On Sat, 30 Aug 2003 02:01:27 GMT, "Gordon Couger"
wrote:


"Torsten Brinch" wrote in message
.. .
On Fri, 29 Aug 2003 21:42:10 GMT, "Gordon Couger"
wrote:

All the data points used in the study except the discarded
ones are in the paper. That should be enough to duplicate
the statistics.

Um. The plotted points you see in the graphics, and the numbers in
tables are not raw values, they are means (n=3), three replicates
per treatment and sampling date.

However, apparently it is your hypothesis while looking at these data
points, that they are not all there, that some data points have been
discarded. But hey, that should be easily verifiable. Check, and you
should find for some sampling dates in plots and tables, that data
points are missing. :-)

Unfortunately for your hypothesis, they are all there.

There is no hypothesis the paper clearly states outliers are discarded.

I am talking about your hypothesis that "data that didn't agree with
the findings was discarded".

God help you, if you have nothing else to base this on, than what is
clearly stated in the paper, that outliers in raw data were removed
from datasets before variance homogenity of data was evaluated
in residual plots. Whatever you may think of removal of outliers at
this particular stage in the statistical analysis, it obviously does
not and cannot constitute the authors discarding of data that doesn't
agree with the findings. There are no findings at this stage, just a
mass of raw values, unfitted to any model, untested for any
significant differences between them.