Rand in 5th in Iowa at 5% leading Bush

[WTLaw]

Just saw a poll where the most recent results had Jim Gilmore beating Rand. What can you say? There is something odd going on here. Maybe Gilmore really is a dark horse...or maybe these polls really are that screwed up. That is pretty screwed up, I must say. I'm not sure I know another human who knows who Gilmore is. Maybe a couple activist friends, but just barely.

[LatinsforPaul]

Just saw a poll where the most recent results had Jim Gilmore beating Rand. What can you say? There is something odd going on here. Maybe Gilmore really is a dark horse...or maybe these polls really are that screwed up. That is pretty screwed up, I must say. I'm not sure I know another human who knows who Gilmore is. Maybe a couple activist friends, but just barely.

Link please.

[WTLaw]

http://polling.reuters.com/#poll/TR1...51124/type/day

Reuters 5 day, Gilmore's past two days have been infuego.

[LatinsforPaul]

http://polling.reuters.com/#poll/TR1...51124/type/day

Reuters 5 day, Gilmore's past two days have been infuego.

The Reuter's polls do not count for anything except for Trump's team to tweet out on a weekly basis. Though to be fair, I think Rand's team once did tweet out one of their 5 day polls.

[WTLaw]

Sure, my comment is aimed less at the debate situation and more at the frustration we are all feeling looking at most/all of the polls i general. There are frequently concerns about methods and who they are polling. How can Reuters even find a Jim Gilmore supporter?

[hells_unicorn]

Just saw a poll where the most recent results had Jim Gilmore beating Rand. What can you say? There is something odd going on here. Maybe Gilmore really is a dark horse...or maybe these polls really are that screwed up. That is pretty screwed up, I must say. I'm not sure I know another human who knows who Gilmore is. Maybe a couple activist friends, but just barely.

The poll has extremely skewed results and is based on an extremely small sample, I wouldn't read anything into this.

[groverblue]

The poll has extremely skewed results and is based on an extremely small sample, I wouldn't read anything into this.

Sample size determination

n = 4/W^2 = 1/B^2 where B is the error bound on the estimate, i.e., the estimate is usually given as within ± B. So, for B = 10% one requires n = 100, for B = 5% one needs n = 400, for B = 3% the requirement approximates to n = 1000, while for B = 1% a sample size of n = 10000 is required. These numbers are quoted often in news reports of opinion polls and other sample surveys.

So, 3,366 Respondents is a pretty good sample size, statistically. Just keep in mind that's it's accurate for the sample, which would be:

Age 18-29: 18% (629 Respondents)
Age 18-34: 18% (629 Respondents)
Age 30-39: 13% (447 Respondents)
Age 35-49: 20% (697 Respondents)
Age 40-49: 14% (474 Respondents)
Age 50-59: 14% (474 Respondents)
Age 60+ : 39% (1,336 Respondents)
Age 65+ : 28% (958 Respondents)


So the sample for 40-60+ is 74% (2,506 Respondents). That's the skewed part. It's heavy with the 40+ age range. It should be ~50%.

[CPUd]

Sample size determination

n = 4/W^2 = 1/B^2 where B is the error bound on the estimate, i.e., the estimate is usually given as within ± B. So, for B = 10% one requires n = 100, for B = 5% one needs n = 400, for B = 3% the requirement approximates to n = 1000, while for B = 1% a sample size of n = 10000 is required. These numbers are quoted often in news reports of opinion polls and other sample surveys.

So, 3,366 Respondents is a pretty good sample size, statistically. Just keep in mind that's it's accurate for the sample, which would be:

Age 18-29: 18% (629 Respondents)
Age 18-34: 18% (629 Respondents)
Age 30-39: 13% (447 Respondents)
Age 35-49: 20% (697 Respondents)
Age 40-49: 14% (474 Respondents)
Age 50-59: 14% (474 Respondents)
Age 60+ : 39% (1,336 Respondents)
Age 65+ : 28% (958 Respondents)


So the sample for 40-60+ is 74% (2,506 Respondents). That's the skewed part. It's heavy with the 40+ age range. It should be ~50%.

I admit I haven't looked into it much, but shouldn't ages in a random sampling follow a normal distribution?

[limequat]

I admit I haven't looked into it much, but shouldn't ages in a random sampling follow a normal distribution?

Only if the population is normal. It is not. Statistically or otherwise.

[Crashland]

Sample size determination

n = 4/W^2 = 1/B^2 where B is the error bound on the estimate, i.e., the estimate is usually given as within ± B. So, for B = 10% one requires n = 100, for B = 5% one needs n = 400, for B = 3% the requirement approximates to n = 1000, while for B = 1% a sample size of n = 10000 is required. These numbers are quoted often in news reports of opinion polls and other sample surveys.

So, 3,366 Respondents is a pretty good sample size, statistically. Just keep in mind that's it's accurate for the sample, which would be:

Age 18-29: 18% (629 Respondents)
Age 18-34: 18% (629 Respondents)
Age 30-39: 13% (447 Respondents)
Age 35-49: 20% (697 Respondents)
Age 40-49: 14% (474 Respondents)
Age 50-59: 14% (474 Respondents)
Age 60+ : 39% (1,336 Respondents)
Age 65+ : 28% (958 Respondents)


So the sample for 40-60+ is 74% (2,506 Respondents). That's the skewed part. It's heavy with the 40+ age range. It should be ~50%.

It should follow normal distribution of likely voters. In the 2012 Iowa caucuses, 74% of the actual voters were age 40+. There's no reason to think it will dramatically shift down to 50% this year.