Did they Intentionally Create a Story? One of the Newsweek Polls was Wrong.

October 5, 2004
By

The following was written by Polipundit reader and guest commentator, Oak Leaf:

Yesterday, Rasmussen released polling that confirms that President Bush is maintaining a lead, after three nights of post debate polling. Rasmussen polls 1,000 likely voters and unlike Newsweek, maintains some stability on their sampling mix. Based on Newsweek’s past performance, they would probably see nothing “wrong” with a nationwide poll of 1,000 voters that did not have a single Republican in the sample and comparing it to a prior poll.

The Newsweek polls of October 2, 2004 and September 11, 2004 provide breakdowns of partisan support, i.e. 89% of Republicans support Bush, which would be useful in determining where the race is going. This is the only useful aspect of these polls in determining trends as it is a constant.

Some interesting points in the partisan support are that Bush has seen almost a doubling in democrat support from 7% to 12%. Also there is almost a doubling of independent undecideds. Bush loses four percent of his base and that is where he will have to work. Remember, this is post debate.

Newsweek states, “the debate erased the lead” which we know now is a complete fallacy based on the almost six percent swing in voter registration demographics between the two polls. If we were to accept Newsweek’s “logic” as correct, that would imply that in the State of Florida, 570,000 registered Republicans changed their registration to Democrat during the last thirty days. One of the two polls is “wrong” and they cannot be compared to each other without adjusting the sample.

I would enjoy hearing from Newsweek as to which poll more accurately reflects, in their opinion, the partisan breakdown of this country. They have certainly given us lots of choices. This data however, allows us to determine various hypothetical voting scenarios.

If we assume that the October 2, 2004 poll reflects correct registration demographics, but the September 11, 2004 poll reflects correct partisan support, the race is Bush 46%, Kerry 46%, Nader 2%, Undecided 6%. This addresses the issue that the September poll had “too many Republicans.” This is useful to see if the “debate” caused a shift in support. This exercise shows the “change” in the race was caused by altering the sample not by the debate. Newsweek created their own “news.”

If we assume that the October 2, 2004 poll reflects correct partisan support, but the September 11, 2004 poll reflects correct registration demographics, the race is Bush 50%, Kerry 41%, Nader 2%, Undecided 7%. Probably unlikely.

If we assume that the October 2, 2004 poll reflects correct partisan support, but the registration demographics are equal among Republicans, Democrats and Independents, the race is Bush 46%, Kerry 45%, Nader 2%, Undecided 7%.

For those interested, the email addresses for Newsweek are: [email protected] and [email protected]

Comments are closed.