The concepts associated with the identification problem are related directly to the probability distribution generating the data.This approach suggests simple examples that can be used for illustrative purposes and avoids the need to introduce simultaneous-equation models before discussing identification.