A Study of the Telephone Survey Method

Abstract

Until now, social surveys (i.e., public opinion surveys) heve been conducted using the face-to-face interview method. This is due to the facts that: (a) resident registrations (or voting registries), which are regarded as a sampling frame which adequately represents total populations, are available in Japan and can relatively easily facilitate obtaining unbiased probability samples, and (b) the reliability of the face-to-face interview method in surveys is regarded as high. Conversely, from the perspective of the spread of household telephones, the telephone survey method has been regarded as one by which it is impossible to obtain probability samples which represent ordinary citizens. Since the 1980s, household telephone coverage has increased (to over 90%). Consequently, the telephone survey method, which is especially useful for its promptness, has been used for voter forecasting surveys, and in the 1990s this method also has been utilized in public opinion surveys. However, not enough research on the telephone survey method has yet been done with respect to the representativeness of its samples or the reliability of survey results. This paper discusses past and current research regarding telephone surveys (specifically the current RDD survey method), and actual survey results, including their errors (except sampling errors). Moreover, by comparing the responses using both the telephone survey method and face-to-face interviews, the differences between the two methods are discussed. The current RDD method is viable; however, we cannot predict this method's future due to developments in communication technology. Finally, because there are many problems to overcome with respect to actual survey methods, further research, e.g., accumulating actual survey results and re-analyzing existing survey data, is necessary.