M Moore's Likely Voters comment about the polls

[liveoaktx]

He says that the company actually skews left and the polls are based on voters who already voted in a previous primary or election as opposed to the 50 percent of people in general who are NOT likely voters who ARE NOT POLLED.

So now I'm wondering. How many polls that I've read over the last 6 months have been based on *likely voters*?

[stewert]

1. That's Why.......

The political polls should only be taken by people who say they plan to vote in November. Likely voter polls are garbage, even polls of registered voters are not totally accurate.

They should only poll registered voters who say they plan to vote in November. That is a poll I would like to see, and zogby should do it because his methods are the most accurate.

[HEFFA]

2. There are so many unanswered questions where polling is concerned

Unfortunately, so many people (especially politicians) give them far too much weight.

This phenomenon that Moore notes is accurate and problematic to say the least. Furthermore, the single most critical question that must be asked before considering the merit of any poll's results is: "What's the sample size?" Have you ever heard mention of this question or been openly afforded an answer to it? It's rare indeed.

Second only in number is the important question of the method used to obtain the sample. There are countless ways to gather a sample, but only one method makes it possible to generalize from the results to the larger population. That method is random sampling. It is rarely used in public opinion polling. Thus, virtually every poll that you see has biases built into it.

I've studied the "science" of polling at length, and I'm skeptical of poll results and the countless ways that these results can and are misused. I wouldn't suggest that they're all useless, but they certainly don't tell nearly as extensive or dependable a story as the press or most of the public would believe.