Our study group, called Survey Methodology Workshop (SMW), consists of various members, who have been concerned with the Public Opinion Research, including scholars, survey methodologists and marketing researchers. The purpose of SMW is that each of the members is beyond his/her deference in the position and cooperates to get over the crisis of the Public Opinion Research. This special issue includes seven research papers written by the group members. From the first to the fourth paper, each of the authors writes about different survey methodologies, which cover most of the interview modes, including face-to-face interview surveys, postal mail surveys, Internet surveys and exit poll in turn. From the fifth to seventh paper, the other authors write about “practical issues” on survey methodologies of the Public Opinion Research. I mean by “practical issues” mean, in other words, the management of the actual research, the technique of sampling, and the design of the questionnaire. It is necessary to realize that survey methodology and measurement method used are both very important, as well as sampling technique, for the development of Public Opinion Research. Changes in our society require us the conversion of methods in the Public Opinion Research. We must build the Discipline of Total Survey Methodology which includes the research and study in area of research planning, survey management, and psychology of survey or interview. Finally, our papers contain a lot of fundamental findings to contribute to the study of the Public Opinion Research.
THE YOMIURI SHIMBUN has conducted monthly public opinion polls, using face-to-face interviews, since March 1978. Since the beginning of the 1990s, the response rate to these surveys has been decreasing, and is remarkably low in recent days. Since 2006, the average yearly response rate has been under 60%. I believe the main reason for the decline is increasing awareness of personal privacy. As evidence of this, there has been a greater rise in the percentage of people who decline to participate, than in the percentage of those who are not at home when interviewers visit. We have taken steps to improve the situation, such as hiring skilled interviewers, but it has not stopped the decline. At this moment, we do not have a sure-fire method for improving the rate of response. The current fashion is for RDD surveys, and if face-to-face interviews are to survive, I believe we must increase understanding of the importance of opinion polls. To that end, in addition to further improving the implementation of our surveys, we should make every effort to prepare fair-minded questions, analyze and report survey results appropriately and make data available to the public.
Many mail surveys have been conducted in Japan since the 1950's. Unfortunately, their response rates have been lower than other methods of administration. Various difficulties that result from the lack of an interviewer are major contributing factors of the lower response rates. This paper provides some guidelines regarding the design of mail questionnaires, the timing of the follow-ups, and other operational details. Mail surveys can potentially achieve high response rates. Further, their response rates in rural and urban areas are very similar. In addition, the response rates of mail surveys were found to be unrelated to the educational level of the respondents. A problem of mail surveys is the omission of questions by some respondents. However, there are usually few omitted questions and the overall quality of answers is usually better. The tendency to leave unanswered some branching questions is discussed and suggestions are provided how to decrease it.
This paper discusses the possibility that an Internet survey using volunteer access panels can measure the transition of public opinion. Any person who is an Internet user and has registered voluntarily can be a member of the volunteer access panels. Thus, a volunteer access panel represents not only a target population but also the Internet users in it. Monthly survey results over a period of eighteen months have been reported in this paper. These Internet surveys were conducted on the members of the volunteer access panel. Further, the survey results were compared with those of the consumer confidence survey that is conducted by the Cabinet Office. Subsequently, deviations as well as some correlations were observed between these two survey results.
Exit polls are excellent sources of information to understand voter behavior and to predict election outcomes. This paper presents the current methods of exit polls, which is conducted primarily by the Mainichi Newspapers, and discusses issues related to exit polls. Exit polls are based on a two-stage sampling design, with a polling precinct as the primary sampling unit and a voter as the final sampling unit. Precincts are stratified and selected systematically with probability proportional to the precincts' forecasted number of voters. The voters of a precinct are first selected systematically, with the interval predetermined based on the forecasted numbers of voters, and subsequently, asked for interviews. The datasets of exit polls that have been collected during 23 governor elections held since April, 2003 are analyzed. Standard errors of the estimated proportions of votes for 90 candidates are, on average, 1.5 times larger than the respective standard errors based on the simple random sampling. The analysis also reveals that the estimated proportions of votes for 20% of the candidates are outside the 95% confidence intervals, which suggests that there exist some sources of errors apart from the sampling error. The issues related to exit polls, including those concerning errors caused by an increasing number of early voters and non-responses, are discussed.
Taking notice of shapes of distributions of clinical laboratory values, we propose an analysis process based on the power-normal distribution as a model which treats variation of clinical laboratory values. Based on the process, we can treat the background factors comprehensively, and estimate suitable reference intervals of the clinical laboratory values. Further, we can consistently and flexibly treat outliers of the clinical laboratory values, classify healthy group and disease group, compare the reference intervals based on this model with traditional ones, and evaluate information loss which results from traditional estimation methods based on asymptotic behavior of estimates of reference limits. Practically, we applied this model to the medical examination data collected at a clinic K in 2003, and estimated reference intervals. As a result, almost all reference intervals were estimated much wider than traditional ones. This result suggests that it is risky to apply traditional masked reference intervals to relevant receivers of clinical diagnosis or actual subject group.
In this study, a secondary factor analysis of multiple populations with a structured mean was applied to the independent, and mean values were decomposed into common and unique components. In this manner, a method for discovering the most effective model was proposed. In order to ensure the practical effectiveness of this method, the data of Brand Japan from 2004 to 2006 was analyzed. This data was composed of 1,000 brands and 15 variables. As a result, some models fitted to the data well in all variations of models, and the best model could be decided by interpreting meanings. This model was applied to the brand data, and factor scores were calculated for every construct. Characteristics of every brand were discovered by observing its movement.
Previous applications using individual differences scaling (INDSCAL) have tended toward higher dimensionalities than in the case of one-mode two-way multidimensional scaling. The interpretation of the results obtained from INDSCAL has not always emphasized the way to visualize the solutions. To interpret the results more efficiently, this study proposes an approach that uses both low- and high-dimensional solutions of INDSCAL. Department store purchase data were analyzed using INDSCAL. This analysis examined the influences of store arrangement and season effects on multiple purchase behaviors. The results showed that consumers' multiple purchase behaviors are based on an image of the shop, the life stage of the consumers, and the layout of the department store. The store arrangement influences long-term purchase behaviors, which differ from period to period.