Designing a Simulation

Some probability questions are hard to solve with a formula but easy to act out. If you want the chance of a family of 4 kids having exactly 3 girls, you could do tricky counting with the binomial setup, or you could just flip a coin 4 times, call heads a girl, and repeat that over and over. The fraction of repetitions that give exactly 3 girls is your estimate. Simulation trades algebra for repetition, and it works because probability is, at its core, long-run relative frequency.

Designing one has four parts. First, assign a random device to match the real situation: a coin for a 50/50 event like boy/girl, a digit from a random number table for unequal chances, a die for 1-in-6 outcomes. Second, define exactly what one trial looks like: for the family question, one trial is 4 coin flips. Third, state clearly what counts as a success, such as exactly 3 heads in those 4 flips. Fourth, repeat the trial many times, independently, and record the results. The estimated probability is successes divided by total trials run.

This topic is the bridge between the two-way tables and basic probability rules you just learned and the binomial distribution coming next. Simulation gives you an approximate answer fast, without needing a formula, and it builds the intuition for why the binomial formula works the way it does. More trials make the estimate more reliable, which previews the law of large numbers and the idea of sampling variability that runs through the rest of the unit.