I was reading on another blog where the writer was overly dependent on poll results as to how Americans perceive George W. Bush's policies. In light of the recent "approval rating" polls, I had to comment.

I have never been a part of these polls. Nobody I know has ever been a part of these polls. Given the number of people I have known over the years, it is statistically unlikely that a random sampling of Americans would consistently avoid every individual with which I have discussed the subject.

Having taken many online tests/polls, I have consistently been frustrated by the selection of answers. Put simply, the answer I want isn't there. This is where it gets interesting.

In a recent episode of the TV show "Numbers" (one of my newest geeky fixations), the math professor points out the flaws of a police lineup. Individuals brought in for identification subconsciously believe they HAVE to pick one of the six individuals set before them. And so they select the individual who MOST resembles the perpetrator, which gives a reasonable margin of error in their selection. With polls it is no different. When given a choice of "a, B, or C", most individuals will take one of the three rather than state no opinion or a different opinion (which, if stated, is likely to be included in the poll data under the ambiguous category of "other"). Third party presidential candidates often get caught in this trap, as they are required to poll at a certain percentage to get invited to the debates, and yet most polls deliberately exclude these candidates and/or shift them to the "other" category.

Add to that the fact that many individuals work shift work or have no landline phones, and the poll data becomes increasingly skewed. There is also the fact that poll responses aren't legally binding, so people don't necessarily take them as seriously as they would a vote.

There are probably a whole slew of variables I have overlooked here. But the fact remains that polls, while a useful litmus test for gauging SOME public interest, should not have the emphasis that they do in our American culture. Pollsters have no legal responsibility for flawed data, and thus, have no incentive to fix this system.