Cross-Cultural Overconfidence

In view of results such as those in Fig. 15.1, it is tempting to hazard a generalization that respondents everywhere in Asia tend to be more overconfident than respondents in the West. However, is this generalization justified? Yates et al. (1989) found that their mainland Chinese respondents exhibited substantially greater overconfidence than their American respondents. Whereas, on average, the over/underconfidence index for the Americans was 7.2%, it was considerably higher for the Chinese, 13.4%. The studies by the Wright and Phillips group included Chinese respondents in Hong Kong and Malaysia. The Wright and Phillips results, along with those of Yates et al. (1989), suggest that Chinese cultures might be distinctive in exhibiting high degrees of overconfidence. To evaluate this possibility, Yates, Lee, Levi, and Curley (1990) compared the general-knowledge overconfidence of American respondents to that of respondents in Taiwan, another Chinese culture. Consistent with the earlier comparisons, they found greater overconfidence among the Chinese in Taiwan. A similar result was found in another Taiwan–United States comparison reported by Lee et al. (1995).

The Lee et al. (1995) study actually involved respondents in several other sites besides Taiwan and the United States. Of particular interest is that it included respondents in India, a South Asian culture (actually, myriad subcultures) that differs in many ways from any of the cultures considered in earlier overconfidence studies. Nevertheless, consistent with the earlier findings, the Indian respondents’ judgments were markedly more overconfident than those of the Americans who participated in the same investigation. In a previously unpublished study, we compared the general-knowledge judgments of Koreans and Americans using the same approach as has become standard. When we controlled for question difficulty, we found that the over/underconfidence index was 5.1% for the Korean respondents but only 2.4% for the American participants (p = .027). That is, once again, overconfidence was greater for the Asian group. Whitcomb, Önkal, Curley, and Benson (1995) broadened the picture even further in a comparison of general knowledge judgments by Americans and Turks. The pattern was the same: The overconfidence evident in the Turkish respondents’ assessments was greater than that in the opinions of the Americans. The chapter also describes work by Price and Murphy, then in press, that examined a Latin American sample. In a comparison of Brazilian and American judgments, they found substantially greater overconfidence among the Brazilians. Whereas the over/underconfidence index for the Americans was 5.1%, it was 12.1% for the respondents in Brazil.

The evidence reviewed so far seems consistent with the generalization that overconfidence really is greater in Asian (and, indeed, in Latin-American) than in Western cultures. Other evidence, however, indicates that the story is more complicated. Lee et al. (1995), for instance, found that their Singaporean respondents exhibited essentially the same degree of overconfidence as their American respondents. Particularly intriguing is the case of the Japanese. Yates et al. (1989) found that the overconfidence of the Japanese was essentially the same as that of the Americans and, thus, substantially weaker than that revealed by the Chinese. Since that initial examination of Japanese judgments, there have been several others, most of them indicating that Japanese exhibit even less overconfidence than Americans. For instance, in the Lee et al. (1995) study, whereas the over/underconfidence index of the American respondents was about 7%, it was only about 3% for the Japanese.

Generalization Beyond “Discrete-Fact” General-Knowledge Questions

All the studies discussed to this point have dealt with cases in which respondents stated how sure they were about their knowledge concerning miscellaneous “discrete facts” (e.g., whether copper is denser than iron; whether Mexico City is more populous than Cairo). Understanding people’s judgments about their own knowledge of such facts is nontrivial (despite the resemblance of the research items to the questions used in so-called trivia parlor games). After all, many of the practical decisions that people make in everyday life are predicated in part on what they assume to be true about the world in general. This seems to be especially the case when people are confronted with nonroutine decisions, such as a request for a change in some organizational policy. There is also the potential significance of such judgments for guiding or motivating people’s attempts to learn more about particular topics. Depending on its foundations, we should expect overconfidence to discourage people from seeking to improve their knowledge in circumstances in which that knowledge is sorely deficient. Even given considerations like these, it is apparent that many of the other practical decisions that people face every day rest on judgments about facts or events different from those examined in previous general-knowledge studies. So the following question naturally arises: What is the character of cross-cultural variations in those other kinds of judgments, if such variations exist at all?