Biology Reference
In-Depth Information
FIGURE 9.2
The effects of size, sex, and region on
alpine chipmunk mandible shape:
(A)
size;
(B)
sex;
(C) region.
measurement error. Systematic errors are most worrisome when they interact with a bio-
logical factor, such as parallax, which is most pronounced at the edge of the camera's field
of view. It may therefore have a greater influence on larger specimens than on smaller
ones if the field of view is held fixed. In contrast to systematic errors, random errors do
not occur in a regular manner, but they are typically modeled by a statistical distribution,
most commonly by a normal distribution. To reduce the impact of measurement error we
can first decompose the error into its sources by a repeated-measures nested MANOVA,
taking multiple measurements of the same specimens and treating each potential source of
error as a factor. For example, we might have a choice between two imaging methods, e.g.
CT scans and photography. We would then image each specimen several times using both
methods, and digitize each image several times. We would then have the factors: (1) indi-
vidual (the biological variation among members of the population); (2) imaging technol-
ogy; and (3) measurement. All are random and “Imaging technology” is nested within
“Individuals” and “Measurement” is nested within “Imaging technology”. By decompos-
ing the variance into these sources we can determine whether the technologies produce
