Why use a Wilson interval?
A raw percentage hides how much evidence produced it. The Wilson score interval remains useful near 0% and 100% and with smaller samples, where a simple symmetric approximation can behave badly.
Read overlap carefully
If the interval overlaps your target band, the sample is compatible with values inside and outside that band. If it sits wholly outside, that is a stronger review signal—but still not proof that one mechanic caused the result.
Keep comparable games together
Do not combine different rules versions, player counts, factions, or teaching conditions just to increase the game count. Use the Seat Advantage Analyzer or Faction Matchup Analyzer when those conditions define the question.