anyone see obama's new attack ad

[bu villain]

9 of 17 is a majority.

Sure, a 9 of 17 is a majority. Can't argue with that math. What I am arguing about is that it is a meaningless sampling and will probably mislead people into thinking that a majority OF ALL ECONOMISTS feel that way. Hey I just asked three people on the street who they were voting for and they said Obama. Would it be responsible for me to write an article and title it "100% of Georgians plan to vote for Obama"?

[David88vert]

Sure, a 9 of 17 is a majority. Can't argue with that math. What I am arguing about is that it is a meaningless sampling and will probably mislead people into thinking that a majority OF ALL ECONOMISTS feel that way. Hey I just asked three people on the street who they were voting for and they said Obama. Would it be responsible for me to write an article and title it "100% of Georgians plan to vote for Obama"?

It's as meaningful as the way the job numbers are polled.
If you looked at the percentage that those 17 make up of all professional US economists, and looked at the percentaged of those polled on the job rate against all individuals over 18 in the US, I suspect that the percentage would be greater on the polled economists. There aren't that many well-known and well-respected economists, and there are a lot of people eligilbe for work in the US. Yet, some people have no problem with the small sampling for the job numbers - when they support their political position.

[bu villain]

It's as meaningful as the way the job numbers are polled.

So then you agree, it's meaningless.

If you looked at the percentage that those 17 make up of all professional US economists, and looked at the percentaged of those polled on the job rate against all individuals over 18 in the US, I suspect that the percentage would be greater on the polled economists. There aren't that many well-known and well-respected economists, and there are a lot of people eligilbe for work in the US. Yet, some people have no problem with the small sampling for the job numbers - when they support their political position.

Actually you are mistaken on the math behind statistical significance. It is not as simple as what percentage of the total.

To determine unemployment for the US Population of 300 million, you only need to ask 1067 people to get a result that has a margin of error of 3% with a 95% confidence level.

To determine the opinion of the top 100 economists with the same 3% margin of error and 95% confidence level, you need to ask 92 of them.

Now the article only asked 17 (out of the top 100 economists lets say). Their result was 9 of 17 or 53%. With a 95% confidence level, guess what the margin of error is. ..... almost 22%. So the conclusion of the article should be "Somewhere between 31% and 74% of top economists think Romney is better for the economy."

[David88vert]

So then you agree, it's meaningless.



Actually you are mistaken on the math behind statistical significance. It is not as simple as what percentage of the total.

To determine unemployment for the US Population of 300 million, you only need to ask 1067 people to get a result that has a margin of error of 3% with a 95% confidence level.

To determine the opinion of the top 100 economists with the same 3% margin of error and 95% confidence level, you need to ask 92 of them.

Now the article only asked 17 (out of the top 100 economists lets say). Their result was 9 of 17 or 53%. With a 95% confidence level, guess what the margin of error is. ..... almost 22%. So the conclusion of the article should be "Somewhere between 31% and 74% of top economists think Romney is better for the economy."

All the polls are meaningless. Only the electoral collee poll matters in the long run. The media just loves to stir up controversy to generate more profits.

Asking 1067 does not get you a cross-section of 300 million, in and of itself. You have to look at different region, economic levels, industrues, age, races, genders, etc. 1067 is not enough to achieve this, period. Do you think that you are going to get the same answer from a well-to-do 70 year old white pig farmer in OK, as from a single mother-of-3 black woman in NYC, who just got laid off from her job? It's not possible to get enough sampling from 1067 for conclusions or relible calculations on ANY subject. f it could, we wouldn't need to have an election at all. We could just make a few phone calls (all to Chicago, of course).

[bu villain]

All the polls are meaningless. Only the electoral collee poll matters in the long run. The media just loves to stir up controversy to generate more profits.

Asking 1067 does not get you a cross-section of 300 million, in and of itself. You have to look at different region, economic levels, industrues, age, races, genders, etc. 1067 is not enough to achieve this, period. Do you think that you are going to get the same answer from a well-to-do 70 year old white pig farmer in OK, as from a single mother-of-3 black woman in NYC, who just got laid off from her job? It's not possible to get enough sampling from 1067 for conclusions or relible calculations on ANY subject. f it could, we wouldn't need to have an election at all. We could just make a few phone calls (all to Chicago, of course).

My point was and still is that the 17 economist poll is useless and will only mislead people. That's it. I'm trying not to get pulled off on a tangent to analyze any other kind of polling. Funny side note though, there are a significant number of economists and mathemeticians who think should decide elections via relatively small samples of polling.