I'm overdosing on polls. One day you hear President Barack Obama is up by 5 over Republican Mitt Romney and the next day Romney is ahead by a point. None of it seems to make much sense.
So I asked a political consultant to help sort this out. He said he doesn't bother much with polls by big-name news outfits, even though they can generate headlines. Instead, he looks at the daily results from Gallup and Rasmussen, which run continuous tracking surveys based on rolling averages of a few days. Here's what their surveys showed Friday.
Rasmussen polls "likely" voters, which tend to vote more Republican. Even so, Rasmussen had Obama at 46 percent and Romney at 45. Gallup surveys "registered" voters. He had Obama and Romney tied at 47 percent.
Both surveys are reputable. Rasmussen, by the way, tied with Pew for the most accurate of the 2008 presidential race.
The other polls are less reliable, sometimes even zany. Too many of them under-sample Republicans. Take a Washington Post/ABC News poll that ran earlier this month. It had Obama leading Romney 49-48 among likely voters, and 50-44 among registered voters.
But turn to the key question — party affiliation — and you learn that the sample was 33 percent Democrat, 23 percent Republican and 37 percent independent.
Which means you take that poll with a large grain of salt.
The party breakdown of the voters who showed up on Election Day in 2008 was roughly 39 percent Democrat, 32 percent Republican and 29 percent independent. In other words, the GOP share of The Post poll was even below GOP turnout for an election that heavily favored Democrats.
Few people believe this year's Democratic share of the vote will match '08 levels, but many of these news-organization polls are using samples that try to replicate the 2008 electorate.
That makes little sense. Consider two groups that were key to Obama's '08 victory — African-Americans and young voters. After nearly four years of feeble economic growth, the enthusiasm of both has undoubtedly waned.
Obama cannot again offer himself as the shining hope who will transform our politics. Most polls give Republicans a big edge in enthusiasm.
Pollsters surround themselves with technical gobbledygook but what they do is more art than science. They try to forecast the makeup of the electorate that will show up Nov. 6 and then construct a random sample of respondents with those characteristics, not only in terms of party preference but all sorts of criteria such as gender, age, racial composition and urban/rural split.
Many pollsters are guarded about how they "weight" their samples — what my political consultant friend called "the voodoo." But flaws in the sample aren't always hidden in the voodoo; they can be glaringly obvious, as in The Post poll cited above.
A CNN/ORC poll earlier this month, for example, had Obama up 52-46. Dean Chambers at Examiner.com looked at the fine print and discovered that only 4.2 percent of the sample was made up of independent voters, a group that has been solidly in the Romney camp. The other proportions were 50.4 percent Democrat and 45.4 percent Republican.
Compare those proportions to Rasmussen's August party-preference survey, which revealed an electorate 37.6 percent Republican, 33.3 percent Democrat, and 29.2 percent independent or "other." Chambers figured that properly weighted, the poll should have had Romney ahead of Obama 53-45.
So where is the race now? Probably where it has been for some time. Obama is slightly ahead, but the prize is still up for grabs — perhaps why all these polls are driving so many people nuts.
E. Thomas McClanahan is a member of the Kansas City Star editorial board. Email at firstname.lastname@example.org.