Considering the states involved and the traditional Rasmussen "tilt" toward GOP candidates in their results, three point leads for McCain in Virginia and North Carolina don't suggest how "closely contested the November election could be." They suggest a Democratic landslide.

Posted at May 12, 2008 9:16 AM in response to Rasmussen: Virginia And North Carolina Looking Close For November

The Pew Research poll (out today) does seem to be a slight outlier in terms of Obama support. But that will be apparent (or not) as more data become available.

In any event, the first comment in this thread is utterly nonsensical. Not only does it refer to the wrong poll, (NBC/WSJ), the comments on the Taylor Marsh website suggest that none of the folks there understand the concept of "oversampling."

Suffice to say that when a poll "oversamples" a subgroup, it's for the purpose of being able to draw more reliable conclusions about that subsample. The results from "oversampled" respondents are then "weighted" to cancel the effect of the "oversampling." A single semester course in statistics is all that's required to understand such a technique. As long as (a) respondents are chosen randomly and (b) appropriate weights are applied to oversampled respondents, the effect is to diminish the poll's "margin of error."

Posted at March 27, 2008 6:29 PM in response to New Poll Shows No Real Imbalance In Potential Dem Defections To McCain

And that's why such a "huge swing" is so suspicious. Only a few days ago, Reuters reported a national Zogby poll that indicated a similar massive swing nationally in Clinton's direction.

The fact is that public opinion doesn't usually swing so wildly. More likely is a poll that is way out of whack followed by one that more accurately taps opinion, or vice versa.

Posted at March 25, 2008 4:45 PM in response to Poll: Obama Ahead 21 Points In North Carolina Primary

A few points.

() The basic point is correct. Obama does run "significantly" (statistically) better against McCain than Clinton does. What is missing from the analysis is the relative vulnerability of Obama to defections. His weaker support among traditional Democrats (compensated by his support among independents) and the greater likelihood that Clinton supporters would defect if he's nominated (See recent Pew Poll) suggests that Obama's nomination is higher "risk," just as Bill Clinton suggested.

() The clear implication of the SurveyUSA 50-state poll is that Obama's nomination puts many more states (and electoral votes) in play for Democrats than Clinton's. With limited resources, it's clear why the GOP would rather run against Clinton. The number (and size) of battleground states is significantly fewer with a Clinton nomination.

() Complaints about the accuracy of SurveyUSA's results are worthless. The data are what they are. I live in Washington State. Given McCain's role in the recent Air Force tanker deal, his chances of carrying Washington are only slightly better than Clinton's (or Obama's) chance of carrying Wyoming. Nevertheless, the data (in late February) suggest he's competitive in Washington. The landscape will undoubtedly change in a few months in a number of states. However, the SurveyUSA data provide a useful starting point for analysis.

Posted at March 10, 2008 9:13 AM in response to Obama significantly outscores Hillary vs. McCain

Given Rasmussen's usual tilt toward GOP candidates, this is important news. In 2006, for example, Rasmussen overestimated the actual GOP share of the vote in 19 of 21 US Senate races. The basis of their bias is unclear. (I'm not inclined to believe it's intentional.) It may stem from weighting their samples to predetermined partisan shares of likely voters, and the bias may have been corrected since 2006. If it has not, however, it's likely that Franken's actual numbers are 2-4% higher than shown.

Posted at February 19, 2008 10:06 AM in response to Rasmussen: Al Franken Takes Small Lead Against Norm Coleman

Does anyone else feel that the Clinton campaign would be far better off by locking Howard Wolfson in a closet somewhere?

Although I support Senator Obama, I hold Senator Clinton in high regard. That regard is tested every time I see Howard Wolfson speak on her behalf.

I'm not blind to the fact that any candidate's flack is obligated to recite the campaign's talking points. But Wolfson is so resistant to responding with anything resembling an answer to a question, I have a very difficult time listening to him.

Posted at February 18, 2008 2:39 PM in response to Hillary Campaign: Obama Plagiarized Speech From Supporter