Is two-tailed testing for directional research hypotheses tests legitimate?