Myth of the Myth of the Month: [This item rescued from oblivion in a previous item] Mark Blumenthal, the pollster formerly known as Mystery Pollster, explains the presence or absence of a pro-Obama poll “bounce” after his State of the Union speech by arguing that there ain’t no such thing:
The answer is that if any of these changes turn out to be more than statistical noise, they are probably incidental to the speech. The State of the Union bounce is largely the stuff of myth.
I’m not convinced. 1) The Gallup “myth” article Blumenthal cites compares polls taken anywhere from a week to over a month apart. Usually they’re two weeks apart. You wouldn’t expect a poll taken every two weeks to pick up a bump that might last a few days. Gallup has only been doing the sort of daily tracking poll that could capture such an ephemeral bump since late 2008, apparently. 2) There’s no bump–except when there is. In Gallup’s table, Clinton got a 10 point bump in 1998, a 6 point bump in 1996, and a 4 point bump in 1994. Why shouldn’t Obama get one of those bumps? In fact, Obama himself did get a “notable” 8 point bump in 2009, says Gallup–the only reason that’s not on Gallup’s table is that it wasn’t officially a “State of the Union” speech.
Blumenthal can easily come up other explanations that would account for such bumps–Clinton’s 1998 address, for example, came in “the context of the Monica Lewinsky story, which had broken a few days before the speech.” But I can come up with a competing explanation: it was a good speech!
Interestingly, Obama’s 2009 bump was captured by Gallup’s new daily tracking poll in the three days following the speech–and all of Clinton’s bumps showed up when the (non-tracking) poll happened to be taken three days or less after the speech. Hmm.