Friday, October 04, 2013

Slimdown update: Dunno. Is they?

The short answer is "no, but thanks for asking." A more specific version is "we have no idea, but none of the evidence in this story suggests that they is, and to the extent it suggests anything, they isn't."

This is actually a fairly interesting bit of news practice, in that the story's gone from the lead position (this afternoon) to oblivion -- not that it's vanished, but it's no longer findable from the home page, and it hasn't picked up a comment in four hours (comments are showing up every few seconds on the current lead story at this writing). And it's largely contradicted by the usual story Fox writes to accompany a Fox poll.

Anyway, here's the now-oblivionated "Throw the bums out? Budget deadlock portends 2014 backlash" tale:

While the House and Senate point fingers at each other over the budget impasse, new polling suggests the only thing they might accomplish is getting thrown out of office.

Hold that thought.

Despite a robust campaign by Democrats to pin the blame exclusively on their GOP colleagues, the American public appears fed up with just about every elected member from 1600 Pennsylvania Ave. to the Capitol dome.
The public sentiment should bring a sense of foreboding for any lawmaker seeking re-election in 2014, with evidence mounting of an anti-incumbent backlash in the making.

That's a bunch of fact claims about "new polling" and "public sentiment" and "evidence mounting." Shall we continue?

Newly released Fox News polls show Congress' approval rating is 13 percent. President Obama's is better, but still not great for a recently re-elected president, at 45 percent. And more people disapprove than approve of either party, though the numbers for Republicans are worse.

Mostly true (though it's one poll, not "polls"), but irrelevant to whether who might be mounting what, sentiment-wise. For that, you need to look at some other polls, which is the sort of thing Fox is actually pretty good at in its regular poll coverage:

President Obama’s job ratings are down-right glowing by comparison -- yet they are still underwater.  Some 45 percent of voters approve of his job performance.  That’s up from 40 percent approval in early September, around the time he was considering military action against Syria.  Forty-nine percent disapprove of Obama, down from 54 percent disapproval last month.

With a sample this size, both those changes are approaching traditional significance (no, I don't know why Fox rounds the margin of sampling error; the Times does too, and they should both stop, because it's stupid and misleading). If you sacrifice baby squirrels at the shrine of the P value, you can ignore the apparent improvement on grounds that it's nonsignificant -- but if you take it as evidence of an anti-incumbent backlash based on this week's events, you should stay away from those friendly games of chance.

Congress? Again, give Fox credit for posting the complete results, which show that Congress is on the low side of its approval for the last 24 months. If you wanted, you could conclude that those results are forming a normal distribution around a stable population value for approval of Congress. I think it's more sensible to conclude that approval of Congress has been moving up and down, leaving a couple of questions about this month's result: Is it an outlier, and if not, what caused it?

If you're curious, you could look at a larger bunch of more current survey results. (RCP doesn't bother to provide sample sizes, but if you still think the "RCP average" is meaningful, imagine what the average of 1,500 likely voters, 952 registered voters, and 2,242 adults must look like.) Again, you could say that's not enough information among these unrelated polls to draw any meaningful conclusions, and I'd agree -- though if the only post-SlimFast sample finds a higher value for approval, it'd be conclusively dumb to suggest that SlimFast had made approval go down.

To be clear, this is an overtly partisan story ("The president is noticeably trying to insulate his side from a political backlash"), but its numerical dishonesty is nonpartisan. It's the kind of crap journalists produce when they want numbers to mean something because, dammit, they ought to mean something. It's odd to have one Fox story demonstrate that another Fox story is pure hackery, but we live in odd times.

Labels: ,


Post a Comment

Links to this post:

Create a Link

<< Home