An estimator which involves the ratio of two variates, i.e. a ratio whose numerator and denominator are both subject to sampling errors. The term occurs particularly in sample survey theory. If the members of a population each bear the values of two characteristics, x and y, and the total of x, say X, is known for the population, the corresponding total of y, say Y, can be estimated by multiplying X by a sample ratio consisting of the sample total of y divided by the total of x.