A statistical axiom that states that the larger the number of exposure units independently exposed to loss, the greater the probability that actual loss experience will equal expected loss experience.
In other words: the credibility of data increases with the size of the data pool under consideration.