Re:Results of yesterday's personality poll: we're strange.
by basman on Tuesday November 17, 2009 @11:10AM
(#946639)
His analysis doesn't depend on how many significant digits there are: he's doing a hypothesis test, so all that matters is whether one number (a p-value which was not shown) is bigger than a threshold (.01). The fact that it is in most of the cases means that the variance in results that you mention has been handled by the sample size.
So I do think the results are meaningful.