Environmental Engineering Reference
In-Depth Information
plausible alternative hypotheses to explain a particular pattern or process
may exist.
To know if a hypothesis is a reliable explanation, it must be tested exper-
imentally by manipulating the putative causal variable(s).This is the domain
of the Hypothetico-deductive method.
Hypothetico-deduction
In the Hypothetico-deductive (HD) method, one deduces testable predic-
tions based on the hypothesis.Then, most critically, one tests the hypothesis.
Let me illustrate by building on our example. Suppose that we chose to
test the hypothesis that higher plant species diversity leads to higher insect
diversity because of the inherent food value afforded by the variety of plant
species. One possible food source for many insects is flowers because they
provide nectar and pollen. Each plant species will have unique flower char-
acteristics that attract different insect species. So, more plant species lead to
a greater diversity of flower types, which then offers a greater variety of
pollen and nectar resources.The logical deduction then is that if we manip-
ulated the diversity of flowers in a patch, we should see an associated change
in insect species diversity.
One way to experiment with flower diversity is simply to cut off selected
flowers from plots in the field containing high plant diversity. Such a ma-
nipulation is called an experimental treatment or experimental perturba-
tion. If our hypothesis (and associated logical deduction based on it) is
correct, then we should see a decline in insect species richness in the plot
following the application of the experimental treatment. If we see the de-
cline, we may conclude that a higher diversity of food leads to higher in-
sect species diversity. But, is this conclusion reliable? The answer is no.This
is because under the current experimental design we have failed to include
an experimental control.
The experimental control serves as a critical, unmanipulated baseline for
comparison with the experimental
treatment. It is established simply by
leaving some plots alone, or techni-
cally leaving plots “untreated.” The
control allows us to tease apart the ef-
fects of the experimental treatment
from random and potentially con-
founding environmental effects. For
The point is if we do not have an ex-
perimental control, then we cannot
draw reliable conclusions about our
hypothesis test. No Control =
No Conclusion.
