When is a low poll number significant? Note the commentary from Fox News Sunday's pundit panel about
the latest AP-Iposos poll (margin+-3%).
It showed a 46% approval rating for Pres. Bush's handling of Hurricane Katrina relief. Weekly Standard editor Bill Kristol said that was "no big deal", that the number is not bad and matches Bush's general approval rating in the low-to-mid 40s.
The responses for how to pay for Katrina relief were as follows:
42%: Cut spending on Iraq war
29%: Delay or cancel tax cuts
14%: Add to federal debt (borrow it)
11%: Cut domestic spending
Fox News Washington editor Brit Hume dismissed these results because none got a majority, and the Iraq poll response was phrased way too abstractly to be meaningful(?!).
So, both being a minority, what makes 46% a satisfactory value, but 42% not worthy of consideration? Both questions were clear and gave a range of balanced answers. I assume the same people were asked both questions, so it's not a size/sampling issue. Significantly, the full 3% margins of error for these figures overlap.
Maybe it's that Kristol has a different standard for polls than Hume.
As a side note, the top two responses were solutions from the left. If you add "Reduce Iraq war spending" to "Delay/cancel tax cuts", 71% of people agreed with a solution contrary to Pres. Bush's stated priorities.
But, of course, that question was just badly phrased. Brit Hume raised an argument to this effect that was both a straw man and implied the responders were clueless. He was confused about the 42% for reducing Iraq spending, because they obviously didn't want to cut money for ongoing military missions, nor for the troops, nor for pulling out immediately. But of course there are more options than those Hume raised, such as a compressed draw down, increased accounting vigilance to avoid billions lost to embezzling, and so on.