Just before the Wisconsin recall election on Tuesday, the Real Clear Politics website posted its final average.  It was a 6.7 % lead for Gov. Walker.  Walker won by 6.9 %. That’s close.

In the final hours of the election, theleft-wing media gleefully announced that the exit polling indicated a dead heat, meaning that Mayor Barrett had real shot at winning.  The exit polling wasn’t necessarily wrong, it was the reporting.  Raw exit poll data should never be reported, because they mean nothing until they are corrected to mitigate the inevitable sources of bias in the sampling.

For example, Jon Cohen wrote yesterday that different types of people vote at different times of day.  It’s not hard to imagine that a lot of people who work for a living will vote after they get off work, resulting in a big surge in Republican votes later in the day.

Mr. Cohen identified several sources of bias, but he missed the biggest one.  He made several references to “random” sampling.  In practice, exit poll sampling is anything but random.  I was at the Shannondale Elementary School on Election Day in 2004.  Nationally, early exit poll numbers were erroneously being reported to suggest a huge John Kerry win.  Democrat operative John Schrum famously asked Sen. Kerry, “Can I be the first to call you Mr. President?”

I heard the reports, but I didn’t believe them because I saw how the exit polling was being done.  Process matters.  The woman doing the polling, obviously an untrained temporary worker, waited behind a table for people to walk over to her.  The people who responded tended to look like her … young, female, and minority.  They also appeared to be the people not in a hurry to get back to work.  All of these factors would have skewed the exit poll results toward the Democrats on the ballot.  There was nothing random about it.  I’m sure the scenario I observed was repeated in many other places.  Temporary workers would be (I’m guessing) disproportionately young, female, and minority, resulting in more exit-poll respondents from those groups.

Biases in the sample can be corrected, but that process surely takes hours, if not days, to accomplish.

The process for correcting exit poll data was described by Sean Trende:

In other words, the exit pollsters in the field missed a lot of Walker voters. Now, exit pollsters have ways to fix this. For one thing, they weight different regions of the state to the actual vote returns. For example, if northeast Wisconsin exit polls are showing a 50-50 race, and the actual results are 60-40 for Walker, they will simply assign greater weight to a Walker respondent in the region, bringing the reported result in line with the actual result.

… if the non-respondents are disproportionately male, white, and older, the exit pollsters will make sure that the final weights account for those discrepencies.

Going back to Cohen, his bottom line was this:  Exit polls tally how different groups voted in an election.  They do not predict results.

I wouldn’t get too exited about the suggestion that Tuesday’s exit polling indicates a huge lead for Pres. Obama in the November election.  First of all, that election is months away.  Further, we never know if they are reporting raw (biased) data or corrected data.  Gov. Haley Barbour said Wednesday that the corrected numbers indicate a dead heat.  Trende also talked about this in greater detail.