Some Possible Issues With Newsweek Poll

While I don't doubt that Huckabee is surging, there's a few grains of salt to take with the latest Newsweek poll showing him at 39 to 17 for Mitt. And, yes, it was before the speech. But what's strange is that compared to previous polls, Huckabee would have had to draw more than 1 - 1 from Mitt, while Thomspon and McCain weren't touched and Paul picked up 3 points. South Carolina has shown something different, where Huckabee cut Fred's support dramatically. It doesn't make sense that he wouldn't touch him in Iowa. That brings us to the sample.

Yes, they're touting a sample of 1,408 people, but that's not really accurate. In terms of possible Republican caucus goers, you're looking at an MOE plus/,minus 7. Most reputable polling organizations would try to stay away from an MOE that high.

275 Likely Republican caucus-goers (definitely/probably will attend) (plus or minus 7)

The Strategic Vision poll from just days before sampled 600 likely Republican caucus goers and had Huckabee at 27 and Mitt at 24, within the poll's MOE. I'd want to see some additional polling particularly after the speech before I jumped to conclusions on this one. And because it is a caucus, Huckabee still needs the ground game to make sure people turn out. As Marc Ambinder points out, this could end up being a gift for Mitt, more because of what it does for expectations, than anything else.

Comments

[LOL]

"you're looking at an MOE plus/,minus 7"

what do you base this statement on?

[LOL]

because a sample of 1408 people from the entire population of Iowa (i.e ~2.9 million people) is 2.61%. Obviously the Republican caucus will be much smaller than that, and so will have a smaller MOE.

I mean, do you actually know anything about anything?

[Sara]

Are there a set number of caucus goers, or is it open to all who show up? If it is open, don't count Mitt out, he has a pretty good organization going in Iowa from what I've heard.

[Dan Riehl]

"what do you base this statement on?"

Thanks for proving you're an idiot. I'm basing it on the number they provided and which I quoted. Learn to read.

275 Likely Republican caucus-goers (definitely/probably will attend) (plus or minus 7)

[BobInStamford]

This Huckleberry fellow sure is progressive.

http://news.yahoo.com/s/ap/20071208/ap_on_el_pr/huckabee_aids

We are lucky that the LORD put such a patriot in our midst.

Amen!

[Dave]

The poll has to be an outlier and the small sample size provides cover. It has had a dramatic and immediate impact among bloggers, as any cruiser of websites can attest. A number of people have posted to the effect that they hope Mitt can stop Huck in Iowa, and that it might be all over if he can't....and I'm talking about people who were strongly for other candidates. Club for Growth is going to start airing a devastating ad of Huckabee championing a wide variety of tax increases while in office. Fred's campaign is sending literature to Iowans attacking Huck's conservative claims vs. his liberal record on things like spending and taxes. Mitt is also sending mailouts explicating Huckabee's record. The news media seems focused on the DuMond case and his statements about quarantining aids patients. I'm confident that Mitt will win about 3 1/2 weeks from now.

[Jim Addison]

ANY survey with a sample so small is suspect. When it is so far out of line with other recent polls with larger samples, raise the bright red flag on it.

Ironically, if the results can't be immediately duplicated by more substantial samples, it only serves to hurt Huckabee. He would be best served remaining a dark horse second in the polls and pulling off an upset win. The media buzz would be deafening. Now, his expectations are raised.

Romney might not have been able to survive losing Iowa, always a keystone of his strategy, had he been leading in all polls going in. This gives him some cover, so even a close second could justify continuing (not to mention his bankroll). Now a second place for Huckabee isn't a big and positive story; it's a disappointment.