A Nov. 14 NewsBusters post by Scott Whitlock rags on CNN's Bill Schneider for stating that the country has "no confidence" in the Bush administration based on a USA Today/Gallup poll that showed President Bush's job approval rating at 33 percent with a disapproval rating of 62 percent. Added Whitlock: "Of course, Schneider didn’t mention that the poll he was referring to sampled only "1004 national adults," as opposed to registered or likely voters. Both of which, would have resulted in more conservative findings."
But Whitlock doesn't explain why the opinions of non-voting citizens are less valid than those of "registered or likely voters," beyond apparently being more liberal. A cross-section of "national adults" does, after all, cover the entire nation -- why is that a bad thing? Nor does he explain why "only" 1,004 respondents is an insufficient or invalid sample for a poll -- in fact, the sample size of many polls is 1,000 or less.
NewsBusters frequently complains that poll results aren't conservative enough, despite the fact that pollsters are using standard sample sizes (as above) and political breakdowns that accurately reflect the American population (see Noel Sheppard).