Ss played a shuffleboard bowling game in which the variability of their performance was manipulated by E. Two response modes were used to assess the Ss' sensitivity to differences in variability: recall of individual scores and verbal report of "consistency." The data supported the hypotheses that statistical tests are more sensitive to differences in variability than are the responses of Ss and that most of the individual elements in a naturally occurring distribution of variances would appear indistinguishable with respect to variability.