Notes that studies in psychology often have low power because of inadequate sample size and that prevalent psychometric distributions are sometimes radically nonnormal. Demonstrates robustness of independent samples t-test with respect to Type I error. Shows that researchers may use power tables based on population normality without modification…

Demonstrates low probability that nonequivalence will produce erroneous inferences in small samples. Sees probability of erroneous inference in absence of true treatment effect as generally no greater than nominal Type I error rate. Seems unlikely that small samples have biased inferences drawn from past psychotherapy outcome research. Cites other…

Demonstrates that magnitude-of-effects (ME) estimates vary in susceptibility to sample-size bias depending on whether they are directional or nondirectional estimates. Also demonstrates that study characteristics that influence size of ME estimates can be explicitly taken into account when comparing studies. Emphasizes need to consider study…