Experimental replications in animal trials

The recent discussion on the reproducibility of scientific results is particularly relevant for preclinical research with animal models. Within certain areas of preclinical research, there exists the tradition of repeating an experiment at least twice to demonstrate replicability. If the results of the first two experiments do not agree, then the experiment might be repeated a third time. Sometimes data of one representative experiment are shown; sometimes data from different experiments are pooled. However, there are hardly any guidelines about how to plan for such an experimental design or how to report the results obtained. This article provides a thorough statistical analysis of pre-planned experimental replications as they are currently often applied in practice and gives some recommendations about how to improve on study design and statistical analysis.