How do you determine directional or non-directional alternate hypotheses? - Answers

Whether you frame your alternative hypothesis, Ha, as one-sided (directional) or two-sided (non-directional) is really up to you, but should be decided before you look at the data. It will affect the calculation of your p-value and ultimately your conclusions from the test. In most cases there will be a sound, obvious reason for choosing one or the other.For example, if you were testing the effectiveness of a new anti-cholesterol drug you'd probably only be interested in testing whether the average of the experimental group was lower than the control group. So Ha is directional, or one sided. If on the other hand you were testing, for example, whether a Group A performed better on a test than Group B, your Ha would be that the average of Group A does not equal Group B. That is, you're not sure, before you run the test, whether Group A should perform better or worse than Group B. So your test is non-directional, or two-sided.