1.14 Conclusions
It should be clear from the above that there are important and close connections
between biological knowledge about systems, logical development of ideas, models
and hypotheses, design of sampling, analysis of data and interpretations of the
information. Of course, like everything else in science (and life itself), there will
be continuing disagreements about the role and purpose of hypothesis-testing in
marine research (see Stewart-Oaten, 1996; Suter, 1996). This can be a rich debate
or a sterile one. Here, it would be a distraction, so it will receive no mention.
Suffice it to say that those who favour risk analysis in environmental assessments
and those who favour estimation of magnitudes of differences among sites, times,
habitats, conditions, rather than structured tests of formally explicit hypotheses
would surely all agree about the need for care in developing sampling programmes.
Without advance thought about appropriate scales of replication, it is no more
possible to put valid CI around estimates of mean numbers of animals (or any
other parameter being estimated) than it is to do a valid statistical test of some
hypotheses about a parameter. Estimators and other describers of the world have
the same issues for design as do experimentalists testing hypotheses in advancing
scientific understanding.
are no reefs; for more details, see the text)