Statistics Question

David Hershey dh321z at
Tue Sep 28 18:40:41 EST 2004

I'm not an expert in statistics but I was taught that
you can often fit a diverse group of curves to a
widely scattering group of biological data. However,
many of them will not fit well nor will they be the
type of curve expected for that kind of data. For
example, it seems to make little sense biologically to
fit a straight line to growth data that are known to
produce an S-shaped curve. 

I recently questioned the validity of data presented
by Rice and Maness (2004) on assaying the level of
toxic compounds in tree leaves. All that was done was
to graph raw data lacking zero concentration controls.
There were no statistics such as means, standard
errors or analysis of variance. Curves were fitted to
the raw data, and those curves used to extrapolate to
zero concentration rather than generating actual data
for zero concentration controls. There were no
correlation coefficients or other statistics to
quantify how well the curves fit the data. They did
not seem to fit well.

When I questioned the lack of statistics (Hershey
2004), Rice (2004) replied that 

"... most classes taught by readers of this journal
would not be prepared for these analyses." 

That seems false and also insulting to most students.
It was also contradicted by Rice himself in the same
issue. Rice and Griffen (2004) contained statistics
and encouraged students to statistically analyze data.

Rice (2004) also first revealed that statistical
analysis was performed for the data in Rice and Maness
(2004). He then claimed that the statistical
differences were proof that the experimental methods
were valid. That seems illogical. If an experimental
method is invalid, then it follows that any
statistical difference based on data produced by an
invalid method will also be invalid.

In July 1999, Ross Koning contributed an item to this
newsgroup titled 
"USING the Supermarket Botany Article."

It suggested teachers assign students to fact-check
flawed articles and discover the errors for
themselves. The articles by Rice cited below seem
useful for that kind of exercise. 


Hershey, D.R. 2004. Brine shrimp "bioassay" problems.
American Biology Teacher, 66, 474-475.

Rice, S.A. (2004). Response to Brine shrimp "bioassay"
problems. American Biology Teacher, 66, 475.

Rice, S.A. and Griffen, J.R. (2004). The hornworm
assay: Useful in mathematically based biological
investigations. American Biology Teacher, 66, 487-491.

Rice, S.A. and Maness, I.B. (2004). Brine shrimp
bioassays: A useful technique in biological
investigations. American Biology Teacher, 66, 208-215.

Do you Yahoo!?
Declare Yourself - Register online to vote today!

More information about the Plant-ed mailing list