A situation in which I would test a directional hypothesis would be a drug company’s new antidepressant. The directional hypothesis I would test is that the drug helps those dealing with symptoms of depression in a positive way, either by reducing or eliminating depressive moods, thoughts, and/or behaviors. The independent variables would be whether or not a participant in the study received the drug or the placebo. The dependent variables would be whether or not the drug affects the participants moods, thoughts, and/or behaviors. This hypothesis is directional because it states that the drug positively affects the participants moods, thoughts, and/or behaviors; therefore, I am operating under the assumption that the drug either does these things or simply does not do them.
Using the same situation, I could create a nondirectional hypothesis: the new antidepressant will have an effect on the participants moods, thoughts, and/or behaviors. The independent and dependent variables remain the same; however, my hypothesis is now a nondirectional one. It is nondirectional because it states the drug will have some effect, but does not say what that effect will be, thus I am not stating which direction I expect the results to go.
The more appropriate test, in my opinion, is a one-tailed directional test. This is because I specifically want to know whether or not this drug is going to positively effect the participants.
It is important, in regards to a two-tailed test, to alter the alpha level to create a rejection region of 5%, instead of having a total of 10%. As stated in our lectures, in a two-tailed test, we would “reject the mean for a total of 10% of the scores and not just 5%.” In addition, the tails on a two-tailed test are commonly smaller than a one-tailed test, thus in order to reject the null, the “test statistic needs to be a more extreme score.”