictions
based on the polls four years ago. Coin tosses would have done
better than they did.
Craig Reinarman is a professor of sociology at the University
of California, Santa Cruz. Harry G. Levine is a professor of
sociology at Queens College, City University of New York.
Return to Front Page