Japanese Aging and Health Dynamics Study
National Survey of the Japanese Elderly

A total of ten waves of surveys were conducted between 1987 and 2021.
All participants newly added to the survey were randomly selected throughout Japan using a technique called two-stage stratified random sampling, although the year they were first added varied depending on the participant (as shown in Fig. 1). Note

(Note) Regions were stratified according to a combination of regional block (Hokkaido, Tohoku, Kanto, etc.) and population, and the number of persons (sample size) selected from each stratum was determined depending on the population of the target ages in the stratum. The procedure for extracting samples involved first selecting 192 survey locations throughout the country, then selecting individuals of the relevant ages from each location using the Basic Resident Register.

Participants who joined the survey for the first time were generally asked to participate in each subsequent survey, unless they died or declined to continue participating. Whether they actually participate is of course left to the discretion of the individual. By the ninth survey, 6,665 individuals had participated at least once.

Central Research Services, Inc. conducted visit interviews from September to December of each year of the survey. The surveys were conducted as interviews of the participants themselves in principle. Beginning from the second wave, however, proxies such as family members were interviewed (for certain items) only if the participant was unable to respond for a reason such as serious illness.

The tenth wave survey was conducted in 2021 during the COVID-19 pandemic. A home-visit survey was conducted for the E and F participants of Fig. 1, while a mail survey was done for the A to D participants. Regarding the home-visit survey, the method of leave-behind questionnaire together with face-to-face interviews was used to shorten the interview length.

The following describes the participants of each survey in somewhat greater detail.

[Birth year of the respondents]
A: 1893-1927, B: 1927-1930, C: 1930-1936, D: 1898-1929, E: 1919-1952, F: 1928-1961

In the first wave (1987), responses were obtained from 2,200 individuals aged 60 years or older throughout Japan (response rate, 67.2%; A of Fig. 1). Follow-up surveys were subsequently conducted every 3 years. Individuals aged 60-62 years and those aged 60-65 years were newly selected at random and supplemented to the panel in the second (1990, B of Fig. 1) and fourth (1996, C of Fig. 1) waves, respectively.

In the second and fourth waves, the number of newly selected individuals (sample size) was specified (second wave: 580, fourth wave: 1,210) so that the selection probability for an age group was roughly equal to its previous selection probability for an older age group. This was done so that the data analyzed were an unbiased reflection of individuals aged 60 years or older throughout Japan.

By the fourth wave, 3,580 individuals had participated (including proxies) and were asked to be participants in subsequent follow-up surveys.

In the fifth wave conducted in 1999, 2,000 new individuals aged 70 years or older were randomly selected (D of Fig. 1), while continuing follow-up of the A to C participants who had cooperated through the fourth wave. Individuals aged 70 years or older were added because the issues of the older old were a focus of the survey (→ see About the Project). Of the 2,000 new individuals, 1,635 who responded were added to the follow-up panel.

The sixth (2002) and seventh (2006) waves continued the follow-up of the A to D participants, without supplementing new samples.

In the eighth wave (2012), 2,500 individuals aged 60-92 years (E of Fig. 1) were randomly selected, while continuing the follow-up of the A to D participants. Of the new E participants who were respondents in the eighth wave (n=1,450), proxy participants and those who did not consent to subsequent participation were excluded from follow-up, and 1,257 remained as follow-up panel. The ninth wave (2017) continued follow-up of the A to E participants, without supplementing new samples.

The A to D participants received a short questionnaire with limited items such as health status, instead of home visiting, and 443 (61%) of the 730 participants returned the questionnaire, including proxy responses. The participants who commenced participation in the eighth wave (E of Fig. 1) as well as 2,700 newly selected individuals aged 60-92 years (F of Fig. 1) were asked to participate in the home-visit survey consisting of face-to-face interview and leave-behind questionnaire. Regarding the new sample F, 1,227 responded to the tenth wave survey by themselves or by proxies, and approximately 80% of them consented to subsequent participation.

The number of respondents (including proxy respondents) for each wave ranged from 2,000 to 4,000 through the eighth wave, but dropped below 2,000 for the first time in the ninth wave. The decrease occurred because death increased as the participants aged. The number of participants confirmed to have died up to the tenth wave (2021) was 2,674 for the A to C participants in Fig. 1 (of 3,580 follow-up participants), 1,301 for the D participants (of 1,635 follow-up participants), and 248 for the E participants (of 1,257 follow-up participants).

The response rate was 67% in the first wave (not including proxy response for the first wave), approximately 85% in the second to sixth waves, approximately 75% in the seventh and ninth waves, 67% in the eighth wave, and 54% in the tenth wave. However, it should be noted that comparing the response rates across waves is difficult because the presence or absence and proportion of newly added participants and the ages of the participants varied.

The response rate for the newly added sample is getting worse in recent surveys, partly due to the increasing privacy consciousness. It was 59% for the newly added E participants (Fig. 1) in the eighth wave and 46% for the F participants in the tenth wave, showing a greater decline than before. In particular, the tenth wave survey conducted in October to December in 2021 showed a larger regional difference in the response rate compared with the previous waves, suggesting a possibility of the influence of the COVID-19 pandemic. Meanwhile, the response rate for the follow-up participants who had experienced the survey(s) before exceeded 75% even in the eighth and tenth waves.

For more information, please see Table 1.