8/7/2006 - Undecided in Wisconsin

The Mystery Pollster has an excellent recent piece on undecided voters.  Some recent data contradicts one of the cardinal rules of polling:  Undecided voters overwhelmingly break for the challenger.

I heard about this notion for the first time during the Bush-Gore election in 2000.  I was cudgeled with it in 2004, when every news reports seemed to begin with a reminder:

"October 23rd, 2004:  As research has demonstrated, time and time again, undecided voters break overwhelmingly for the challenger.  In other news, the Red Sox swept the Cardinals to win their first World Series in 86 years…"

Even as I originally listened to the networks claim how undecideds would break, I was dimly aware that this was an example of biased reporting -- most of these networks supported the challenger and they surely influenced decisions.  The morality of this can be debated elsewhere.

Well, MP just revealed that this conventional wisdom was based on data as recent as 1982.  An update of the analysis reveals something noteworthy: that percentage has been trending downwards.  In 2002 and 2004, the split was closer to 50/50 (with the challenger still having a slight edge).  What drove the change?  Well that's far from clear and pollsters will be happy to grab any number of snowflakes out of the thin air (from post 9/11 theories to improved sampling methods).  But my question is this?

If the "fact" that undecideds break for the challengers is now entrenched in the fourth estate, I imagine that we will hear much of this in the upcoming mid-term elections.  If the media still relies on professional polling organizations to provide the bulk of the analysis, you may start hearing a new tune in the near future.

12,100 late-breaking steps today.  Cheers,