Just because you're paranoid doesn't mean that you aren't making stuff up

I've got a few more posts on the recall aftermath. Over at the Marquette University Law School Faculty blog and this blog's home page, I add a little more legal detail to what others have written about the "we got outspent" narrative offered by some Democrats. The shorter version, for this general purpose audience, is that 1) the margin wasn't as large as is generally reported, 2) the margin had nothing to do with Citizens United unless it was to help the forces supporting Barrett close the gap and 3) the margin doesn't seem to have made much difference.

In general, I think it is a mistake for the side that loses an important election to look for scapegoats. Blaming everyone but yourself is a good way to stay a perennial loser.  But at least talking about money and politics is a serious response. There is a narrative bopping around some of the left-leaning blogs that is beyond silly.

Someone is messing with the votes! The exit polls showed that the race was much closer but then the networks called the race for Walker. Exit polls are accurate. What gives? Who stole the strawberries? 

This turns out to be a common meme on the far left, beginning with a really bad article by Robert F. Kennedy, Jr. on the 2004 exits in Rolling Stone. The argument begins with the premise that exit polls can;'t be wrong. They survey only actual voters who report how they have voted as opposed to how they will vote. Exit polls have been used to detect fraud in the counting of votes in other countries. They tend to show more support for Democrats than the actual numbers. The difference is cheating. Toss in a reference to Diebold.

This is moonbattery. Exit polls have their own difficulties. Charles Franklin, Director of the Marquette University Law School Poll, directed to me a good discussion of the issues here. While the discussion relates to exit polling in the 2004 Presidential elections, the issues are the same.
Exit polls must predict the extent of early voting and try to account for it by a telephone survey. They are not, strictly speaking random, in that the polling organizations must identify certain precincts  and then figure out what weight to assign precincts. They must attempt to sample the voters at each precinct randomly - which can be a challenge when the pollster must stand where she is told to stand and voters may try to be polled - or to avoid being polled. Like all polls, they must adjust their numbers to reflect what the actual composition of the electorate is (or is presumed to be).

One significant problem is that exit polls cane be plagued by a large refusal rate. The extant evidence says that there can be nonresponse bias. As noted above, Democrats tend to exit poll better than they actually poll. In the 2008 Democratic primary, Barack Obama did better in exit polls than in his actual match-up with Hillary Clinton.

Exit polls used to verify election results in other countries are often different animals than media polls used to predict the results. They generally involve far more extensive - and often more concentrated polling - to reduce some of the problems described here.

Complaining about the exit polls is black helicopter stuff.

Cross posted at Purple Wisconsin.