Environmental Engineering Reference
In-Depth Information
Variability and associated uncertainty introduced by random environ-
mental effects can be dealt with by repeating experiments enough times that
we can calculate the odds (or risk) that an experimental treatment (or a
management prescription based on it) provides a rare or unusual outcome.
For example, if we conducted the same fragmentation experiment twenty
times under different environmental conditions and found that it does not
alter species diversity (that is, we could not reject the hypothesis) nineteen
times out of twenty, then we say that there is a 5 percent chance that habi-
tat fragmentation will collapse bird species diversity.The implication of this
calculation for policy is that if the goal is to develop land and safeguard
species diversity, then we can claim that there is a 5 percent chance that frag-
mentation will be harmful.This information is valuable because it allows
policy makers to decide whether the benefit of implementing a particular
management prescription is worth the risk of failure.
But uncertainties still underlie decision-making, which opens the pos-
sibility for making errors.The first kind of error, known as a Type I error,
arises when we reject the null hypothesis when the null is in fact true.The
likelihood of doing this can be kept marginally small simply by being very
stringent about the criterion we use to reject the hypothesis.The norm in
ecology is a 5 percent risk of making a Type I error. Criteria for ensuring a
small likelihood can be calculated whenever it is possible to quantify the
mean and degree of variability in experimental response among different
replications—often this can be done with as few as three replications.The
second kind of error, known as a Type II error, arises when we accept the
null hypothesis when in fact it is false.Type II errors can only be controlled
by gaining understanding of typical experimental responses versus compar-
atively rarer ones. Our confidence in discerning what is typical and what
is rare is only boosted by increasing the number of times an experiment is
replicated. But, most experiments conducted on scales relevant to ecosys-
tem management are neither easy, nor practical, nor often affordable to repli-
cate many times. Consequently, there is often a greater chance of
committing a Type II error than a Type I error.These different risks of error
carry different ethical implications.
The normal practice in science-informed policy and management is to
weigh the consequences of action versus making a Type I error. Under this
condition, policy favors the interest of land development if the decision is
to develop in light of scientific evidence that it will not damage the natu-
ral system. In turn, the burden is placed on the public, or regulatory agen-
