[Strider]

Now when I took statistics, a “random” sample was just that “random”. When a poll is taken where “supplemental voters” and added to provide additional voters in one category that is a problem, the conclusions reached cannot be trusted.

It all depends on how it's done.

For instance, if I want to survey a population and compare let's say three age groups. If one of the age groups is underrepresented in the population, and I draw a random sample, I may not get enough of the oldest age group.

There just aren't that many of them and a random sampling will give me about 30 old respondents. That's not a big enough sample to compare them to the other groups.

So, I purposely interview 100 old respondents even though a random sample would not give me this result.

Now all I have to do is weight the old respondents down from 100 to 30, when extrapolating to the entire population.

That is only one of the many ways populations are over and under sampled. As long as you weight them back to their proper proportions, before extrapolating to the whole, your fine.

[DiamondDon1]

Point well taken.

But exactly how did the Times find these Latino voters? One would think they would call in Latino areas which are most likely Democrat in nature. It would be very difficult to right this ship, the water inside is above the waterline outside...

DD