Epidemiologic studies have been the cornerstone of the development of public health research programs and continue to make important scientific contributions to public health understanding and policy. Unfortunately, the nature of these programs often precludes the random allocation of exposure.P-values have a rich history of use in observational studies, and although, at one time they were embraced, p-values have fallen into some disrepute. The difficulty with p-values in these important, nonrandomized efforts is that they have erroneously but commonly been interpreted as binding up the truth of the trial—sample size, effect size, sample variability and effect attribution—into one number. Workers who interpret p-values in any research study, imbuing them with interpretative powers beyond an assessment of sampling variability, unfortunately do so at their own risk. The situation is more acute in nonrandomized studies due to the absence of the random allocation of exposure in the sample chosen from the population. P-values can play a useful role in the causality debate if the researchers are clear that p-values were not designed to measure sample size, effect size or effect attribution, only sampling error.
Weitere Kapitel dieses Buchs durch Wischen aufrufen
Bitte loggen Sie sich ein, um Zugang zu diesem Inhalt zu erhalten
Sie möchten Zugang zu diesem Inhalt erhalten? Dann informieren Sie sich jetzt über unsere Produkte:
- Mistaken Identity: P-values in Epidemiology
Lemuel A. Moyé
- Springer New York
Neuer Inhalt/© Stellmach, Neuer Inhalt/© Maturus, Pluta Logo/© Pluta, Avaloq/© Avaloq Evolution AG