A footnote on the Newsweek poll

Drudge has flagged an interview disclosing the results of a forthcoming Gallup/USA Today/CNN poll of likely voters. The poll shows President Bush with a small lead and no Kerry bounce out of the convention.
The Drudge item serves to emphasize the point I sought to make yesterday in “The flubber bounce” about the necessity of looking at the survey sample to determine the screens used by the pollster in conducting the poll — adults, registered voters, or likely voters. In my view, polls of “adults” such as the Newsweek poll and the ABC News/Washington Post poll are worse than worthless.
Reader Greg Nesmith has kindly forwarded a link to an item from Polling Report.com. The linked item breaks down the sample of “adults” in the Newsweek poll that we discussed yesterday. Here’s the party affiliation of the sample:

Republicans 30%
Democrats 38.5%
Independents 31.5%

Query: Is anyone aware of an instance in which a member of the elite media has oversampled Republicans in one of these highly promoted presidential horse-race polls?
UPDATE: Here is USA Today’s story on the Gallup/USA Today/CNN poll: “Poll: No boost for Kerry after convention.”


Books to read from Power Line