Abstract
Importance: Reproducibility, or the long-term validity of findings, is a precondition for the credibility of scientific results in several scientific disciplines. If different experts were asked how much published work in their field is reproducible, more than fifty percent of researchers in chemistry, physics, biology, medicine, and others said we have a replication crisis. This means that the scientific credibility of many disciplines in the eyes of the public is at risk, with significant consequences for the reputation and funding of science. Challenges: It is therefore necessary to tackle the causes of the replication crisis, such as Questionable Research Practices (QRP), publication pressure, and weaknesses in the planning and statistical analysis of studies. The latter is the subject of this article, in which it is emphasised that many hypotheses do not correspond in their complexity to the phenomena studied, either in terms of the possible influencing variables or in terms of the measures of association. Measures: It is suggested that the hypotheses should be more differentiated, take greater account of the presumed effect structure, and the variety of logical relationships in the empirical phenomena. This article uses several examples to show the extent to which more precise hypotheses have an impact on the accuracy of statistically reliable results. One computer program that can be used in the next time for these purposes is Relation Analysis (RELAN), which allows logical analyses, statistical tests, explorations and simulations of relations between variables. Conclusion: In future, it will be necessary to adapt scientific hypotheses in the biological, human and social sciences more closely to the complexity and the structure of empirical phenomena.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call