Rand Paul: 51%; Jack Conway: 45%

[qh4dotcom]

Rasmussen:

Opposition to the national health care law is higher in Kentucky than it is nationally.In Kentucky, 65% favor repeal of that law and just 29% are opposed.

Hopefully they won't forget that on Nov. 2nd

 

[BamaFanNKy]

I still haven't seen much of a Conway movement. It's basically a Vote for Rand or a Vote against.

 

[specsaregood]

How to Read Political Polls Like a Pro

I'm not buying it as it makes no statistical sense. It is a single question, not two questions. So their is a single standard deviation to it. I know its been a long time since my statistics classes but I'm pretty sure we never doubled it like you propose.

 

[low preference guy]

I'm not buying it as it makes no statistical sense. It is a single question, not two questions. So their is a single standard deviation to it. I know its been a long time since my statistics classes but I'm pretty sure we never doubled it like you propose.

I'm not proposing it, I'm just explaining how people think. If you want to go deeper, the "margin of error" is just a rule of thumb, not anything meaningful (as far as predictive power). The result of an election could in fact vary with more than twice the margin of error.

 

[sailingaway]

I'm not proposing it, I'm just explaining how people think. If you want to go deeper, the "margin of error" it's just a rule of thumb, not anything meaningful (as far as predicting power). The result of an election could in fact vary with more than twice the margin of error.

With all respect, I think you are thinking too hard. I think they just wrote the article while still tabulating the results and THOUGHT the lead would be within the margin of error like the Daily Kos's was....HOPED it would be, I should say. That is never how margin of error has been used in any other poll I have ever seen.

 

[low preference guy]

With all respect, I think you are thinking too hard. I think they just wrote the article while still tabulating the results and THOUGHT the lead would be within the margin of error like the Daily Kos's was....HOPED it would be, I should say. That is never how margin of error has been used in any other poll I have ever seen.

The article I just posted disagrees with you. But it was written by John McManus. Maybe he is crazy.

 

[specsaregood]

If you want to go deeper, the "margin of error" it's just a rule of thumb, not anything meaningful. The result of an election could in fact vary with more than twice the margin of error.

Uhm yes it is absolutely meaningful. Margin of error is a critical mathematical component. It is a measure of the confidence interval. It isn't just a rule of thumb. The larger the poll, the higher the confidence level, the smaller the MOE.

Edit: here you go from the wiki

 

[low preference guy]

Uhm yes it is absolutely meaningful. Margin of error is a critical mathematical component. It is a measure of the confidence interval. It isn't just a rule of thumb. The larger the poll, the higher the confidence level, the smaller the MOE.

I know it has a statistical meaning, I'm saying it's not meaningful as far as predictive power of the actual result. The actual result of an election could vary by more than three times the margin of error of a poll.

 

[sailingaway]

Uhm yes it is absolutely meaningful. Margin of error is a critical mathematical component. It is a measure of the confidence interval. It isn't just a rule of thumb. The larger the poll, the higher the confidence level, the smaller the MOE.

Edit: here you go from the wiki

I know the margin of error is meaningful, I just think that a margin of error of 4.2 is not a margin of error of 8.4. I'll read the article, though.

--
ok, looking at your wiki I stand corrected. I have never seen it used that way before, though, but maybe others who discussed it just got it wrong. Well, that means Rand could be 14% ahead, as well.

 

[Epic]

Rand could be 4.2 lower than he is in the poll, and Conway 4.2 higher than he is in the poll, so the real preference could be 8.4 points off.

Still, this was a push-poll and Rand has at least 47%. It's a really good result for Rand.

No, this isn't how it works.

 

[specsaregood]

I know it has a statistical meaning, I'm saying it's not meaningful as far as predictive power of the actual result. The actual result of an election could vary by more than three times the margin of error of a poll.

Ok, that wasn't the impression I got.
It does appear you are correct though. I was operating under the impression that the MOE was the entire confidence interval. According to wiki the MOE is usually reported as half the confidence interval which means the MOE is effectively double the reported percentage. Cheers!

No, this isn't how it works.

Evidently it does. The good news is this means Rand could be leading by as much as 56 - 41. 15%

 

[low preference guy]

Ok, that wasn't the impression I got.
It does appear you are correct though. I was operating under the impression that the MOE was the entire confidence interval. According to wiki the MOE is usually reported as half the confidence interval which means the MOE is effectively double the reported percentage. Cheers!

Yeah, and don't discount the questions they asked. They asked about private business discrimination and BP, so I categorize this as a push-poll.

 

[Stop Making Cents]

Fantastic news!

If Rand was only up 6 points it wouldnt' be a big deal, but the fact that he is over 50% is a HUGE DEAL!!!

All he has to do is hold steady!

 

[specsaregood]

Yeah, and don't discount the questions they asked. They asked about private business discrimination and BP, so I categorize this as a push-poll.

Yeah, I think this poll is positive for Rand. With that said, any poll that doesn't release the crosstabs isn't worth reporting on. Usually it is push polls that don't release that stuff so......that makes sense.

 

[Matt Collins]

How could a major newspaper say 6 < 4.2 on the front page?

Because this is the same company (Gannett) who lied about Ron Paul even being in the race back in 2008. And this company is very close to John Seigenthaler, Al Gore, and the Kennedy family.