Posted by cccp1014 on 3/15/2018 10:32:00 AM (view original):

Posted by bad_luck on 3/15/2018 10:06:00 AM (view original):

Posted by cccp1014 on 3/15/2018 9:56:00 AM (view original):
Yes. It could be misleading. Polls can be misleading. We agree.

I still say my poll of ~100 persons was more statistically significant than his poll of 15 people.

Do we agree that a sample of 32 is not good enough for large populations?

Define large

But yes it can be misleading when you have over 20,000 persons. "Not Good Enough" is not a term I would use. I would say "could be misleading".

For a 32 voter sample? Any population over 50.

I would definitely use "not good enough." The entire point of a poll is to give you an idea of what will happen. If your margin of error produces a range of 30 percentage points, it's useless.

What? In a population of 51 its not useful?

Posted by cccp1014 on 3/15/2018 6:13:00 PM (view original):
What? In a population of 51 its not useful?

The MOE on a 32 voter sample with a 51 voter population is 11. That gives you a 22 point range. Pretty high. If the poll comes back John 46 - Steve 54, you’ll have no idea who’s actually going to win.

I think you have a margin for error of course but I think you would more often than not get a good idea of the outcome. But not even the outcome if you asked for their top three issues I think that part would definitely help if you asked ~63% of the population.

Posted by cccp1014 on 3/16/2018 11:20:00 AM (view original):
I think you have a margin for error of course but I think you would more often than not get a good idea of the outcome. But not even the outcome if you asked for their top three issues I think that part would definitely help if you asked ~63% of the population.

A) you said 32. All you need is 32 people regardless of population to get an answer that is, in your words, "statistically significant." That's incorrect. A sample of 32 from a large population is not going to give you an answer that you will be able to rely on as not a result of random variance or error.

(side note--"statistically significant" is not really the wording I would recommend for this...I'd use "useful" or "predictive")

B) In a 32 voter sample out of a population of 51, you only get a "good idea of the outcome" if the result of the poll is extreme. 90% to 10%, etc. If the result is relatively close...even as close as 60/40, you really don't know because just a handful of votes swing the election.

C) This is dumb, because no one is running a scientific poll with a population of 51. But even if you are polling a small town with 800 or 900 people and you want a reasonably predictive poll, you're going to need a couple hundred people. Not 32.

I concede.

well done boys........... almost Jefferson vs. Hamiltonian............... well, not quite. hahaha

more like sponge bob vs patrickonian!

Doug... Still waiting on your report about 538. I can't wait to hear all about why they are a hate site!!

tangplay is a hypocrite and a liar. Don't waste your time confabulating with him.

I'm still waiting for an explanation of his meltdown on the Dino site the other day. He seems a bit unhinged at times.

Posted by cccp1014 on 3/18/2018 2:47:00 PM (view original):
tangplay is a hypocrite and a liar. Don't waste your time confabulating with him.

I am not a hypocrite and not a liar. Please stop trying to undermine me.

Posted by tangplay on 3/18/2018 2:34:00 PM (view original):
Doug... Still waiting on your report about 538. I can't wait to hear all about why they are a hate site!!

What page was that on?