Generic polling flaws

Susan Page:

...

A general election many Democrats assumed six months ago would be difficult to lose seems a little less of a sure thing.

At the beginning of the year, many polls showed a generic, unidentified Democratic presidential candidate thumping a generic Republican by close to 20 percentage points.

In the USA TODAY Poll taken April 18-20, however, McCain kept the contest against Obama and Clinton within the survey's margin of error. Obama led McCain 47%-44% among registered voters. Clinton led 50%-44%.

Republicans lined up more solidly behind McCain than Democrats did behind Obama. Nine of 10 Republicans backed the Arizona senator, compared with eight of 10 Democrats who supported the Illinois senator. Each got equal support, 8%, from members of the other party.

They split independents: 46% for Obama, 42% for McCain. (The divide in a Clinton-McCain matchup was similar.)

"Some of it defies the philosophy or ideology of John McCain and gets into John McCain the American hero, John McCain the maverick Republican, John McCain the antithesis of the Democratic and independent voters' stereotypes of Republicans," says Bill Carrick, a Democratic strategist based in California. "He looks much stronger in these polls than any other imaginable Republican nominee would be."

USA TODAY combined the responses in nationwide surveys taken over the past two months to get a large enough sample to analyze the demographics of some of McCain's surprising backers: Americans who disapprove of Bush. Those voters, who make up more than a third of McCain's support, presumably would be the first target of a Democratic opponent.

Almost all of them were white, and most were middle-aged. Many were blue-collar workers. Slightly more were male than female. Almost half were independents; one in five were Democrats.

...

A poll sponsored by the Democratic National Committee in March tested McCain's standing among undecided voters in 17 swing states. The bad news for Democrats was that it found 59% of them said McCain had some or a lot of appeal. The good news: After giving them "factual information about his record" — designed to raise questions about him — that number dropped, to 44%.

The DNC now attacks McCain's standing as a maverick in a daily barrage of e-mails, including a series labeled the "McCain Myth Buster." They spotlight examples of him changing positions or hewing to the GOP line — for instance, voting against the Bush tax cuts in 2001 and now calling for them to be extended.

McCain's high-profile alliances with Democrats, dry sense of humor and coziness with some of the reporters who cover him prompt some voters to assume he has less conservative policy positions than he does. A survey by the political arm of Planned Parenthood in 16 swing states found that 23% of women who support abortion rights and support McCain over Obama said they believed McCain agreed with them on abortion. Eighteen percent said, accurately, that he opposes abortion rights.

"As soon as people know where he stands on the economy, Iraq and health care, support for him drops off dramatically," Democratic National Chairman Howard Dean says.

...

There is much more. The invalid assumption of the DNC is that they can frame the debate about McCain's positions in a vacuum which assumes there will not be countervailing arguments. Democrats are prone to do that. In fact they get pretty surly when their opponents fight back and reframe the debate.

The flaw in the generic ballot is that it starts with a Democrat bias over sampling Democrats. It also uses registered voters instead of likely voters which also over samples Democrats. Then it leaves out the baggage that each of the candidates brings to the campaign. It has been a misleading indicator in almost every election. I think they only use it to make Democrats feel good in the run up to the race.

The polling on the war and President Bush can also be misleading. Republicans can be dissatisfied with how aggressively the war has been fought or Bush has responded to Democrat attacks. The question on the war should be framed as do you want to lose. If framed that way the Democrats will lose.