Magnitude

Most studies rely on statistical tests of theoretical propositions. The authors generally report whether or not findings are statistically significant, but the reader is not given a sense of the magnitude of the effect. We believe it would be useful for authors in many areas to attempt to make estimations of easily compared and understood proportions and probabilities for outcomes, using meaningful metrics for independent variables so readers can better grasp the impacts being discussed. Rather than simply reporting that abuse has a significant indirect effect on violence through its effect on academic achievement, for example, we urge authors to report the proportion of abuse victims (of each type) who commit violence, the proportion who have academic problems, and compare the proportion of those who commit violence across categories of abuse and achievement. If many studies employ this approach, we can begin to get a sense of the size of the impact and the “dose-response” of adverse outcomes, and emphasize those with the largest magnitude for policy.

 
Source
< Prev   CONTENTS   Source   Next >