On the reproducibility of scientific research