Top view of a variety of different coffees in cups on a wooden table

“If you torture data enough, it’ll confess to anything” my manager always used to say. She was on point.

With enough slicing, dicing, and tweaking, you can make data say whatever you want. Once more for those in the back – even true data can be made to lie. Want to prove coffee causes success? Find a tiny study where CEOs drink coffee. Want to “prove” it’s bad? Find one where people who drink six cups a day have heart issues.

This is called p-hacking, and it’s why results should always be questioned. Were different angles tested? What’s the sample size? Were results handpicked to confirm the narrative?

Good analysis isn’t about getting the answer you want, it’s about getting the answer that’s real.