Robust preclinical replications through uncertainty informed sample size planning

In recent years, preclinical replication studies have been conducted to assess reproducibility rates of preclinical research. A crucial aspect of such studies are sample size choices. In replication studies, only sufficiently high animal numbers will result in reliable conclusions. This is offset by ethical constraints inherent to animal experimentation that demand animal numbers to be at a minimum. To develop guidance on how to balance the diametrically opposed goals, we analyzed sample size choices in three large-scale preclinical replication projects that assessed preclinical reproducibility. We identified conceptually different choices regarding replication sample sizes across projects. To explore the consequences of these choices, we conducted simulations based on data from the three preclinical replication projects. Here, the declaration of replication success depended strongly on different approaches to sample size planning. Through incorporating uncertainty of exploratory experiments into sample size planning, reliability of replication studies increased markedly. We discuss how sample size planning informed by the uncertainty of exploratory effect sizes will benefit testing specific hypotheses through replications in preclinical research settings.