Tolerance for errors in intuitive mean estimations

Abstract

Two groups of subjects examined 20 lists of 15 numbers with either low (SD = 7.0) or high (SD = 13.0) variance. Group 1 first estimated the mean of each list and then, for each mean, specified an interval of acceptable error for intuitive estimations of that mean. Group 2 specified acceptable error intervals for the same lists without prior estimation of the means. The estimates of Group 1 were significantly less accurate for the high variance lists than for the low variance lists. Correspondingly, both Group 1 and Group 2 specified significantly wider intervals of acceptable error for the high variance lists than for the low variance lists. It was concluded that subjects are responsive to task contingencies that affect accuracy, showing more tolerance of error for task components of greater difficulty.