1. Said of an estimated parameter if it is sufficiently different from zero, relative to an estimate of its probability distribution, that the probability of the actual parameter being zero is below some small threshold, such as 5%. 2. An estimate that is more than twice, in absolute value, its standard error.