The Retraction Watch web site shared an article about null hypothesis testing — “Almost 40% of peer-reviewed dietary research turns out to be wrong” — which illustrates why direct, experimental studies on humans (like our study) can yield more accurate and reproducible data than statistical methods.
This is not to say statistical methods are wrong to conduct. After all, our companion web site has noted that epidemiology has saved millions of lives through statistical associations.
However, we also emphasized that correlation is not causation. And to obtain solid, credible evidence, statistical correlations must be tested (whenever possible) by direct human testing.