How pollsters can rig the polls

An interesting experiment out of Wisconsin today that shows why you should be wary when you hear about poll results.

A prominent Wisconsin pollster, Charles Franklin, conducted a poll of 716 Wisconsin voters asking their opinions on four things: The current state of the economy, how the economy will fare in the coming year, President Obama and Wisconsin Gov. Scott Walker. But there was a wrinkle. Half the voters were asked the economy questions first, then were asked about Obama and Walker. The other half were asked about Obama and Walker, and then about the economy.

David Lauter with the Los Angeles Times reports on the results: Those who were asked about Obama's job performance first favored him by a stout 52 percent to 39 percent. But those who were asked about the economy first were evenly split on Obama, 48 percent to 48 perent.

Pollster Franklin says that just mentioning the economy plants enough of a seed to make voters express a less favorable opinion of Obama, Lauter reports.

It's a good reminder: Poll results can hinge on how a question is worded and, as this experiment showed, even on how the questions are ordered. So organizations seeking a certain outcome can nudge the results their way by crafting the poll a certain way. Even polling organizations with no hidden agenda will inevitably produce results that are colored by how the questions are worded and ordered (and on many other factors, including how representative their sample is).

So when you hear Obama or anyone else is up or down, look for the full questionnaire, and consider how it might have affected the results.

-- Taylor Batten