Intuition tells us that small samples lead to fluke results. Let's double thesample size and imagine that the study had revealed 14 errors in one year and 6 the following (instead of 7 and 3). Now regroup your twelvesimulations of ten tosses each into six trials of 20 tosses each. Combine thefirst and second sets, the third and fourth, etc. Then do six more trials of 20tosses each for a total of 120 additional tosses. You should now havetwelve sets of 20 tosses. Did you ever get fourteen (or more) heads in a trialof twenty tosses?