The extent to which an experiment shows convincingly that that changes in the dependent variable are a direct result of manipulation of the independent variable.