Year: 1992
Type: article
Abstract: There seems to be a controversy in the credibility given to the results of epidemiologic studies.On the one hand, some feel that more credibility should be given to the results from the 'basic' (laboratory) sciences than to those of epidemiologic studies.On the other hand, others have claimed that epidemiology offers the ultimate proof of a proposed mechanism or therapy.In our view, there is only one body of empirical science; therefore, a hierarchy of fields of sciences is not only unjustified, but also counterproductive, since it may lead to wrong conclusions, based on only part of the available empirical evidence.The Information gathered in empirical studies may be false, because of bias, measurement error or chance.Therefore, the results from several studies may be contradictory.This forces us to weigh the results of different studies.Generally, there are two approaches in judging whether an association, äs found in a study, represents a true causal relation.One might apply criteria, for instance of plausibility and coherence, that might be helpful in considering whether an association is causal.Nevertheless, the judgement will always remain subjective to some extent.One might apply an approach, that incorporates this subjectivity.In this Bayesian view, scientific studies are seen äs procedures that modify our prior belief in certain associations into a posterior belief.With this view plausibility and coherence, äs well äs the unavoidable subjectivity are incorporated in the prior belief, whereas the posterior belief still allows for uncertainty.
Author: F.R. Rosendaal
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W56346225