For those of us who aren't bilogists (or biological investigators) and
have therefore to exercise some imagination, could you be a tad more
explicit in describing the problem? E.g.:
1. "abundance" = "number of winkles counted", or "proportion of
individuals (in algal species, presumably) that have winkles", or
"number of winkles per individual", or ... ?
2. For your several sets of data to be comparable, one would like to
suppose that whatever the measure of "abundance", it was observed in
each of 45 instances (a) for the same length of time and/or (b) over the
same area or volume (whichever is appropriate) of space; or that the
measure of "abundance" is expressed (possibly as some kind of %) in
terms for which variable durations and extents do not much affect the
quantity observed (or reported).
3. You refer to a "Chi Squared test". There are a variety of
so-called chi-square tests. One (or more) of them may be appropriate
for the question(s) you want to ask in the context of the quantities you
have measured. But the ambiguities above make it difficult to offer
useful advice or comment on possible styles of analysis. (Of course,
it's possible that when I understand those details I'd _still_ have
difficulty offering advice...)
4. Your hypothesis that "the winkles are equally distributed...":
Is that your research hypothesis -- that is, what you'd rather like to
be the case, in the best of all possible worlds --; or is it your null
hypothesis -- that is, what one might expect to be the case if there
were no systematic influences (of the kind one might be seeking evidence
for) inducing different distributions between species -- ?
To put a few specifics into _my_ thinking, so you can judge how far
astray I've gone:
(i) If "abundance" = "counted frequency", then your data can produce
the total number of winkles observed on each species (call those numbers
A, B, C for convenience), and the total number of winkles observed in
the experiment is T = A + B + C. There is a chi square test with which
one could ask whether A = B = C approximately (equivalent to your
expression, "33.3% on each". But this would ignore the fact of 15
different readings in each species, and that information would (one
supposes) be useful in addressing the inherent variability of
"abundance" within each species.
(ii) If "abundance" is some kind of proportion, as suggested by your
use of "33.3%" and references to Mann-Whitney and "t" tests, it would
be possible to address the question of whether these proportions are
equal across species by means of a variation of analysis of variance
(ANOVA). This form of analysis would explicitly use the 15 different
readings within each species.
On Fri, 4 Apr 2003, CK Christopher Kent wrote:
> I've recently carried out a biological investigation in which I was
> comparing the abundance of winkles on "three" different algal
> species. I have three sets of data for each species each containing
> 15 results. So a total of 45 readings. My hypothesis is that the
> winkles are equally distributed on all three wrack species i.e.
> 33.3% on each.
>
> I was trying to verify my results and see if this is true.
This sort of begs for the question, "What results?" (And why do they
need "verigying"?
> I am not sure if Chi Squared test works with three sets of data.
> Is it possible?
See comments above...
> Moreover, I thought of using Mann-Whitney or T-test
... which _could_ be appropriate, depending ...
< snip, the rest >
HTH. -- DFB.
-----------------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
