JGSS survey methods differ slightly each survey year, but generally the overall methodology is as follows. The survey subject population consists of men and women aged not under 20 to 89 as of September 1st of the particular survey year, and subjects are selected using a stratified 2-stage sampling method. The stratification divides Japan into the six blocks of Hokkaido/Tohoku, Kanto, Chubu, Kinki, Chugoku/Shikoku, and Kyushu, and dividing those into the three groups of largest cities, other cities, and towns/villages according to the size of the cities and districts in each block (JGSS-2006 and later uses four groups of largest cities, cities with populations of 200 thousand or more, cities with populations under 200 thousand, and towns/villages). Using census divisions as the unit of sampling, survey locations were sampled from each stratum. The number of survey points was set such that the maximum number of subjects in a single survey point was approximately 15. Voting registers (or if not permitted, basic resident registers) were systematically sampled in each survey point for subjects.
Data was collected through a combination of interviews and self-administered questionnaires. In other words, survey items overall were divided into 1) questions appearing on the interview questionnaire and 2) questions appearing on the self-administered questionnaire. We rely upon the respondents to answer both questionnaires. When we decide into which questionnaires the questions would be incorporated, we took the characteristics of each questionnaire into account in order to use it to maximum effect. The interview questionnaire incorporates questions with complicated answers, such as ones with many nested sub-questions. The self-administered questionnaire incorporates questions with simple answers, or ones which strongly call for privacy considerations. The duration of each questionnaire is approximately 20 minutes, and we aim to conclude the surveys in about 40 minutes total. We leave it up to the respondents to decide which questionnaire to answer first, but in the average year, the majority performs the interview questionnaire first.
JGSS started with using one type each of the interview and self-administered questionnaires, but for JGSS-2003 we prepared two self-administered questionnaires (From A and B) and assigned from A to one half of the subjects, and form B to the other half. This method became standardized starting with JGSS-2006. To ensure that a sufficient number of responses on each of the two types of self-administered questionnaires was obtained when those questionnaires were prepared, the overall sample size was set high.
JGSS survey items were divided into core questions that essentially appeared on every survey, and topical questions that appeared once, or only once in several surveys. Core questions included questions related to the respondents’ occupations, household composition, and other basic attributes, and ones concerning the respondents’ daily activities, basic values and behavioral patterns, political attitudes, and other issues. In principle, the same core questions appeared again and again on every survey, but were sometimes adjusted slightly from survey year to survey year. The topical questions included questions related to events that had attracted public attention in the particular survey point, and ones that narrowed the focus to specific themes so as to make it easy to perform focused analysis. Starting with JGSS-2005, questions solicited from general researchers are also included among the topical questions. (For details, see "List of Survey Items by Year")