With nearly six months remaining before the November election, Bush led Kerry 48 percent to 47 percent in the survey — a reversal of a poll taken last week, which found the Massachusetts senator with a 1-point edge, 49-48.
Uh, but as the story says two paragraphs later,
The poll, conducted Friday through Sunday, had a margin of error of plus or minus 4.5 percentage points, meaning the presidential race remains close.
The +/-4.5% margin of error doesn’t just mean the race remains close — it means that the 48%-47% pro-Bush result is not a “reversal” of the 49%-48% pro-Kerry result, but is no statistically significant change at all. Both results are dead heats; the seeming “reversal” could easily be the result of random variations in which respondents happened to be picked. A poll conducted on the very same day last week as the 49-48 Kerry poll could easily have reached a 48-47 Bush result.
Someone at CNN must understand this better than most readers do. Shouldn’t it be part of CNN’s responsibility not to mislead the public this way?