I had a very embarrassing experience, when I presented my early draft paper on “Inequality and Corruption as Correlates of Social Trust��? at a Work-in-Progress Seminar at the Kennedy School of Government last fall. Professor Edward Glaeser came to my talk, but I was not aware of him although I had read his articles including one about “measuring trust." He asked a question about measurement of social trust without identifying himself. Since I had already talked about the problem of measurement (apparently he did not hear that because he was late) and was about to present my results, I did not want to spend much time about the measurement issue. He was not satisfied with my brief answer and repeated his questions and comments, saying that the typical trust question in surveys, “Do you agree that most people can be trusted or you can’t be too careful?��?, may reflect trustworthiness rather than trust “according to a study.��? Because I assumed that trust and trustworthiness reinforce one other, I did not think that was a great problem.
Our encounter was an unhappy one for us both. Probably he had an impression that I did not respect him and did not give adequate attention and appreciation to his questions and comments, and I was also kind of annoyed by his repeated intervention. One thing that made the things even worse was that I am not a native English speaker; I have particular difficulty with husky voices like his, a difficulty made the interaction even more problematic. After the talk, I asked him to give the reference for the study on measurement of trust he mentioned. He wrote down Glaeser et al. (2000), and I realized that I had read the article he cited. Even then, I was unaware who he was. I asked a participant of the seminar who he was, and to my surprise, he was Edward Glaeser, the lead author of the article on measuring trust. If I had recognized him, I would have paid much more attention to his questions and comments and tried to answer them better. How big a mistake I made!
Although I still think that the typical trust question captures both trust and trustworthiness, Glaeser et al.’s experimental results may indicate the trust question needs to be designed better. One thing to note in this regard is that caution is not the opposite of trust, as Yamagishi et al. (1999) argued. In my case study of social trust in Korea, I found that inclusion and exclusion of “being careful��? option in trust questions produced substantially different results. More respondents agree that most people can be trusted when they were simply asked, “Do you think most people can be trusted��? than when they were given the two options “trusting most people��? and “being careful.��? Average percentage of trusting people was 42.9 per cent for the former type of questions, and 32.2 per cent for the latter type of questions. I looked at the GSS, and the same was true there. The trust question was given without the option of being careful once during 1983-87, and 55.7 per cent of respondents agreed that most people can be trusted. When the “being careful��? option was given, only 42.1 per cent of respondents did so.