How much confidence do we need in animal experiments? Statistical assumptions in sample size estimation

ABSTRACT

Statistical sample size calculation is a crucial part of planning nonhuman animal experiments in basic medical research. The 3R principle intends to reduce the number of animals to a sufficient minimum. When planning experiments, one may consider the impact of less rigorous assumptions during sample size determination as it might result in a considerable reduction in the number of required animals. Sample size calculations conducted for 111 biometrical reports were repeated. The original effect size assumptions remained unchanged, but the basic properties (type 1 error 5%, two-sided hypothesis, 80% power) were varied. The analyses showed that a less rigorous assumption on the type 1 error level (one-sided 5% instead of two-sided 5%) was associated with a savings potential of 14% regarding the original number of required animals. Animal experiments are predominantly exploratory studies. In light of the demonstrated potential reduction in the numbers of required animals, researchers should discuss whether less rigorous assumptions during the process of sample size calculation may be reasonable for the purpose of optimizing the number of animals in experiments according to the 3R principle.

KEYWORDS:

• Exploratory statistics
• hypothesis generation
• power
• type 1 error

Acknowledgments

The authors would like to thank the reviewers for their valuable comments, which really improved the manuscript, as well as George Taurman for proofreading.

Funding

This work was partially funded by the German Federal Ministry of Education and Research, grant number 031A573.

Additional information

Funding

German Federal Ministry of Education and Research