Learn the rules before you break them!

As someone who teaches statistics and evidence-based decision-making, I find that one of the most difficult balancing acts is to emphasize the importance of certain rules in statistical reasoning (e.g., p-values), but at the same time not emphasize them to the point of straitjacketing the students.

It’s the age-old rule of “learn the rules before you break them”. The problem is, different people have different standards for what “learned the rules” mean. It reminds me of the famous Kohlberg’s system on moral development stages. As children, we first learn what’s morally good and bad. Next we learn to apply these rules and apply them strictly and without exception. Finally, we learn to tradeoff between the various rules that may be in conflict.

The system holds up pretty well if we change the words “moral” to “statistical” or “data-driven”. For many people, going from Stage 4 to Stage 5 is frightening, since one is leaving the clear black and white world of rules and order. The most challenging as an instructor are those who think they are at Stage 5 but actually at (say) Stage 2. The only one we need to add is a Stage 0, which corresponds to “Stats? What's stats?”.