Monday, September 29, 2008

You are now (almost) allowed to pay attention to daily tracking polls

A not just because Obama is up by 5+ points(heh)... but an interesting post by John Judis over at The New Republic, noting that the predictive power of daily tracking polls increases as of October 1.
Since 1960, Gallup’s tracking poll registered the winner in the popular vote (including Al Gore in 2000), eleven of twelve times. The one exception is 1980, when Jimmy Carter still led Ronald Reagan by 44 to 40 percent. The late September-early October polls have not necessarily predicted the final margin. Third party candidates usually screw up the total, because their support usually drops by the final election. and generally the race narrows somewhat by the end. In 1996, for instance, Bill Clinton led Bob Dole by 14 percentage points on October first. Clinton’s final margin would be 8.5 percentage points. In 2004, George W. Bush led John Kerry by 8 percentage points. His final margin would be only 2.4 points. But in six of these elections--1960, 1964, 1976, 1984, 1988 and 2000--the final margin was different from the October first polling results by less than three percentage points.

October 1 seems like a bit of an arbitrary cut off and 12 elections isn't a terribly large sample, but it does make sense that people are paying significantly more attention now than in July... regardless, we can all now officially remove the caveat that "daily tracking polls are meaningless" before we start obsessing about them. You can now obsess with a clear conscious!