Politico has a story today on the Employee Free Choice Act, and how the numbers each side is citing aren't matching up very well:

• “Nearly three-quarters of the public — 73 percent — support the Employee Free Choice Act,” crows the AFL-CIO on its website.

• “Seventy-four percent of voters oppose the Employee Free Choice Act,” concludes research cited by the pro-business Coalition for a Democratic Workplace.

No, the electorate hasn't gone mad. Those who crafted the questions simply framed them in a way that favored the outcome they desired -- a tactic that is all the more effective when the sample doesn't have much of a predefined opinion on the matter either way, as the article posits.

The takeaway:

With questions like these, surveyors are doing something more like what campaign pollsters do: testing messages to see where public opinion is likely to go.

There’s nothing wrong with that per se, but when advocates or opponents use these numbers out of context in statements or press releases, they are essentially attempting to form public opinion by asserting that they already know what public opinion is — which can produce the kind of absurdist discrepancies that have been a hallmark of the labor/management smackdown thus far.

So what’s a poll-conscious legislator to do? The minimum one needs to know for a number to be useful, says Blumenthal, is who commissioned the study, when it was conducted, who was asked the question, what the exact text of the question was and what questions were asked previously.