Geoscience Reference
In-Depth Information
be. When haphazard spatial samples are used, meaningful general-purpose
analyses are difficult. Haphazard and unplanned sample sites may provide
interesting observations but should not be considered part of scientific moni-
toring plans.
10.2.4 Convenience Samples
An extremely popular form of nonscientific designs are convenience sam-
ples. Convenience samples are those that include sites because they are easy
and inexpensive to reach. Common convenient sample sites include those
located close to research facilities, roads, or other access ways (such as trails).
Because budgets are always limited, convenience samples are a tempting
and popular way of reducing costs. The danger with convenient samples is
bias. Convenient sites, or easy-to-access sites, may or may not represent the
entire population. Sites that are easy to access can and often are changed by
the fact that they are easy to access and therefore are different from more
remote sites. Furthermore, to justify that the convenient sites are similar to
remote sites, it is necessary to place at least some sites in remote areas, and at
that point, it would be easier and more defensible to select a (perhaps strati-
fied) probability sample from both areas.
Convenience samples can employ probability samples of the areas that are
easy to access yet remain a convenience sample if the study area is not rede-
fined to correspond to the region that was actually sampled. In other words,
drawing a probability sample from easy-to-access areas is perfectly valid
unless inference to a larger study area is desired. For example, drawing a
probability sample near roads is perfectly acceptable as long as it is clear that
estimates apply to regions of the study area near roads, where
near
must be
defined. It is a mistake, however, to represent estimates constructed from a
near-road probability sample as applying to regions far from roads as well.
10.3 Monitoring versus Research
In addition to the differences between scientific and nonscientific designs,
there are differences between monitoring and research studies that have an
impact on the spatial sampling designs of both. The designs described later
in this chapter are classified as generally appropriate for either research or
monitoring, and this section defines both types.
The primary defining characteristic of research studies is that they answer
a specific question or estimate a specific parameter. Research studies gener-
ally focus estimation efforts on one or two target variables measured over
a relatively brief period of time (say 1 to 5 years). They are convenient for
graduate students because objectives are usually focused so that the study
