Zendo is often described as "Science: The Game." (More discussion here)

Lots of biases come up. You quickly learn to avoid positive bias if you play this often. You start to deal with confirmation bias and illusory correlation and neglect of sample size. Almost any bias that affects hypothesis generation and testing affects how well you play Zendo, and you can run through single rounds in as little as 10 or 15 minutes. I cannot recommend it enough.

If you're serious about using it for didactic purposes, have players work together, collaborating aloud. This way, you can cover some of the social biases, and have a clearer record of what people were thinking and when they thought it. (If you're really serious, record the play session, and show insight-generating clips as you go. When you play as the master, you get to see these all the time.)