Network News

X My Profile
View More Activity

Putting the Horse Race Numbers in Perspective

On an almost daily basis some media outlet or another releases a national poll that tests hypothetical 2008 presidential primary match-ups.

These polls invariably produce headlines like, "Clinton Slips, Obama Gains," "McCain Faltering Among Conservatives" or "Romney Fails to Gain Traction."

But is that what the poll numbers are really telling us? Not really.

Let's Parse the Polls!

We start with a quick example. Several polls released recently -- including one in The Washington Post -- showed Rudy Giuliani widening his lead over John McCain.

A CNN survey in the field from March 9-11 showed Giuliani ahead of McCain 34 percent to 18 percent. That same poll showed Giuliani leading 32 percent to 26 percent in January and 29 percent to 24 percent in December. The NBC/WSJ poll showed Giuliani with a 38 percent to 24 percent lead over McCain earlier this month; a far wider margin than the five-point spread that separated the two men in a December WSJ survey. In The Post poll, Giuliani went from 34 percent to 27 percent margin in January to a wider 44 percent to 21 percent edge in February.

Rudy widens lead, McCain slips, right? Well, sort of. Tuesday's CBS/New York Times poll shows Giuliani's lead over McCain actually shrinking over the past month from 21 points to nine points.

So, which is it? Is McCain slipping or gaining?

That question illustrates the widespread misuse of national primary polls at this early date in the cycle. Any pollster of repute will tell you that national polls are more telling in measuring the mood of the country than they are in predicting which candidate is up or down in the horse race.

There are several reasons for this. First and foremost, the presidential nominations for each party are decided on a state-by-state basis, not at the national level. So even though Democrat John Edwards trails far behind Hillary Rodham Clinton and Barack Obama in national polling, his regular first-place showing in Iowa polls means that he belongs in the first tier.

It also follows that at this early stage candidates with a national profile -- Clinton and Obama for Democrats, Giuliani and McCain for Republicans -- will fare better in national polls than those presidential candidates with regional or state bases like Mitt Romney or Bill Richardson.

Second, the samples used by nearly every major media outlet (Fox News is the lone exception) are comprised of adults -- not registered voters or likely voters. Campaign pollsters almost always use likely voters because their goal is to predict what the electorate will look like on Election Day, not test the overall mood of the country (nonvoters, registered voters and likely together).

Third, many voters are getting their first glimpses of these candidates, so their views can swing wildly over these early months. Take, for example, a voter living in New Hampshire who thinks of herself as a Clinton supporter but happens to be at a diner where Obama is speaking. If she is polled the next day, she might well say she is for Obama but a month later when Clinton (or Richardson or Edwards or even Dennis get the point) appears at a house party in her hometown, she may well go back to Clinton. Thus, polls taken even a week apart could show differing results while accurately measuring voters' opinions at that particular moment in time.

In short, the results of national polls are often misused -- by members of the damned media! -- to show incremental gain or loss in the horse race ballot question. But the true purpose of these surveys is to provide a broad atmospheric look at the state of the country.

Does this mean national polls are useless when it comes to analyzing the 2008 race? No. The best way to use national polls is to take the long view. Don't draw any conclusions from a single survey about which way the race's tectonic plates are shifting.

Instead, look at the trend in a particular poll or polls over a three- or six-month period to get the best perspective on the state of play in the party primaries. Remember that we are still more than 300 days away from the Iowa caucuses, so how a candidate looks today in a national poll isn't terribly predictive of how he or she will finish in the Hawkeye State and beyond.

By Chris Cillizza  |  March 14, 2007; 9:35 AM ET
Categories:  Parsing the Polls  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: Fix Picks: Abortion vs. Gay Marriage
Next: Edwards: Shrum was wrong

No comments have been posted to this entry.

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company