and some research reports could profit from rounding too…
Just shy of 4 years I own an electric “maxi scooter”, a Vectrix #vectrix http://michaelbach.de/misc/roller/ When today I rode this really fine machine, the odometer numbers passed 15000.0 [kilometers]. It's amazing: (nearly) any number on an odometer occurs only once in a lifetime (loosely stated), yet some are special to us – we can't escape seeing patterns.
Some researchers could learn a little from this: When I read a paper giving age of a 20odd group as 57.86±12.77 years, or giving thickness of a “fibrous layer” as 8.666±2.08 µm (from a histologic section) I mentally downgrade that manuscript immediately – and I get a lot of those. There's actually a statistics site on this topic: http://www.sportsci.org/resource/stats/digits.html.
The last example above also reports different precision for mean and SD; following Hopkins, Karen and I criticised such recently as well: http://dx.doi.org/10.1007/s10633-010-9249-7.