Who chose the early workshop?
1A community workshop offered an optional early session before a model-building activity. Participants chose whether to attend. At the end of the activity, those who had attended the early session generally produced more accurate models than those who had not. A poster described this pattern as proof that the early session caused better performance.
2The organizer was not ready to make that claim. People with more experience might have been more willing to attend early, and the records did not include their previous experience. The groups had not been assigned at random. The observed difference could therefore reflect the session, differences that existed before it, or a combination of both.
3The organizer suggested a further comparison that would account for prior experience and assign participants to sessions more carefully. This did not show that the early session was useless. It meant that the current records were insufficient to isolate its effect. Reporting the pattern honestly required keeping an observed association separate from an established cause.