Is there anyone on the list who might have an answer to this question?

We have data on the body mass of parent birds feeding young in the nest
collected over the whole day (dawn to dusk) using an electronic balance
hooked up to a computer under the nest. The data include 10  males and
10 females, with mass estimates for each of 30 time periods (0.5hrs)
throughout the day, for 3-5 days per bird. Our question is what
determines the shape of the trajectory of increasing body mass for
individual birds over the day? Is it the particular bird, the type of
day (i.e. weather), or the work rate of the parent feeding the young?
Some birds gain lots of mass early in the morning, some gain most late
in the afternoon, and others show more individual trajectories.

Statistically, we think could approach this either as: 1) a
curve-fitting exercise (e.g. 3rd order polynomials?) and perform an
analysis (e.g. MANOVA?) on the parameter estimates from the curves (e.g.
slopes, etc.), with bird, day and nest as factors; or 2)  perform an
analysis (e.g. repeated measures ANOVA?) on the data itself, with time
period as the repeated measure (or within-subjects effect) and bird, day
and nest as factors (or between-subjects effects). The second approach
would be preferable, because it allows us to look for a wider variety of
differences within the mass trajectories (i.e. deviations over just a
few time periods, as well as general daily trends). However, the second
approach does have the problem that the mass of a bird at one time
period is not necessarily independent of its mass in the previous and
subsequent time periods - i.e. there is more covariance between data
from the most adjacent time period. Various other possibilities have
been suggested to us, such as time-series analysis or profile analysis,
but as non-experts in statistics we've had problems finding out more
about these, and especially finding solutions within the stats packages
we have access to (i.e. SPSS, MINITAB, etc.).

I would therefore be grateful for any help or advice anyone can provide
on this one.

Jon Wright.


_______________________________

Dr Jonathan Wright,
School of Biological Sciences,
University of Wales, Bangor,
Gwynedd LL57 2UW, UK.
Phone: +44 (0)1248 382313
Fax: +44 (0)1248 371644
E-mail: [log in to unmask]
_______________________________




%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%