Theoretical vs. experimental probability