I once met a PhD in statistics. He said ‘give me the option to chose a method and a sample, and I can prove anything I want.’