Email Bookmark and Share


Often in Error, Rarely in Doubt

Saturday, May 23, 2009, 12:00 PM - 12:55 PM
Yerba Buena 14 - 15

Chair: Don A. Moore
Carnegie Mellon University

Excessive confidence in the precision of one’s knowledge is both the most robust and the least understood form of overconfidence. This symposium investigates its ultimate causes. The evidence suggests that overprecision is caused by limitations on the working capacity in human memory, conversational norms, and social pressure.

Overconfidence and the Representation of Uncertainty
Jack B. Soll
Duke University
We compare uncertainty about repeatable cases (e.g., a random person’s weight) with uncertainty about specific cases (e.g., Rush Limbaugh’s weight). We find that perceptions of repeatable cases are noisy but accurate. In contrast, for specific cases people underestimate the variability around one’s best guess—they are overconfident in their answers.

Overconfidence in Intuitive Confidence Intervals: Effects of Assessment Format, Short-Term Memory Capacity, and Aging
Peter Juslin
Uppsala Unversity, Sweden
The naive sampling model suggest that one cause of overprecision is that, although people accurately describe the small samples of similar observations they activate within short-term memory, they are naive about more sophisticated statistical relations between these small samples and the properties they are assessing.

The Role of Conversational Pragmatics in Reporting Interval Estimates
Craig R. M. McKenzie
University of California, San Diego
When people report high (e.g., 90 percent) subjective confidence intervals, the intervals are often too narrow. We report evidence indicating that different interval elicitation methods can affect interval width not for “cognitive” reasons, but because the different methodologies signal whether accuracy or informativeness is relatively important.

Michael Liersch, New York University

Competing To Be Certain (But Wrong): Social Pressure and Overprecision in Judgment
Don A. Moore
Carnegie Mellon University and University of California, Berkeley
We find that the competitive social pressure of a market contributes to overprecision among those competing for influence. When advisors must compete with each other, we find that the confidence they claim escalates over time. This overprecision helps advisors’ sell their advice.

Joseph R. Radzevick, Carnegie Mellon University

Go back

How Do We Measure Pleasure? - Lindo Bartoshuk