Environmental Engineering Reference
In-Depth Information
bias relative to the investigation. Clear, concise, and definite methodology must be
produced to not only guide one's study but allow future investigators to replicate
and reproduce the original study. During this step, the investigator determines the
study population, limits to inference, available resources (e.g., funding, personnel),
sampling approach, and initiates an evaluation of the literature relative to the
system to be studied. The succeeding step is to state the independent and dependent
variables of interest. Each must be expected to have a measureable response or
linkage to the treatments, disturbance, impacts, developed hypotheses, or other
elements of interest. Again, careful consideration must be made to not attempt to
measure all possible variables but only those that are meaningful and unrelated (i.e.,
not correlated and thus redundant). Pertinent variables can be considered based on
literature, prior experience, and results from a pilot study. In addition, one should
identify pertinent covariates at this stage so that the influence of typically nuisance
or potentially confounding variables can be minimized through appropriate design.
Usually there are a number of potential methods or approaches available for
recording data relative to a specific variable. It is most appropriate to choose a
measurement method guided by the questions and objectives of the study. Other
aspects to consider include techniques used in comparable studies to which col-
lected data will eventually be compared, availability and cost of equipment, type of
data being recorded, precision of the measurements, and identification of any
identified biases relative to proposed methods. Development of clear, structured,
and reliable data recording forms cannot be overstated. Such forms are typically the
foundation for data recording, storage, and transfer. A considerable amount of data
and information loss occurs with the use of poor data forms. Basic to all data forms
are: (1) information to be collected, (2) data collection strategy, (3) order of data
recording, and (4) structure of data recording (Levy and Lemeshow
1991
). All data
must be recorded in a meaningful and legible format that minimizes the probability
of recording error. Order of recording data is important for efficiency of data
collection and subsequent transfer to electronic format (i.e., data bases). Order of
data recording simplifies data collection and minimizes observer effects. The
recording structure includes a condensed explanation for sampling protocol that
can be referenced in the field; use of “check” boxes or other approaches to minimize
mistakes in recording data; and development of variable “keys” to define any short-
hand notation or acronyms that can be used on the data sheet. Observers must be
trained to consistently complete data forms. Furthermore, during the publication
process, a common reason to reject manuscripts is due to poor or inappropriate
methodology; therefore, careful consideration of variables to be measured and how
to measure and record the variables is necessary for a successful investigation.
Use of pilot studies or some sort of preliminary data collection is recommended,
especially for studies where the investigator has little or no experience. As previ-
ously mentioned, a pilot study can be used to estimate the variation within data
recorded for dependent variables. In addition, a pilot study should be used to
develop a suitable sampling protocol allowing all observers to be trained and
become familiar with methodology and recording data. Observer bias can be
recognized and accounted for by using the results of pilot study usually which
