Uncertainty from internal and external sources: A clear case of overconfidence.

Abstract

If, as earlier work suggests, people tend to be especially overconfident of uncertain outcomes which depend upon their own performance, they should show a preference for situations in which the total uncertainty is more internal (I) or skill-based than external (E) or environment-b ased. This hypothesis was tested by manipulating I-E composition of objectively equiprobable choice alternatives in a paired-comparisons paradigm. Each alternative included a specified criterion on a dart-throwing task (I component) and a specified zone on a probability spinner. Favorable outcomes on both components of the chosen alternative paid \i; failure on either or both paid Oji. Fifteen 5s, trained to a stable dart-throwing level under one of three different set conditions, made a series of choices at three overall uncertainty levels. Neither variable modified the chief result, which was a clear-cut verification of the hypothesis (87% of all choices were in the predicted direction). Scaled preference values for the various I-E mixtures were generally as expected, except that psychological distances were disproportionately large under the highest level of overall certainty. In two earlier papers, (Howell, 1967, 1970) the writer has argued that man's perception of uncertainty, and therefore his probabilistic decision behavior, is very much dependent upon the source of that uncertainty. Evidence was presented suggesting that people tend to be overconfident in the outcome of events over which they have some control, i.e., where uncertainty is largely internal or skill derived. On the other hand, the literature on external or environment-derived uncertainty, which encompasses the vast majority of subjective probability and decision research to date, reveals frequent instances of underconfidence especially at higher objective probability levels (Peterson & Beach, 1967).

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 60 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.