33 university students made intuitive estimates of the means of 27 lists of two-digit whole numbers. The lists varied independently with respect to length (3, 5 and 7 numbers), standard deviation (approximately 6, 12 and 24) and distribution shape (symmetrical, positively skewed and negatively skewed). Multiple regression analysis showed these three variables taken together to be a fairly efficient set of predictors of absolute estimation error. List standard deviation was the best predictor, with list-length and distribution-shape accounting for much smaller proportions of criterion variance.
