There are two types of car in a company car fleet. Car type A (say there are 10 cars) and car type B (say there are also 10 cars). If a car from car type A breaks down, does this make it less likely that a car in car type B would break down in the same week? Another way of putting it is ... say there are a group of cars which we haven't differentiated (which are generally reliable) and 3 have broken down in one week (very rare), I assume it would be quite unlikely that another one would break down in the same week (i.e. the probability decreases each time - am I correct in this assumption?). But if those 3 cars are all car type A, and the fourth car we are observing is car type B, would this car be any more likely to break down or not, considering the back 'luck' of the past week, and considering we have now differentiated it to another group of cars?