Biology Reference
In-Depth Information
CHAPTER
8
Statistics
Organisms vary for reasons beyond our control and often beyond our understanding.
Variation is of obvious biological importance because evolution could not occur without it,
but variation is also a source of frustration for biologists, as evident in what has been
termed the Harvard Law of Biology: “under the most carefully controlled conditions, bio-
logical material does whatever it damn well pleases” (quoted by Ellen Larsen [2005,
p. 115], in a topic entirely devoted to the subject of variation [
Hallgr
´
msson and Hall,
2005
]). Because organisms, even when reared under carefully controlled conditions, vary in
the outcome of development, we cannot assume that all those outcomes are due to the treat-
ment that we applied experimentally. The problem of interpreting experimental outcomes
is obviously much more difficult when nature did the experimenting, not us. Given that
there will always be variation that we cannot explain, we cannot safely ascribe all experi-
mental outcomes, whether the experiments are controlled or natural, to the treatments. This
inexplicable variation is the “error” term in statistical analyses
any variation that we can
explain is not error so long as the factor explaining it is included in our statistical model.
Variation further complicates drawing inferences about the experimental results
because we rarely, if ever, measure every single individual in the population of interest.
Almost always we instead draw a sample from that population and hope to infer some-
thing about the population from the sample. For example, if our experiment is run in the
laboratory, we are rarely asking questions about the response of our particular laboratory
population to the specific treatment that we applied. Should we give some mice liquid
diets and others regular laboratory pellets and measure their jaws to see the impact of die-
tary consistency on their jaws, we are not asking whether these particular groups of mice
differ. Rather, we want to know if mice, more generally, will differ in their jaw morpholo-
gies because of differences in dietary consistency. Similarly, when we analyze natural
populations, we are rarely interested solely in the specific organisms that we measure, we
want to generalize from those samples to the population as a whole. For example, we do
not ask whether adult chipmunks (Tamias alpinus), collected between 1911 and 1919,
whose skulls are contained in the mammal collection of the Museum of Vertebrate
Zoology, vary in jaw shape because they vary in size, or if these particular chipmunks are
