Reproducible Research from Noisy Data: Thoughts from Agricultural Statistics

Reproducible results define the very core of scientific integrity in modern research. Yet, legitimate concerns have been raised about the reproducibility of research findings, with important implications for the advancement of science and for public support. With statistical practice increasingly becoming an essential component of research efforts across the sciences, this talk highlights the compelling role of statistics in ensuring that research findings in the agricultural sciences are reproducible. We revisit foundational statistical concepts relevant to reproducible research in the context of the agricultural sciences, raise awareness on common statistical misuse undermining it, and outline recommendations for statistical practice.