This is one example of statistical sloppiness in reporting, comprising: (a) sloppiness about the quantity or proportion being asserted (e.g. "many" rather than "at least 4000" or "10%" or "24"), and (b) sloppiness about the confidence interval.

I'd love to see a sidebar to articles asserting trends providing both a clear quantity estimate and the confidence interval around it. I'd even settle, in early articles on a supposed trend, for something like "We're asserting that this has happened more than N times." where N might be as small as 0.

I'm aware that my proposed standard could even be applied to this comment:), so I'm simply asserting that this sloppiness exists. Here's one exemplary takedown: http://blogs.crikey.com.au/contentmakers/2010/03/27/did-the-...