Developing A Formal Hypothesis

In the planning stage of research, the researcher should generate a formal hypothesis or set of hypotheses based on previous research and/or theories that have yet to be formally tested. The hypothesis is usually a statement or prediction that the researcher seeks to evaluate with a study. For example, suppose a researcher wants to test the hypothesis that the average lifespan of mouse line A is longer than the average lifespan of mouse line B. Implicit in this hypothesis are two distinct possibilities represented individually by the null hypothesis, denoted H0, and the alternative hypothesis, denoted Ha. The null and alternative hypotheses can be expressed as follows:

H0: there is no difference in average lifespan between mouse line A and mouse line B (^a = MB); Ha: the average lifespan of mouse line A is longer than mouse line B (^A > MB)-

This is an example of a one-sided hypothesis (i.e., the researcher expects mouse line B to live longer, on average.)

If the researcher has no particular reason to believe that the average lifespan of mouse line A will be longer than the average lifespan of mouse line B, or vice versa, he or she will need to test a two-sided hypothesis. The null and alternative hypotheses would then be as follows:

H0: there is no difference in average lifespan between line A and line B (^A = MB); Ha: there is a difference in average lifespan between line A and line B (^A = MB)-

Depending on the form of the hypotheses, the researcher uses either a two-sided or one-sided statistical test to determine the statistical significance of any observed difference in average lifespan before either accepting H0 or rejecting H0 in favor of Ha. Most statistical analysis software will calculate both one-sided and two-sided tests at the same time, and the researcher simply needs to pick the proper one based on the nature of their hypothesis.