Previous 11 - 20 Next
Darn no good Polacks.
It's certainly not basic math. That you would say so indicates that your knowledge of statistical analysis is minimal. If a poll could be conducted in which the sample was everyone who will vote in the election, then yes, it would be basic math, however this would be far too expensive. Pollster instead try to get a random sample of the voters. A truly random sample will give a very accurate approximation of the opinions of the entire pool. The accuracy is determined largely by the sample. Obviously, if the sample is of 90% of the pool it will be more accurate than if it were of 10% of the pool. The smaller the sample the cheaper the poll is to do. So, various statistical methods are employed to try and increase the accuracy.
I have a hard time seeing a Paul supporter ever voting for Obama. I see them holding their nose and voting for Romney or staying home. So, if it is true that the reactionary youth who voted in droves for Mr. Obama are now Paul fans, then yes, Obama is in trouble.
But, the pollster make their methodology available. We know how they poll. One major poll calls people on land line telephones (and probably over the telegraph system as well) between the hours of 4 and 6. This obviously will introduce a bias. This poll is Rasmussen.
Eddie, Go ahead throw your vote away! Don't blame me, I voted for Kodos.
I think that's the point of the article. If the initial demographics have changed significantly, the polls weight their results inaccurately. If there is reason to believe that this is the case (I think there is), then perhaps there is reason for optimism.
Legitimate polls are a science. Exactly. However, I don't buy that the polls are deliberately misleading. Pollsters want to be accurate. My guess is that the weights are introduced to compensate for a previously determined bias in the sample. So, if I previously determined a pool of voters to be split 34% to 21% and poll them again less accurately (such as in a small sample daily tracking poll) and find that my current sample is 34% -34%, a pollster can make the reasonable assertion that SSS has skewed his results and that he should weight it to the previously determined split. Of course, this seems to fail to account for changes to the voting demographic between the polls. Most pollsters truly want to be accurate.
Nate Silver gives this about a 4% chance of happening based nearly 40 years of data. In other words, in the past, in races looking like this in an environment such as this, the chance of the true result being a landslide for Romney is about 4%.
I don't understand why a random sample would need to be weighted. Weighting would seem to introduce a lot of error and seems to be very subjective. Which polls weight? Why do they weight the way they do? No offense to the belief in your gut, Mr. Towery, but could you please give a reasonable explanation for why pollster (even generally conservative ones like Rasmussen) find a need to introduce the bias and why are they wrong to do so? Why is a random sample not the best way to eliminate bias?
In response to:

Should We Obey All Laws?

bglick4 Wrote: Jul 13, 2012 10:22 AM
I think that is pretty much impossible to not violate the law in some way. Many traffic laws, if followed, create unsafe conditions. In PA, the speed limit on entrance ramps is 35 mph. It is extremely unsafe to attempt to merge at 35 mph, so, of course, no one does this. However, if a cop sees one going 65 on an entrance ramp prior to merging, he can pull this individual over and levy a massive fine. Does one obey the law or drive safely? There are three things every man needs in this country; a good wife, a good accountant, and a good lawyer. Not necessarily in that order.
In response to:

Should We Obey All Laws?

bglick4 Wrote: Jul 13, 2012 10:05 AM
I think the 1994 election was, in some ways, a response to Waco. I guess it also inspired the lunatic response in OKC.
Previous 11 - 20 Next