Is Obama trying to destroy the Democrat party?

[Redfish]

give the boy a prize. he can do some basic math. Yes, 17 million is what would be required for a statistically accurate representative sample, and yes, a poll of 17 million would be virtually impossible to do.

So the pollsters play little games and claim that they structure their 1000 person polls to include every possible demographic within the same ratios as they appear in the total population. Its bullshit, but if you choose to buy into the bullshit, go right ahead.

I love ya man, but you are wrong on this

no I'm not. research it. what I said is exactly mathematically correct. what I said in red above is exactly what the pollsters do.

If 1000 person, 2000 person polls in a population of millions were meaningless, which is what you claimed,

then final election polls would be no closer to the actual results than a random selection of numbers.

We KNOW that is false.

I said it was mathematicallly meaningless, and it is.

It cannot possibly be 'mathematically meaningless' if it is FAR more accurate than a random selection.

You're effectively claiming that a 1000 person presidential election poll is no more accurate than if we put every percentage point from 0 to 100 in a hat,

and drew out one for Obama and one for Romney,

and posted that.

LOL----------carefully selected random sample-------------LOL. You don't even understand the words you are using.

 

[kaz]

the fact is that political polling in the USA is designed to influence public opinion, not to report on it. We can give credence to polls if we choose, but we are naive if we think they are statistically pure.

On Sampling:

Actually, the sample size is not really dependent on the size of the population being sampled. You are clearly out of your element in statistics. You and I generally agree on things, what you are saying statistically wrong.

You realize statistics is built on probability, right? That's why I gave you the example of coin flips. Does it really matter if you skip a coin 10,000 times ore 10,000,000 times? Sample size is really driven by probability.

Harvard MBA is impressive, but I don't know what it has to do with statistical modeling.

If you know what a six sigma black belt is, that should sway you a lot more than either of our math majors. I stopped in math with an undergrad degree. My graduate degrees are in computer science and business. So I have no problem believing you took more than I did.

On bias:


On bias though I agree you have a point. Hence the Kerry fiasco where the pollers didn’t want a fair evalution of exit polling. They were hoping to give an impression Kerry was doing better than he was hoping to gain momentum for him. Lack of bias is critical to accurate polling.

Leftist media polls similarly are constantly skewed because of how they phrase questions. For example, if you ask the population the following questions:

1) Do you think abortion should be banned in all cases, you will get a no

2) Do you think abortion should be available in all cases, you will get a no

The majority think abortion should be banned other than in certain cases, like rape, incest and for medical reasons. So depending how you phrase the question, any side can get the answer they want. The media is well aware of that, and it’s hard to believe with the constant results they don’t use that freely to project poll results

 

[kaz]

I love ya man, but you are wrong on this

no I'm not. research it. what I said is exactly mathematically correct. what I said in red above is exactly what the pollsters do.

If 1000 person, 2000 person polls in a population of millions were meaningless, which is what you claimed,

then final election polls would be no closer to the actual results than a random selection of numbers.

We KNOW that is false.

I said it was mathematicallly meaningless, and it is.

It cannot possibly be 'mathematically meaningless' if it is FAR more accurate than a random selection.

You're effectively claiming that a 1000 person presidential election poll is no more accurate than if we put every percentage point from 0 to 100 in a hat,

and drew out one for Obama and one for Romney,

and posted that.

LOL----------carefully selected random sample-------------LOL. You don't even understand the words you are using.

I did a double take on hat one too. How can a sample be more accurate than a random sample? Random sample is the whole point of how you pick the sample. Again, the foundation of statistics is probability