Suppose a random sample of n items have been administered to a random sample of m examinees. Small-sample formulas are given for estimating the mean and variance that would have been found for the number-right scores, These formulas are useful when it is impossible to administer all items to all examinees.