Double counting is a fallacy in which, when counting events or occurrences in probability or in other areas, a solution counts events two or more times, resulting in an erroneous number of events or occurrences which is higher than the true result. This results in the calculated sum of probabilities for all possible outcomes to be higher than 100%, which is impossible.
For example, what is the probability of seeing at least one 5 when throwing a pair of dice? An erroneous argument goes as follows: The first die shows a 5 with probability 1/6; the second die shows a 5 with probability 1/6; therefore the probability of seeing a 5 on at least one of the dice is 1/6 + 1/6 = 1/3 = 12/36. However, the correct answer is 11/36, because the erroneous argument has double-counted the event where both dice show 5s.
In mathematical terms, the previous example calculated the probability of P(A or B) as P(A)+P(B). However, by the inclusion-exclusion principle, P(A or B) = P(A) + P(B) - P(A and B). The principle is used to compensate for double counting by subtracting those objects which were double counted.
Another example is made in the joke where a man explains to his boss why he has to be an hour late to work every day:
All of the numbers are correct, but the man is counting them incorrectly. Sleeping, bathing and eating are part of the weekends, holidays and vacation time that are also being included, making these hours double counted.