Different animals. Instatrends polled 386 registered voters. Rasmussen polled 500 likely voters.

Actually, if you look at the actual poll on the InstaTrends website, it states that the people polled were highly likely voters (those who voted in the last 5 elections)

It said:

"A total of 386 registered Missouri voters were polled in a random sample telephone survey from September 17-19, 2005"

It also said it was performed using computer recordings and voice recognition software.

Upon further review:

Both were non-human. InstaTrends describes their sample better. It is not clear what Rasmussen describes as "likely voters." Neither poll explains how it handles call-backs.

On a statewide race including Missouri, the pollster should have 600 people that it talks to make the poll fairly accurate.

Both polls are consistent on one item, the incumbent(Talent) is under 50% and that means good news for Claire. Usually the vast majority(80%) of undecideds move toward the challenger.

Talent can not win, a two way race, without 50%. Yes, Missouri we have a U. S. Senate race.

In the interest of fairness, kcindsider is right about the InstaTrends survey methodolody. This is from the methodology link listed on thier website. Though I agree the sample size is lacking, the quality of the sample is excellent.

From: http://www.instatrends.com/press/2006MissouriSenate2005-09-19.pdf

· Survey Conducted By: InstaTrends, L.L.C.
· Target population: Missouri highly-likely voters.
· Sample Population: Registered Missouri voters who voted in all five of the
following elections: 1998 General, 2000 Primary, 2000 General and 2000
Presidential.

Of course, that kind of screen does exclude every registered Missouri voter who moved here after Nov 1998 or who wasn't old enough to vote in 1998.

(Also, if JW copied that sample description properly, somebody can't count to five.)

The sample may be good, but the voice recognition software and recorded messages are not nearly as accurate as live phone callers. We should wait until there is a live telephone poll. I also think that in the type of poll that this is, there should be some weighting to take into account some error rates in the recognition of voices, etc.

I checked the numbers, and the sample size is correct, and I had no real problem with the questions, however, I wouldn't rule out Rasmussen quite yet, He has been closer than any other pollster in the country on the last few Presidential elections. He is quite talented and is one of the most highly respected Pollsters in the country, he is also independent.

I have not heard of instatrends, I do not know where or how they get their information. I know that they state five elections but only list four in their methodology, I would assume 2002 General but I don't know. Their methodology on the website leaves me with questions:

How good are their lists, (I hope better than their editing skills)?

Did they call purely randomly or use systematic random sampling?

Too many question marks, robo poll, who paid for it, and who are they? If I am wrong about Instatrends skills as pollsters please let me know.

It's pretty much impossible to determine their methodology without knowing their screen--as Public Eye already pointed out, typical screens for Highly Likely Voters would exclude a good many people. And more than Highly Likely Voters are likely to show up. Given Democratic base voters are often not highly likely, it doesn't give you much to go on.

http://ambienrx.zikforum.com/ http://ambienrx.ephpbb.com/ http://ambienrx.forumup.org/?mforum=ambienrxAmbien rx Ambien rx Ambien rx Buy ambien Buy ambien Buy ambien Ambien online Ambien online Ambien online Ambien cr Ambien cr Ambien cr Ambien side effects Ambien side effects Ambien side effects