Statistical power is influenced by the "direction" of the hypothesis. Remember that we can pose a directional or non-directional hypothesis. In our example, we have been posing a directional hypothesis; we think that college students today have less face-to-face social interaction than previous groups of college students. This is directional, because we have specified that they will have less face-to-face interaction. If we thought that students might have more or less face-to-face interaction, we could test this hypothesis with a non-directional test. We would hypothesize that they were different, rather had less face-to-face interaction. Directional tests are known as "one-tailed" tests because all of the error is is one "tail" of the distribution (less than). Non-directional tests are called "two-tailed" tests because we must include the possibility that the alternative population could be less than m or greater than m.
Directional or "one-tailed" tests are more powerful than non-directional or "two-tailed" tests. Let's see why.
Here is our statistical power graph. Let's conduct a two-tailed test and see what happens. The alpha-level, sample size, and effect size will remain the same.