THE ACCOUNT TAKEN OF STATISTICAL POWER IN RESEARCH PUBLISHED IN THE BRITISH-JOURNAL-OF-PSYCHOLOGY

Since approximately 1925, researchers in psychology have evaluated the ir hypotheses against the probability of making a Type I error. Attemp ts to persuade researchers to augment this information, with details s uch as effect size and confidence intervals, and in particular to take the probability of a Type II error into account, have largely been ig nored. The present paper reviews the degree to which statistical power has been explicitly considered in papers published in the British Jou rnal of Psychology in 1993 and 1994. It summarizes the power for small , medium and large effect sizes for 54 papers, based on the sample siz es which were employed. The analysis confirms that the power of statis tical tests is not taken into account by researchers and that, accordi ngly, they continue to run a high risk of rejecting their research hyp otheses.