I've always wondered how often reporters, writers and others in media bother to check facts given to them by often hardly-disinterested parties.
Yet more evidence that the answer is "hardly ever." To be fair, I can see how the initial error here could have been an honest mistake. The problem seems to be that once it was made, a long line of people never bothered to "show their work" because the data just sounded so good (or too bad, depending on how you look at it).