How big is big enough? A Method of Determining Sample Size

Introduction: If you ever have conducted a sample census, the question what is the smallest sample size that you can get away with has entered your mind. At one extreme, you want the sample to be large enough to be representative of the species that are present in the area, but at the other extreme, you don't want to have to spend a huge amount of time collecting data. This exercise focuses on how ecological sample sizes are determined based on a method developed by Henry S. Horn (1993). The taxonomy of the species involved is not important in this exercise. For species that students are not familiar with, descriptive names and samples may be taken for uniformity when pooling data. After data is pooled, it is graphed and an optimal sample size is extrapolated from the graph. Note that the optimal sample size for one habitat will not be generalizable for another habitat.

Materials: Meter stick or measuring wheel, twine or rope to mark your boundary, pencil and paper

Procedure:

1. Given a specific environment - green field, forest, garden, ect., hypothesize what size square plot would be a representative sample for plants found in that area.

2. Share hypotheses with the class. Assign each group a unique area size to sample. You may even want to assign groups sample sizes that differ by powers of two. It depends on how open-ended you want to conduct the activity.

3. Have different groups survey the types and number of species found in their sample size.

4. Construct a data table with the following headings: Plot area, Number of Species, and Cumulative Species.

5. Pool data with other groups and plot Cumulative Species vs. Area on a graph. From the graph you will notice a clustering of data. At some point, you will increase the area, but won’t gain many new species. At this point, the sample size may be determined by extrapolating the area value on the x-axis.

Extensions:

1. To mathematically determine the exact area, you may use a graphing calculator to plot your points. Then using the function for best fit line, have the calculator draw the best-fit line. The formula for this line should have the general equation:

S = cA^z where S represents the total number of species observed, A represents the optimal area and c and z represent constants fitted to the data. (c is similar to the slope of the line

and z > 1)

2. Try circular or linear plots to see if an optimal area can be determined in the same manner.

3. See if your calculated sample size is generalizable for other types of habitats.

Source: Horn, H. S. (1993). Biodiversity in the backyard. Scientific American 268: 150-152.

Back to main page

b
The Woodrow Wilson National Fellowship Foundation CN 5281, Princeton NJ 08543-5281 - Tel:(609)452-7007 - Fax:(609)452-0066 Technical contact: lpt@woodrow.org