Facebook Twitter

Pollsters as divided as the rest of nation

Who’s ahead? Web sites tell different stories

SHARE Pollsters as divided as the rest of nation

In the mood to decipher the indecipherable or explain the inexplicable this political season? Try reconciling two poll-tracking Web sites. Each of the two claims to offer a peek inside the minds of American voters coupled with a not-so-veiled prediction of who is most likely to occupy the Oval Office next year.

I've been driving myself batty (according to my husband, a short drive) for several months, "motoring" as it were to rasmussenreports.com. This Web site's daily and rolling tracking polls provide not only national numbers, (to which so-called experts this year are paying scant attention) but also tallies of voters' inclinations in the all-important battleground states. The site then projects the likely outcome of the prized electoral college vote. More recently, I found electoral-vote.com, which unlike rasmussenreports.com, does not produce its own polling but collects and cites a variety of other polls.

As of this past Sunday, here's how the two Web sites were calling the presidential race. Rassmussenreports.com does not describe itself as Republican-leaning but owner Scott Rassmussen has been widely identified in the media as a Republican pollster. Its latest nationwide tracking poll revealed President Bush drew the support of 50 percent of likely voters and Sen. John Kerry drew 46 percent. Rassmussenreports tallied the electoral college votes as follows: Bush 240, Kerry 179. With 270 the magical number for victory, rassmussenreports.com projected no clear winner, but declared the president way ahead of the challenger, Kerry.

Electoral-vote.com, whose creator is a self-described Kerry supporter, has Kerry at 270 electoral votes (that magic number again) and Bush a "fer piece" back at 248.

We've all come to expect deviation in polls. Last month several national polls on Bush's "bounce" coming out of the Republican National Convention showed him ahead of Kerry by spreads ranging from 2 to 12 points. That's a chasm-sized gap, given pollsters are all supposed to be polling similar representative samplings of likely or registered voters.

But the difference in battleground state leanings as proclaimed on these two Web sites is even greater.

What explains the difference? Pollsters offer a number of explanations.

First, the same exact question asked in different order, or with the modification of one changed word, can produce polar opposite results. Go figure.

Then, there's this season's cachet of the "undecided voter." People who are not truly undecided may still claim to pollsters that they are, as a way of primping for additional media attention. After all, it's the "undecideds" who are starring on national television, in thoroughly scrutinized and fawned-over focus groups, etc. It's the "Hi mom, look at me," suddenly famous syndrome.

Then there are new oddities in the way pollsters poll and changes in voting patterns that combined may be throwing the precision of polling way off its usual formulaic goal.

These are as follows. Pollsters tend to cull registered voters and likely voters (the people they prefer to poll) from well-worn lists of phone numbers they've used in the past. Pollsters traditionally rely on land-line home phone numbers. The trend among younger voters (the 30-and-under crowd, which is to say, a sizable segment of the U.S. population) is away from land-line home phones, toward total reliance on cell phones. Cell phones are unlisted in telephone directories. Therefore these voters' opinions, the belief goes, are not being adequately represented in pollsters' samplings.

Then, there's the huge push by both parties to register tens if not hundreds of thousands of new voters per state, especially in battleground states. These voters, too, do not generally appear on pollsters' lists. So their opinions may not be fairly represented.

Democrats claim to have registered larger proportions of new voters than Republicans. But then again, just because people register does not necessarily equate with hordes of new voters trudging to polling places on Election Day. Still, this change in registration patterns may be enough to throw the reliability of polls into question.

In sum, there are more questions than answers this election season about the reliability of polls. Where we end up is, not only is the American public deeply divided but our pollsters are, too.

Bonnie Erbe, host of the PBS program "To the Contrary," writes this column for Scripps Howard News Service. E-mail bonnieerbe@CompuServe.com.