Redfish
Diamond Member
Harvard MBA?I understand statistics quite well. I also understand that the current polling is designed to influence public opinion, not to report on it.
When they use a sample of 1000 out of 330,000,000, it is not a statistically valid representative sample. I know that the pollsters claim to proportionally represent every demographic in the country in their tiny sample. But it is simply not possible. True, some of them make pretty good guesses and some of them are flat out rigged.
you are free to believe them if you like.
LOL.....Harvard MBA doesn't understand statistical sampling
'
clearly much better than you do. go to your local library and check out a stat 101 text book, then read about representative sampling, you might learn something.
I fully understand what the pollsters CLAIM. But it does not pass a math test.
Even Harvard MBAs understand that each sample has a margin of error of plus or minus a few percent. To think that a Harvard MBA does not understand that a sample of 1000 can be relevant is embarassing.
Individual samples of 1000 can each be challenged on their standard deviations. However, if repeated individual samples of 1000 are yielding the same results, those samples collectively are more valid
Even a French model like myself understands it
Amazingly ignorant. a sample of 1000 out of 330,000,000 cannot ever be valid, the margin of error is too great. The pollsters try to overcome that by carefully selecting the sample to (theoretically) proportionally cover all the demographics within the 330,000,000 population. The problem is that there are more than 1000 different demographics in the USA so it is impossible to proportionally cover all of them in such a small sample.
Now if you make the sample 2,000,000 or more the margin of error might be acceptable. But they never use that large of a sample because it would cost too much and take too long to gather.
I am right about this, you only make yourself look foolish by continuing your rants.
No, the margin is not too great
If you random sample 1000 people and every one says they do not support Redfish for President do you think you need to sample 100 million before you can be confident that Redfish has no chance?
Sampling error will provide your statistical confidence band.
One sample may be an outlier and need to be thrown out. If random samples consistenly yield the same result, they provide valitity
^^^^ignorance on display.
If I go to the lower ninth ward in New Orleans and ask 1000 people if they think Obama is a good president, probably 900 will say yes. Should I then conclude that 330,000,000 americans think Obama is a good president?
The pollsters do not do "random" sampling. Do you even know what random sampling is?