Readings & Instructions
- Read the following on-line information to
gain a basic understanding of Random, Systematic, Selected (or
Subjective), and Stratified Sampling:
Allocation of Sample Units - Sampling Designs
- Read Chapter 7 (pages 97-141) "Sampling
Monitoring and Measuring Plant Populations.
What this Reading Covers
Once a researcher or manager has decided "what" to sample, then "where" to
sample must be determined. Three basic sampling schemes exist:
Subjective (or Selected), Systematic, Random, and Stratified.
Subjective - Selectively place a set of plots or plants (a sample) in
specific areas that meet research interests or management objectives.
Select sites that are considered
representative of change or responsive to management.
May or may not reflect the larger area. How
well the sampel reflects the larger area depends on judgment of person selecting sites.
Difficult or impossible to make statistical
inferences about a whole pasture, park, watershed, or management unit.
However, use of subjective sampling may be very effective in determining if
management strategies are working to meet management goals.
Monitoring key areas or critical
areas is a type of subjective sampling.
Key Areas are a portion of
land which, because of its location, grazing or browsing value,
or topography, serves as an indicator of land conditions, trend,
or degree of seasonal use by animals. These Key Areas are
considered indicators of what is happening on a larger area as a
result of on-the-ground management actions.
Areas are units that
contain unique or special values such as:
- fragile watersheds
- sage grouse nesting grounds
- riparian areas
- habitats with rare plants
Systematic - Placing plots or selecting plants for research or
monitoring by systematically and regularly spacing plots according to a
Rapid and easy to use in the field.
Assures good distribution and uniform coverage
of the target population.
Limitations in statistical analysis because
units in the sample are not independent of one another. For example, if
plots are placed every intersection in a ½-mile grid, once the first sample
is selected, the location of all other plots in the sample is known. In
other words, the location of all plots is dependent on the location of the
first plot of the placement of grid.
Random - the location of plots or areas to be studied are selected in
a way that each sampling unit (i.e., plot or plant) is selected completely
at random and any potential area or plot to be studied has an equal chance
of being selected.
All observations in the sample are independent
so normal (i.e., parametric) statistics can be applied with confidence.
The best way to select a random sample is
generally to apply a grid (with two coordinates) to the total area of
* A pair of random numbers can then be selected for the x- and y-coordinates
on the grid.
* One could also number all the intersections on the grid and select a
random number from 1 to the total number of intersections.
* Random numbers can be selected in several ways:
▫ roll dice
▫ put all units or
coordinates in a hat and draw a number
▫ use a random numbers
table in a statistics book
create a list of random
numbers with the random feature in a spreadsheet (like excel)
One problem with random sampling is that it
may result in poor distribution of sampling units as the units that are
randomly selected may not be evenly distributed across the landscape or
target population. Plant populations and habitats are rarely distributed
evenly and randomly across the landscape, therefore a sample of random units
may or may not represent the landscape.
Stratification - If there are aspects of the landscape that will
clearly result in differences among the plots sampled it is often good to
stratify the area and sample within these sub-units. For example, if a
pasture has 3 major ecological sites that vary in biomass productivity, the
proportion (%) of samples examined in each ecological site could be based on
the proportion of the total area occupied by each site.
Stratification often overcomes the problem of
poor sample distribution.
At least 2 sample units (preferably 3 or more)
must be drawn from each sub-unit to determine the variation within each
Can yield information about variation among
and within sub-units.
The samples within each sub-unit can be
applied in a random fashion to create a Stratified
Random sample, or systematically to create Stratified
Systematic sample, or subjectively to create a Stratified
Selecting Study Areas
locations of study sites is crucial to success of an inventory or monitoring
program. Selection should be documented. In other words, explains
somewhere in your survey notes, why and how you selected the sites for evaluation. The site selection should clearly
reflect the management or monitoring objectives. Criteria used for
selecting sites are generally based on:
seral state of
water, fences, & natural boundaries
areas of animal
statistical comparisons or interpretations intended
Remember: Study sites must always be clearly mapped and documented.
- When should random sampling be used over
systematic sampling schemes?
- Why are key areas used so frequently in
management and seldom used in scientific investigations?
- In assignment 1, you were asked to find an
example of a sampling protocol. In the example you found, who
were samples placed on the landscape? Random, Subjective,
Stratified, or something else.