題目： A Cross-National Survey on Generalized Trust and Assurance in the United States and Japan
氏名： Tomoko Mitamura
担当教官： Alan S. Miller
In this research I will present three main issues, which argue that there are biases in the way people respond to the WVS question. Specifically, first, the WVS question involves two different concepts: trust and caution. Second, people might conceptualize “most people” differently. Third, people might conceptualize the meaning of the word “trust” differently. Since this research suggests that one’s social environment might play a role in affecting how people respond to the WVS question, we conduct a detailed analysis of these three respondent biases by comparing: A) The effect of living in different social environments within a country and B) Social environments across societies to examine differences between living in an individualistic and collectivist society. There is a good reason to believe a bias exists. A signal that there might be a problem with the WVS question (dichotomous question) appeared in the 1995-1997 WVS result which claimed Japanese respondents have higher levels of trust than Americans (Dentsu-soken, 1999). However, this result was inconsistent with Yamagishi’s survey results as measured using a 7-point scale (Yamagishi & Yamagishi, 1994). According to Yamagishi (1998), generalized trust plays a much more important role in a society where social uncertainty and opportunity cost is high, such as in the U.S. In such an open environment, it is essential to have high levels of generalized trust, but also believe in the need to be cautious when dealing with other people in general. In other words, choosing caution does not imply being distrustful. In this case, although people living in an open environment have higher levels of trust than people living in a closed environment, since the WVS question is conflating trust and caution, there might be a large number of people living in an open environment who choose caution, and therefore appear to have lower levels of trust than people living in a closed environment. Based on this theoretical background, I will begin to consider first potential biases below.
The first problem involves conflating the concepts of trust and caution. To clarify the discrepancy in results listed above, I hypothesize that people living in a closed environment will score higher than people living in an open environment on trust when measured by the WVS question. However, the order will be reversed when measured as a trust versus distrust variable. In other words, when the choice is between trust and caution, people living in a closed environment such as Japan will appear to have higher levels of trust than people living in an open environment such as America (since Americans are more cautious), however, when “caution” is removed and trust is measured on a scale, Americans will appear to have higher levels of trust than Japanese. These results were confirmed. However, when open and closed environments were compared within each country, the expected differences did not emerge. This is probably due to sampling error, since schools within each country did not differ significantly in the degree to which their environments were open. Thus the remainder of this study will focus on a comparison of Japan and the U.S. and the role that differences in culturally/socially-specific variation might play in the way people respond to the WVS question.
In the second issue, the problem of ingroup out-group bias might be involved. The WVS question is meant to measure a type of default trust level. However, if this measure taps trust in people one knows, it is not measuring what it is supposed to measure. Since people living in a closed environment have a strong distinction between in-group and out-group members, if respondents conceive “most people” as in-group members then they will be more likely to evaluate them favorably, and if they conceive of most people as out-group members, they will evaluate them more negatively. To test this, I asked respondents who they were thinking of when they answered the WVS question and asked them to chose either people they know well or people in general. The results show that nearly half of the Americans were thinking of people they know well, and over 20 % of Japanese were doing the same. Thus, the meaning of this question is not clear to many respondents.
Related to the above differences, the third issue concerns whether whom respondents were thinking of led them to answer the WVS question differently. If it did, then their responses might not indicate a generalized trust since they are referring to people they know rather than people in general. Thus to test whether respondents who conceived of “people they know” had the same levels of trust as those who conceived of “people in general,” we hypothesize that people living in a closed environment who conceptualize “most people” as in-group members will be much more trusting than those who conceptualize them as out-group members. However, people living in an open environment will show similar levels of trust regardless of whether they conceptualize “most people” as in-group or out-group members. The result shows that whom Japanese respondents were thinking of greatly affected how they responded to the WVS question. Of those who were thinking of people they know well, nearly 80% chose trust, as opposed to only 55% of people who were thinking of people in general. For Americans, however, whom they were thinking of had no significant affect on how they responded to the WVS question.