I think online polls are all frauds. I think they're a sleazy trick to get people to come to websites. What's more, I think everyone knows they're phony; that the sampling that votes is utterly unscientific and that many groups stuff the online ballot box, often because they figure out how to vote repeatedly. This is not usually difficult to do. Amazingly, someone has now figured out how to make an unreliable poll even less reliable than usual. This is done by changing the question that people are voting on.
Yesterday, a poll appeared on the website of Senator Majority Leader Bill Frist. Here's a screenshot of how the results looked after the thing had been up for a few hours. Note not only the totals but the way the question is worded…
Ordinarily, a poll on the website of a politician should wind up overwhemlingly reflecting that politician's position, since folks who agree with him are far more likely to visit that site. In this case, a couple of Democratic websites urged their readers to go over and push up the "no" vote on Senator Frist's poll, and droves apparently did. At the time the above screenshot was taken, it looked like Frist's view might lose on his own home page. Fat chance of that happening. Even as Conservative sites were urging visitors to go drive up the "yes" vote, the operators of the Frist site were changing the wording of the question to slant it more their way. Here's a screenshot of the same poll from later in the day…
I'm not sure if they reset the total to zero and started the count over or if they just changed the question. But either way, the first question disappeared. The new wording plus the Conservative vote-stuffing apparently did the trick because soon, Frist's position had pulled into the lead. That apparently wasn't enough for the operators of the site because later, they changed one more word in the question. Here's a screenshot of the final total…
"Perform" is an even better word to skew the vote than "exercise," since we expect our elected officials to "perform" their duties, whereas "exercise" sounds a lot more optional. In any case, I'm guessing the final result had more to do with how many people managed to vote 200 or 300 times apiece. In no way do any of these questions relate to what any fair sampling of Americans believe. In fact, the questions may not even relate to what the people who voted thought they were voting on. I'd sure be pissed if I took the trouble to rig a poll and vote 600 times and then they changed the wording on me.
P.S. Added a few minutes later: I just did a search and found that other blogs have picked up on this matter, which was called to my attention by an e-mail buddy. The mysterious Atrios claims that at one point, the wording of the poll was inverted to ask, "Should the Senate minority block the body's Constitutional duty to provide the President's judicial nominees with an up or down vote?" so that the old Yes votes now meant No and vice-versa. Here's the Atrios post on the matter, followed by readers reporting even more rephrasing. Sounds like someone's having some fun screwing with the questions to frustrate enemy ballot-stuffers. Either way, it makes my point that online polls are bull. You can vote once and have it cancelled out by someone voting twice…or even by a webmaster who reverses the meaning of the question.