Stick number labels on random outcomes, and use a distribution table to fully capture how likely each value is