Environmental Engineering Reference
In-Depth Information
B. Collected data are unbiased, but not
precise with the sample mean representing
the population mean with high variability,
which represents an inaccurate sample.
A. Collected data are unbiased and precise,
with the sample mean representing the
population mean with low variability, which
represents an accurate sample
D. Collected data are biased and not precise
where the sample mean is not
representative of the population mean and
high variability is present, which represents
an inaccurate sample.
C. Collected data are biased and precise with
the sample mean not representative of the
population mean but with low variability,
which represents an inaccurate sample.
Fig. 1.2
A graphical representation of the concepts of bias, precision, and accuracy where the
center ring
represents the true mean of the target population and
black dots
represents data
generated by sampling with the intent to estimate the true mean
does not mask the ecological process and data collected through sampling
characterizes the ecological process of interest albeit usually in a much more
simplified manner.
Basic to any study design is the goal of sampling randomly-selected experimen-
tal units to measure variables of interest. This ensures unbiased inference about
some set of population parameters based on a statistic (e.g., mean, variance,
standard error) that describes some attribute of interest. The purpose of sampling
is to estimate the variable of interest and describe its variation in space and time.
When not all members of a target population can be measured (i.e., a census), a
sampling design is used to estimate the population value of the variables; statistical
methods are used to describe the data and make comparisons regarding tendencies of
the data. If one is able to do a census by measuring all subjects of interest (i.e., entire
target population), then statistical tests are not necessary. Sampling can be used to
both select experimental units for study and control of nuisance variables through a
