an approach to drawing conclusions from statistical samples that is based on the number of times an event is expected to occur in the long run if conditions for observing the event are held constant. It considers any research study to be one of a very large possible number of replications. Also called classical inference. Compare Bayesian inference.