Study Design and Logistics - Wetland Techniques - page 26

Haphazard sampling , also known as convenience sampling , is frequently
justified due to cost, time, and logistics, or, on occasion, historical merit. This
sampling approach greatly limits the number of experimental units within a target
population available to be sampled because the strategy employs a protocol that
limits sampling only to a limited number of potential experimental units, which can
have substantial influence on subsequent inference. Results from haphazard sam-
pling must be placed in the context that the data were recorded and are valid only if
the target population is homogeneously distributed (Gilbert 1987 ). Examples of
haphazard sampling would be to only sample wetlands on public land or adjacent to
field stations, conduct roadside surveys of wetlands, rely on volunteer reporting of
flora and fauna outside a defined study, or use an inconsistent temporal sampling
schedule. Similar to pseudoreplication, inference from haphazard studies is limited
to the sampled experimental units. For example, if one only sampled wetlands with
public access, then reliable inference can only be made to similar wetlands with
public access. However, at times, investigators may be interested only in the
wetland represented by haphazard sampling and thus, inference can be considered
valid if the remainder of the study design is appropriate. Haphazard sampling can be
used for initial assessments of an area or hypothesis development (Morrison
et al. 2001 ).
Because of the sheer number and small size of available wetlands, Babbitt
( 2005 ) used haphazard sampling across a microhabitat gradient to relate wetland
size and hydroperiod on the occurrence of amphibians rather than random sam-
pling. She justified the efficiency of the sampling approach by noting that no new
species were found in subsequent sampling efforts. In wetland ecology and man-
agement, frequently one is interested in the effects of impacts to wetlands. Hornung
and Rice ( 2003 ) haphazardly selected grazed wetland treatment locations to evalu-
ate the relationship between the presence of Odonata species and wetland quality in
Alberta, Canada. They also used haphazard methods to sample invertebrates.
Unfortunately, one conclusion from the study was that the haphazard sampling
was insufficient to detect a trend for aquatic macroinvertebrate abundance, diver-
sity, and composition along a gradient of grazing intensity. Due to wind conditions,
Pierce et al. ( 2001 ) haphazardly sampled fixed sampling stations to document the
littoral fish community in Spirit Lake, Iowa. Their results indicate a native species
decline of 25 % during a 70 year period, which was attributed to a decline in littoral
vegetation and other habitat changes. In wetland science, haphazard sampling is
relatively common primarily due to access restrictions precluding random selection
of sampling units. It is incumbent upon the researchers to declare the context of the
sampling effort, which will appropriately restrict inference of results.
Judgment sampling is based on the presumption that prior experience allows
for representative selection of a study area or target population (Gilbert 1987 ).
Deming ( 1990 ) stated that judgment sampling was a type of nonrandom sampling
based on the opinion of an expert. This approach can be considered subjective and
representativeness of results relative to the target population difficult to assess (i.e.,
uncertainty regarding what population is being sampled). As with haphazard
sampling, judgment sampling can be used to assess an area, generate questions