A Monte Carlo program produces a very small statistical standard error, but its simulated collision law omits an interaction known to occur in the material. What is the strongest conclusion?