Below the Belt: A Biweekly Column by NOW President Kim Gandy

The Perils of Polling

October 1, 2004

Stop the presses! Breaking news! MANY OF THE POLLS ARE MISLEADING! (And while we're debunking myths—there's no such thing as "security moms," but more on that later . . .)

Over the past two weeks, I've heard great despair over recent polls showing George W. Bush with a double-digit lead in the presidential race.

I'm here to tell you: TRUST YOUR INSTINCTS. Don't let polls or commentators or spin-doctors discourage you, demoralize you, or distract you by making you think the battle's lost. It ain't over till the fat cats sing, and I for one can't wait to hear Karl Rove in a high-pitched squeal.

Here is the truth about what you've been hearing:

1. Many of the polls are wrong and biased,
2. Many of the polls are misleading—what "security moms"?
3. Polls of the popular vote don't matter anyway—the electoral college is all that counts.

Many of the polls are biased.

You may have heard about the Sept. 13-15 Gallup poll touted by major media outlets, which gave Bush an astonishing 13-point lead. Want to know why? Gallup assumed that 40% of those voting in November will be Republicans and only 33% will be Democrats, with the rest voting Independent. However, in the last three national elections, more Democrats have voted than Republicans, which means Gallup is over-sampling one party and under-sampling the other party—a tactic likely to get anyone the survey results they want.

Other polls, which do not give Bush nearly as large a lead but do put him ahead, have problems of their own. Some of the errors include: undersampling young and low-income voters (who are hard to reach by telephone to answer poll questions); only polling "likely voters" and therefore ignoring the hundreds of thousands of new voters who are signing up in droves even as you read this column; and using telephone polling companies that do not call cell-phone users. This means that the millions of (overwhelmingly young) people who use cellphones as their primary phones are not represented (or under-represented) in the polls.

So here is a recipe for bias: Undersample Democrats, young people, immigrants, first-time voters and people who move often, and oversample Republicans, ignore the other data, and claim victory! Why not? Loudly claiming victory early and often worked well enough in the 2000 election. In the absence of opposition, power belongs to those who grab it.

At best, polls are weathervanes of public opinion, subject to both sampling and measurement error. At worst, they are statistics that become self-fulfilling prophecies. Remember the famous aphorism (alternately attributed to Mark Twain and Benjamin Disraeli): "There are three kinds of lies: lies, damned lies, and statistics."

Or recall W.H. Auden's 11th Commandment:
"Thou shalt not sit
With statisticians nor commit
A social science."

Social science is far more social than scientific, for the simple reason that people are not atoms or planets: our thoughts and actions do not always follow immutable laws like gravity and electromagnetism.

Polls can be used to say most anything the pollster wants, and woe to us if we assume that pollsters are above partisan politics.

Many of the polls are misleading.

The big talk lately has been about these so-called "security moms," and how the gender gap no longer favors Democrats. Let's take a closer look at these myths.

Myth #1: "Women are more worried about security than men."
Is this true? Of course! Women are more worried about their personal and economic security than men—with good reason. We are more likely to be the victims of violent crimes, and we still make only about ¾ of the income men do. We're more likely to be part-time or temporary workers, without health care benefits or job security—and in one of life's crueler ironies, at the same time we are more likely to be supporting children on our own! And with Bush and his cronies cutting funding for the Violence Against Women office, cutting overtime pay and making life harder for families on welfare, we get more worried by the day.

Of course women are concerned about security, but that doesn't translate directly to concerns about terrorism. The Sept. 26-28 Greenberg/Quinlan/Rosner Research poll finds that men are much more likely to make the war on terrorism a part of their voting calculus. Women are more worried about the domestic economy and (surprise!) health care. 56% of women want to hear more from the candidates on those subjects, compared with only 34% wanting to hear more on terrorism and national security. Men are more divided, with equal numbers (around 42%) wanting to hear more about the economy or more about terrorism.

Myth #2: "Women give Bush better ratings than Kerry on security issues."
Well, yes and no. It depends on how you phrase the question.

If you ask about "the war on terrorism," women as a whole prefer Bush slightly. But if you ask about "foreign policy," women prefer Kerry by six points. And if you ask about "being respected in the world," women prefer Kerry by 10 points.

Saying that women give Bush better ratings on security is like reading an article under a big headline proclaiming "BLONDES CAN'T NAME SENATE MAJORITY LEADER," only to find that most people in the country, no matter their hair color, cannot name the Senate Majority Leader. And that, in fact, a higher percentage of blondes came up with the right name than the percentage of non-blondes. The reality is that MEN give Bush better security ratings than women do across the board—and that's an example of the gender gap, not a false phenom of "security moms"!

Men prefer Bush on the war on terror by 27 points, on foreign policy by 17 points, and on being respected in the world by 10 points. Surprise, surprise, it is the men who really want a macho leader. Women (and especially unmarried women and women of color) see beyond the Texas swagger to the dangerous cowboy mentality behind it.

Finally, which women are you asking? Turns out, it makes a difference. Married white women with children (only about 26% of women) trust Bush somewhat more than Kerry on security issue, but that's not news! They're a conservative-voting group that has supported Bush all along—they voted for Bush in 2000 against Al Gore. Women on the whole support Kerry—so don't let them tell you there isn't a gender gap!

The popular vote polls don't matter.

As we found out (painfully) in 2000, it's not the popular vote that counts, it's the electoral college. As of Sept. 21, Zogby Interactive polls showed the electoral college breakdown favoring Kerry, but with thin margins in battleground states.

We must focus on getting voters—and especially women—to the polls. This election will turn on TURNOUT, pure and simple. Reports from all over the country show that new voters are signing up in droves in battleground states. New registrations have jumped this year and county clerks are hiring extra workers to process the forms.

There is a tidal wave of the disaffected—young people, peace activists, low-income women who usually don't vote, immigrants whose rights have been trampled on. They are taking the first step, they are signing up.

Now let's help get them to the polls! Visit www.10forchange.org to join in! or send us an email and let us know how you'd like to help!