Friday, September 10, 2004

Polls are deceiving - Stop the polling!

As any statistician will tell you, statistics can be twisted to say what you want them to say. The same is true with polling, especially when "who" was polled, "what" exactly was asked, "how many" were actually polled, and what the personal views of those polled were before being polled, are all pieces of information that are NOT INCLUDED in the publishing of poll results.

Here is an example:

A poll after the Republican convention was highly publicized as showing an 11 point lead by Bush. What was less reported was that this was a poll of only 1000 people. What was never made clear was who these 1000 people were. Were they all Republican Delegates? Were they all in a mostly Republican section of the country? Were they folks who were leaning toward Bush before the convention but had not previously been polled? We don't know any of these things, and so, polling reports should be totally ignored and should not be reported in the first place, because they are, at best, misleading.