Sift: October 27
The Unlikely Truth about Polls
At about this time in every campaign,
failing politicians are saying something like: "I don't believe in the
polls. The only poll that counts is on Election Day."
This year you have to wonder if they might be telling the truth: Should we believe these polls?
The worrisome thing is how diverse they are. On Thursday, for
example, the
IBD/TIPP national poll had Obama's lead shrinking to a mere 1.1%, while the
CBS/NYTpoll
it growing to 13%. Something like that happens almost every day. How is
that even possible? If the polls are all measuring the same underlying
reality, shouldn't every poll be within the margin-of-error
of every other poll?
It turns out that's not what margin-of-error means. A poll's MOE is really only the margin of one kind of error, known as
sampling error.
The other kinds of error are nearly unmeasurable (at least until
there's a real election), but that doesn't mean they're zero.
Sampling error amounts to this:
Ifthe pollster does everything right -- chooses a truly random sample of
the electorate, questions them in a way that doesn't influence their
answers, and so on -- and
ifthe respondents don't lie, then the results still might be off just
because of the luck of the draw. You might happen to pick a few too
many McCain or Obama supporters in one day's sample, just as you might
flip ten coins and get four or six heads rather than five.
But sampling error is just the beginning of what might go wrong with a
poll.
Today,
pollsters do lots of tricks with the data -- tricks that are supposed
to make the poll more accurate. One of those tricks is called
re-weighting:
If you look at your sample and it conflicts with known demographic
data, you make a correction. So if your sample is 55% male and the
electorate is only 49% male, then you might count each male
response as slightly less than one vote to make the overall sample
representative. Or you might re-weight by age or race or income.
Different polls juggle the numbers in different ways, and so they might
report different results
even if they interview the same people and get the same responses.
A particularly difficult thing to guess, especially this year, is who
is actually going to vote. That's what polls do when they report
results on "likely voters". Different polls have very different ideas
about what makes a person likely to vote. Nate Silver, the polling wonk behind 538.com, gives us some much-needed
insight into the likely-voter modelsof various pollsters. He notes, for example, that on Wednesday both CNN
and ABC showed Obama with a 10-point lead among registered voters.
Among "likely" voters, though, Obama's CNN lead shrunk to 5%, but was
still 9% on ABC. That 4-point difference amounts to CNN and ABC having a disagreement about who is going to vote.
In most years, the likely voters are more Republican than the merely
registered voters. The relatively powerless get discouraged about
voting, but if they vote, most of them vote Democrat. This year,
though, Obama has invested massively in what the pros call the "ground
game" -- registering new voters and making sure they get to the polls
come election day. If that works -- and it did in the primaries -- then
any previous cycle's model of likely voters is probably wrong.
Silver suggests a very simple check pollsters could do in this era of early
voting: Ask people whether they've already voted. If a large number of
people that your model says are "unlikely" have in fact already voted,
then you need to fix your model.
Because of this modeling uncertainty, Silver advises taking any single
poll with a grain of salt. If it looks like an outlier, he says, it
probably is.
"Not Because They're Racists"
The weirdest event of the week was that young white female McCain
worker in Pennsylvania who made up a story about being attacked by a
big black male Obama supporter. I think the whole story is sad and has
little to do with either McCain or Obama, so I'm not going to repeat
the woman's name or link to a picture of her. But there is one person
here who deserves to be singled out for shame: John Moody, the
executive vice president of Fox News.
Writing on the
Fox Forum blog, Moody called the incident a "moment of truth" and a "watershed event" for the presidential race:
If [the woman's] allegations are proven
accurate, some voters may revisit their support for Senator
Obama, not because they are racists (with due respect to Rep. John
Murtha), but because they suddenly feel they do not know enough about
the Democratic nominee.
If the incident turns out to be a hoax, Senator McCain’s quest for the presidency is over, forever linked to race-baiting.
Both of these assertions are perverse. Let's take the second one first.
The incident turned out to be a hoax, but the strange idea that
McCain's quest for the presidency is over now has quietly vanished at
Fox. Moody has not even done a follow-up what-was-I-thinking blog post.
Now re-read that first paragraph. If some black man in Pittsburgh
really did attack a white woman, that would cause some voters to
"revisit their support for Senator Obama."
What other than racism could possibly connect those dots?
But of course, open racism is not acceptable in America today, so Moody
offers his readers a rationalization: If this incident makes you want
to vote against Obama, you're not a racist, you just "don't know
enough" about Obama -- as if there were something you could "know"
about Obama that would disconnect him from this incident that your
racism has connected him to.
Short Notes
You shouldn't let desperate politicians distract you this week, but if you're
looking for a way to distract yourself, try Neil Gaiman's new novel. Gaiman is a
writer of twisted fantasy for all ages -- just not all ages at the same time. He
has written for children (
The Day I Swapped My Dad for Two Goldfish),
youth (
Coraline), and grown-ups (
American Gods), and his 1990s
series
Sandman is legendary among comic-book readers. Well, what would
happen if such a writer did a homage to Kipling's
The Jungle Book?
The
Graveyard Book, of course: Nobody Owens is an orphaned toddler who is
adopted and raised by the ghosts from an ancient and all-but-abandoned English
graveyard. With the help of a protective vampire and werewolf, the dead keep
Nobody safe from those truly scary beings, the living.
If Sarah Palin can see Russia, maybe the Russians are watching her too. Vlad and Boris sing of their unrequited love in "
Song for Sarah".
I've told a lot of stories about McCain/Palin supporters behaving badly, so I should point you to at least one where
they behave well.
The
Chicago Tribuneprovides a little local perspective on Bill Ayers. Eric Zorn found 60
local-news references to Ayers during the 90s. "the record shows he
just wasn't a very controversial figure." Mike Royko thought he was a
jerk, but associating-with-jerks isn't much to smear a guy with.
One of cutest get-out-the-vote tactics I've ever seen: A customizable
video blaming an individual voter for Obama's one-vote loss. Send them to your friends.
