Dramatic Swings, Wide Disparities Undermine WI Polling

America, your heroic pollsters have spoken! The Wisconsin primary will be won by...somebody. Maybe even Brett Favre! As always, nobody seems to know a cotton-picking thing, and they're prepared to trumpet their uncertainty with authority. Maybe you're the quaintly-named Research 2000, and you have Obama with a small lead in Wisconsin. Or maybe you are Public Policy Polling, and you have Obama up big. Or maybe you are ARG and you've decided, "What the hell? Let's just cover all bases!"

No polling organization has swung as wildly as ARG has since Valentine's Day. Their February 15-16 poll of Wisconsin voters looked like bad news for the Obama campaign and showed enough of a Clinton lead that it made one wonder why the candidate would choose to leave the state:

Just days later, however, ARG came out with a poll that suggested a crushing lead for Obama:

Of course, to some degree, it's completely understandable that there should be wide disparities. Wisconsin allows voters to register on the spot when they come to the polls, which means a late surge of new voters is always possible, and always unquantifiable.

How important are the vagaries of Wisconsin voter regulations? They could be significant: the conventional wisdom is that the Obama campaign is better poised to bring new voters into the process, but as we've seen, time and again, in these primaries, Clinton has had extraordinary success landing the late-deciders. Still, we're talking about a single poll that's flipped from one side to the other by a double-digit degree. With that in mind, ARG's stated margin of error of 4% seems like a naive underestimation.