抄録
Between 1996 and 1998, the National Institute for Japanese Language conducted a survey among 28 nations, the main theme of which was “A General Study of the Japanese Language in Global Society”.
Including only the 18 of the 28 nations whose samples were nationwide, the purpose of this report is to focus on the “International Census on Japanese Language Usage”, a sub-theme of the General Study. To do this, the report introduces and summarizes (1) the items which relate to sampling design and interviewee selection, and (2) the items which relate to the survey's conduct, such as the sampling methods used by interviewers. These two items then form the basis for our investigation of data quality.
In cases where data containing differing total samples are compared, ideally the collected data are homogeneous. In cases where the samples being surveyed are nations, whose cultures, institutions and environments naturally vary, problems of comparability are inevitable.
Data quality is affected by various factors at each stage of the survey. This is especially true when cross-national surveys are conducted where sampling designs and interviewee selection procedures vary, as these are two of the most notable items to be affected. In our survey design, we emphasized to each person in charge of the individual national surveys that data quality would be enhanced by maximizing the homogeneity among nations. However, in the cross-national surveys in our survey field, there are some issues which can rarely satisfy this condition. Here, we present concrete outlines of sampling designs for surveys which we have conducted among several nations in our cross-national studies.
