Re: test for small paired sample size

On Wed, 27 Dec 2006 18:56:08 +0100, Knut Krueger <Etron777@xxxxxx>
wrote:

Hi to all
is there any meaningful analysis for the following data:
a: 62 57 230

b: 10587 5721 3550

we have only three pairs of data and everybody can realize that "a" is
increasing and "b" is decreasing. But is there any statistical function
for such a small sample size to get any "p-value" (for the journals...)

Typographical error? The "a" is not monotonically increasing.

The Pearson r can be computed for n=3, though it could
be hard to justify wanting to do it.

If these represent a time-series, then the p-level is not correct,
because the measures are presumed to have serial correlation.

--
Rich Ulrich, wpilib@xxxxxxxx
http://www.pitt.edu/~wpilib/index.html
.