When subjects must perform in both conditions, it is customary to use (14) , reversing the order of the conditions for an equal number of all subjects. The present t test assumes that the population of difference scores is normally distributed. Violations of this assumption will not matter much as long as (15) is sufficiently large. When the researcher wishes to determine whether there is a relationship between two variables, the appropriate measure is the t test for the population correlation coefficient. samole size must be