Intuitive statistical inferences about variances

Subjects saw samples from each of two populations of numbers and made intuitive inferences about which population had the larger variance. Then they either estimated the ratios of the variances or stated their confidence (subjective probability) in their inferences. The ratios were used to infer the subjective magnitudes of the sample variances; they were systematically inaccurate because of a tendency to underweight deviant sample data and because the subjects regard variance among large numbers as less variable than variance among small mmlbers. Then, confidence in the inferences about the population variances was compared to the probabilities that would have resulted if the ratios of sample variances had actually been the ratios that, the subjects reported. Confidence was systematically related to these probabilities but it was always lower. The results are discussed in terms of the conservatism findings reported in other investigations of intuitive statistics. A Bayesian F4est is appended. Information about the factors that govern occurrences of future events is a fundamental prerequisite for optimal decision making. When the domain of interest can be described in statistical terms, the required information consists of the parameters of the populations from which future samples will derive. In most decision experiments, particularly those that involve gambling, the subject is told the parameters of the relevant popu