A measure of the trustworthiness of a test procedure. Usually measured by the correlation between the values obtained in apparently identical tests of supposedly identical experimental units. For example, when individuals are presented with two different IQ tests, or with the same test applied at two points in time.