The effect of characteristics of the data distribution on subjective probability judgments.

Accuracy in making probability judgments about the characteristics of data distributions was examined in this study. In particular, the effects of four characteristics on probability estimation: problem type, variability of the data distribution, amount of information, and anchor point were investigated. Previous research has examined the impact of single factors on single measures of subjective probability (usually the mean), but has not studied the interaction of factors or used more than one dependent measure. A questionnaire containing two problem scenarios was given to 747 introductory psychology students. One scenario dealt with the amount of rainfall in cm. per year in Buenos Aires, Argentina, and the other scenario was concerned with the average number of points scored per game in one season by a pro basketball team. Both problem scenarios were fictitious but appeared genuine to the subjects. A list of numbers representing data from preceding years followed each problem scenario. Each list was derived from computer generated random normal distributions and had either low or high variability, five or 25 years of data, and a last data point of 90, 100, or 110 units. Subjects were asked to generate a three category subjective probability distribution for the present year for each scenario. They also indicated their level of confidence in these probability judgments on a six point scale. Subjects also made point estimates of the mean from memory for each scenario. Four dependent measures resulted: the mean and standard deviation of the subjective probability distribution, the confidence judgment, and the point estimate of the mean from memory. The major findings of the study were: (1) The mean of the subjective probability distribution was affected by the interaction of anchor point and amount of information. The mean estimate was biased in the direction of the anchor point in the low amount of information condition but not in the high amount of information condition. (2) The standard deviation of the subjective probability distribution was affected by problem variability. Standard deviations were larger for the high variability condition than for the low variability condition. Also, females\u27 judgments were more variable than males\u27 judgments. (3) Confidence judgments were lower for the high variability condition than for the low variability condition. Confidence judgments were also affected by gender and by a gender by problem type interaction. Overall, males were more confident about their judgments than females. Further, males were more confident about the basketball problem than about the rainfall problem while females were equally confident about both problems.Dept. of Psychology. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis1990 .K545. Source: Dissertation Abstracts International, Volume: 52-11, Section: B, page: 6114. Thesis (Ph.D.)--University of Windsor (Canada), 1989