Researchers have found more than half of the public datasets provided with scientific papers are incomplete, which prevents reproducibility tests and follow-up studies. However, slight improvements to research practices could make a big difference. ...