Quote:
Originally Posted by sycamore
Depends on the source...a lot of the bigger polls seem to be pretty solid from a scientific standard.
That one that UT posted looks so wrong though. But I don't think it's an issue with the mechanics of the poll...just the people polled.
|
It comes down to sample size and the Z. For example, a telephone poll of 400 people on the streets of San Francisco, Calif hardly can be extrapolated to "the people of the US would vote for..." The margin of error is the other thing that is important to note.
Telephone polls are some of the worst. All you did was sample people who have telephones.
The most amazing thing is often how small the sample size is. You hear it all the time on TV. "The number of Americans that would vote for X,Y, or Z is 45a%". Looking further you see the sample size was 846 people. Ok, please tell me how you extrapolate opinions of 846 people to 32 million people. It can't be done. The statistical validity hovers near zero.
Statistics, lies, and more statistics.
Polls are easily constructed through the pointed questions they ask to extract the information that the pollsters is after. Political polls and polls by special interest groups with very bland sounding names are some of the best at doing this. Did you ever get that telephone calll from some tighty-righty or lefty-loosey political organization? Listen carefully to the questions being asked. Often only yes or no answers with no clarification or middleground choice.
Polls.... spitoooie....
