a non-directional hypothesis

 Since obtaining statistical significance is easier to obtain with a directional hypothesis (one-tailed test) than with a non-directional hypothesis (two-tailed test), why would anyone ever design a study with a nondirectional hypothe