The extent to which a replication of a measurement process will yield the same results. In forecasting, the extent to which a method will produce similar forecasting accuracy when used in similar situations. Tests of statistical significance do not provide good measures of reliability.