Something odd with the latest CNN Generic Congressional Poll (R52-D45)

[FBaggins]

Their September poll showed republicans up by nine among likely voters and by six among registered voters. This poll shows a seven point gap among likely voters (no real change), but a 47-47 tie among registered voters.

The likely voter screening goes from providing a three-point bounce (net) to a seven point bounce?

http://i2.cdn.turner.com/cnn/2010/images/10/08/rel14a.pdf

[ProSense]

1. Whe the hell did CNN poll?

Obama is a better President than Bush by 2 percent?

Biden is a better VP than Cheney by 3 percent and Palin by 10 percent?

Ridiculous.

[FBaggins]

4. Some people have really short memories I guess. n/t

[Mass]

2. There are more Democratic voters in the RV of the Oct poll than in the Sept one.

My guess is that their likely voters include an enthusiasm factor that penalize Democratic voters because they appear less enthusiastic, even when they fit the other criteria of likely voters.

The biggest weakness of all these polls this year (whether they lean Democrat or Republican) is that their likely model is secret and therefore you cannot know which model are reasonable or not. Depending on the weight CNN/Opinion research puts on the enthusiasm factor, it may explain the phenomenon you describe.

[Gman]

3. Given the volatility of this year's elections

that volatility is showing up statistically. Bottom line, nobody can say anything for sure, including and specifically, Republicans taking back Congress. I think Democrat losses in the House are probably anywhere between 15 and 25 but no more than 30. And the vast majority of the losses are blue dogs.