Monday, September 22, 2008

Rogue story: Shoot on sight

If this story shows up in your paper tomorrow, someone has done a very very bad thing. Take it away, McClatchy Washington Bureau:

A majority of Americans think the United States isn't winning the war on terrorism, a perception that could undermine a key Republican strength just as John McCain and Barack Obama head into their first debate Friday night, a clash over foreign policy and national security. [Whose opinion is "undermine a key Republican strength," and is it based on anything real and measurable?]

A new Ipsos/McClatchy online poll finds a solid majority of 57 percent thinking that the country can win the war on terrorism but a similar majority of 54 percent saying that the country is NOT winning it. [Hmm. So any second now, we'll not only find out what makes a majority "solid" in this poll, but how Ipsos/McClatchy defined "war on terrorism" so as to assure us that respondents were talking about more or less the same thing?]

Skip a few grafs of debate-related setup for this:

... If Americans are turning more pessimistic about the so-called war on terrorism, it could present a challenge for McCain. [All of a sudden, it's become the "so-called war on terrorism"? That's an interesting and important change -- but I doubt it reflects the question you asked, and that underscores why you need to say what you mean by it.] Voters traditionally trust Republicans more than Democrats to handle terrorism and national security, but a loss of confidence in the results of the fighting so far could erode that edge. [One fiction, one guess. There is a body of stuff out there suggesting that, at least in the modern polling era, voters trust Republicans more on defense; it's the "issue ownership" concept. But it doesn't extend to terrorism, because there is no "tradition" of handling terrorism in US politics. And asserting that Americans are "turning more pessimistic," or that there's a "loss of confidence in the fighting so far," would require, at the least, one earlier poll that actually measured pessimism and "confidence in the fighting so far."]

... The survey also found that Americans think by 57-43 percent that Afghanistan is now a more important front in combating terrorists than Iraq is.

Why the "now"? Where's the comparable poll to support the relevance of this distinction? Does McClatchy think Iraq has always been part of the "war on terrorism"? How is the question phrased? And how about some numbers so we can tell a little more about those majorities the writer is flinging around?

...The poll has no statistical margin of error [news flash: That's the only kind there is] because the online sample isn't a random one that mirrors the population within a statistical probability ratio, although Ipsos weights the sample to resemble U.S. demographics.

Are you saying that it's a convenience sample? Because if it is, all the conclusions are bogus. You can't say anything about what "a majority of Americans" think. If Ipsos has a way of explaining how its "online panel" (which, a later blurb declares, "isn't a random sample") is chosen that might mitigate this otherwise pretty obvious conclusion, we need to hear it. Otherwise, just go down to the mall and ask a few dozen people what they think; it's as good as any other convenience sample.

Broken record time again. The McClatchy Washburo (specifically, the pre-merger Knight-Ridder side of it) won a lot of well-deserved praise for its prewar skepticism and its critical coverage of the Iraq runup. That ought to be a proud tradition. Why give it away by running made-up conclusions about bogus data?

Labels: , ,


Post a Comment

Links to this post:

Create a Link

<< Home