Pre- and post-data are collected and analyzed to examine the effect of interventions or programs on processes (e.g. IPM), sites (e.g. fields of crops), or subjects (e.g. freshwater ducks or program participants). Pre- and post-data can represent relatively continuous data (height of plants to the millimeter), interval data (# of trees dying; frequency of a behavior based on a 5-point Likert scale of frequencies), or categorical data (favorite choice for a career). The analyses you perform on your data will depend upon the type of data you collect and the questions you wish to have answered from the analysis. For this reason, data collection tools and processes are best set up with analysis already practiced. Conducting a “test-run” of data entry and analysis reduces the likelihood of encountering unwanted surprises or wasted data once data collection and analyses begin.
You will want to review the kinds of analyses used for different data types before determining your best options. Because most Extension work involves humans and other living organisms, a great resource is Zar’s Biostatistical Analysis (1).
As an example of a common analysis for pre- and post- data when you want to know if participants have changed behavior as a result of a program intervention is to use a pre-survey of behaviors participants identify or rate before the program begins and then compare it with results using the same survey and same group of participants at the end or after the program intervention.
If you have a large, representative sample, you may want to run a paired sample student’s t-test (2). To do this, you enter data as “matched pairs” of pre- and post-scores for each individual. If you cannot match the tests, you should run an independent sample t-test. The database is set up differently for these two types of tests, so refer to the user manual for your statistical package before entering data. The results of a t-test will tell you if the difference between the pre- and post-test is significant. Educators typically seek results with significance levels less than .05.
If your group of participants is small and/ or does not necessarily represent the population you are targeting for your intervention, you may just want to examine and compare the frequencies and mean scores of the pre and post data without using statistical tests. Changes in mean scores will tell you if participants’ knowledge has increased or decreased for the whole group, though statistical significance cannot be ascertained without using statistical comparison methods.
(1) Zar, J.H. 2009. Biostatistical Analysis, 5th Edition. Prentice-Hall, N.J.
(2) Fisher Box, Joan (1987). “Guinness, Gosset, Fisher, and Small Samples”. Statistical Science 2 (1): 45–52. DOI:10.1214/ss/1177013437
