Identifying the Japanese general population: a sampling method.

HOME | About the Study | Questionnaire | FAQ (Japanese only) | Author

Identifying the Japanese population in Australia for epidemiological mental health research: a sampling method using residential telephone directory and statistics of Japanese family names.

KOICHI TAKADA
Centre for Mental Health Research, The Australian National University, Canberra

Abstract

Objective:To establish a practicable, efficient sampling method to access the Japanese population for a population-based cross-sectional postal survey in Melbourne. Background:No epidemiological mental health survey has been conducted in the Japanese community in Australia, partly due to its small size (0.2% of the Australian population) and the difficulties in identifying and accessing study populations. For instance, the large Japanese Society Inc. in Melbourne has a relatively small membership that covers less than 20% of the estimated Japanese population. Most of them are not in the general population, but are company-related workers stationed here with their family. Method:A Japanese surname directory written in Roman alphabets was re-produced from statistics of Japanese family names based on 1.8 million residential telephone lines in Tokyo Metro Area and neighbouring Kanagawa Prefecture. These are the most densely populated areas in Japan and its top 1000 family names share 72% of population. To search potential subjects in Melbourne the electronic residential telephone directory was used in conjunction with the name directory, in which 868 surnames were specified. Results:Using this method, 887 households with Japanese sounding surnames were identified. The cumulative curve of those identified was quite similar to that from original statistics. The questionnaire was sent out to all potential subjects. Of the 887, 210 (23.7%) addresses were undeliverable or no longer available, 38 (4.3%) were non-Japanese, and 38 (4.3%) refused. The age distribution and sex ratio of 255 (28.8%) of responders had some difference from the 1996 Census data. Conclusion:Disadvantages of this method are 'silent' numbers and people whose names are no longer Japanese. However, it seems the best way at this time because it can provide the most practicable sampling method.

[Presented at Annual Scientific Meeting of the Australasian Society for Psychiatric Research, Adelaide, 7-8 December 2000.]

Objective:

To establish a practicable, efficient sampling method to access the Japanese population for the Study on QOL and Mental Health among Japanese living in Australia (QMJA), a population-based cross-sectional postal survey. (See Box 1)

Box 1:

Study on QOL and Mental Health among Japanese living in Australia (QMJA):

Study on QOL and Mental Health among Japanese living in Australia (QMJA) is the first population-based mental health survey in the Japanese community in Australia. This comprehensive survey includes several topics such as their subjective perception of quality of life, mental health status, impact of culture shock and acculturation, prevalence of psychiatric symptoms, mental health literacy, and attitudes towards and utilisation of mental health resources. A pilot study was carried out in Sydney with 414 potential Japanese subjects who were sampled by this method. No statistically significant difference was found across three different version (8-, 16- and 24-pages) of questionnaires which include WHOQOL-BREF, GHQ-12, Culture Shock Questionnaire, Social Support Questionnaire, AUDIT and newly developed instruments related to such topics. Postal survey in Melbourne was conducted in September 2000 with 24-page version. The results due out early in 2001.

Definition:

Japanese refers to either permanent or temporary residents of Australia who had origin in Japan and included those born in Australia regardless of their legal nationality. Those who have difficulty in reading Japanese language were excluded.

Site:

Melbourne was chosen. Size of estimated Japanese population is appropriate for QMJA.

Background:

No epidemiological mental health survey has been conducted in the Japanese community in Australia, partly due to its small size (0.2% of the Australian population) and the difficulties in identifying and accessing study populations. Door-to-door survey such as Census is not practicable for a small population in which language other than English is spoken.

Melbourne has two major groups of Japanese people: the Japanese Society in Melbourne Inc. and Japan Club of Victoria Inc. Even the largest group has a relatively small membership that covers less than 20% of the estimated Japanese population (Fig. 1). Most of them are not in the general population, but are company-related workers stationed there with their family. Samples from such a group are therefore potentially biased. Residential roll of Japanese nationals kept by Consulate of Japan may not be released even for research purposes. Therefore, no databases of whereabouts of Japanese general population are available.

Figure 1 Groups of Japanese in Melbourne

Method:

The most frequent 1000 surnames were quoted from the Japanese surname statistics (See Box 2) to re-produce the Japanese surname directory.

These 1000 surnames were transliterated and were classified into 889 surname categories in Roman alphabets with careful attention to homonyms. Frequency of each surname category was re-calculated.

Using 889 surnames, potential subjects were searched in the electronic residential telephone directory. To check their mobility and update the mailing list, they were searched again 9 weeks later.

Twenty-one surnames found with exceptional values were excluded from the surname directory (Fig. 4). They can be seen as surnames from certain ethnicity other than Japanese. Eventually, 868 surnames were specified as Japanese.

Finally, 887 potential subjects (households) were identified. Postal survey (QMJA) was conducted with them. Non-Japanese were encouraged to return a deletion form to avoid receiving further survey documents (e.g., reminders).

To test the practicability of this sampling method, demo-graphic data of the respondents were compared with that of 1996 Census.

Box 2:

Statistics of Japanese Surnames^[1]:

Statistics of Japanese surnames were based on 1.8 million residential telephone line database in Tokyo Metro Area and neighbouring Kanagawa Prefecture where 16% of Japanese population live. About 50000 surnames were identified in this area. According to the statistics, each Japanese surname generally has low frequency (e.g., Suzuki, the most frequent surname share only 2.0% of population) and the most frequent 1000 surnames share 72% cumulatively. (Fig. 3).

Figure 3 Frequency of Japanese Surname

Figure 4 Number of Potential Subjects Found

Results:

Generally, Japanese surnames are highly distinctive. Only 21 (2.4%) of 889 surnames are not suitable to specify Japanese.

After 9 weeks interval, 8.4% of potential subjects no longer appear on the telephone directory and other 8.9% are newly listed on.

Eight hundred eighty-seven households with Japanese sounding surnames were identified. Figure 5 shows cumulative curve of those identified is quite similar to that from original statistics (Fig.3).

Of the 887, 210 (23.7%) addresses were undeliverable or no longer available, 29 (3.3%) were non-Japanese, 10 (1.1%) were Japanese descendants but could not read Japanese, and 38 (4.3%) refused.

Among 258 responded households (29.1%), 330 individuals (male, 144; female, 186) answered the questionnaires. Figure 6 shows age and sex distribution of the responders.

Comparing the total sample of people born in Japan from 1996 Census data (Fig. 7), the female dominance in sex ratio (1.26 female to male) in responders is consistent. (The sex ratio for age 18 and over from Census Data is calculated as 1.61 female to male.)

Overall response rate for population is 11.4% (13.0% for male, 10.4% for female).

In particular age groups, there are discrepancies between response rates and population (Table 1). These are high response rates for male of aged 50-59 and 60-69 (36.5% and 30.0% for relevant population, respectively) and for female of aged 50-59 (22.3%), and on the other hand low response rates for female of aged 60-79 (0.1%).

Figure 5 Cumulative Number of Potential Subjects Found

Figure 6 Age & Sex Respondents

Figure 7 People Born in Japan (Melbourne, 1996 Census)

Table 1 Age, sex and response rates

Discussion:

Advantages and disadvantages of this sampling method are shown in Box 3.

Representativeness is the key issue of this method. No appropriate demographic data of Japanese is available to evaluate the representativeness of this sample. Census data is the only useful source even though it uses classification of 'people born in Japan' not exact Japanese in cultural context. Non-Japanese born in Japan is small in number and Japanese born in Australia has not yet been a big population because Japanese community in Australia has been recently settled (after 1980s).

Some possible explanations for discrepancy of sex and age between responders and population are;

There is an observation variance between the study and Census (actually, 15 of 22 individuals aged 50-53 had lived in Melbourne and had been under 50 at the time of 1996 Census), and
Most of females who marry non-Japanese take their husband's surname (especially, "war brides" aged mid-60s and over).

Two thirds of responded households were not involved in any Japanese community groups. In other words, this method could access to people in general population.

In conclusion, this sampling method provides pragmatic level of representativeness and has much more advantages than disadvantages in practical aspects.

Box 3:
Advantages:

The latest data of subjects are available compared to any other written sources. This increases accessibility to subjects with high mobility.

Able to search surnames much more efficiently than to pick up each names alphabetically.

Applicable in any part of Australia.

Able to obtain less biased samples than other sources (e.g., member lists of organisations).

Low proportion of false positives (non-Japanese with Japanese names) increases specificity and saves mailing cost.

Applicable to people from other ethnic origins whose surnames have certain distinctiveness.

Disadvantages:

Unable to sample Japanese with non-Japanese surname (false negatives). This reduces sensitivity.

Unable to access people who have no residential telephone lines registered with their own names (e.g., students in halls of residence) as well as those who have silent numbers.

On condition that factors mentioned above are not negligible, representativeness of the samples will be weakened.

Addresses from telephone directory are not always accurate (e.g., no flat/unit numbers). In such cases telephone contacts are needed to confirm their postal addresses.

Reference:

[1] Tsuyama T. Frequency distribution of Japanese surnames by kanji and kana: 1.8 million samples in Tokyo and Kanagawa. URL: http://www.alles.or.jp/~tsuyama/name.htm, 1998. (in Japanese)

Ethical Clearance and Copyright:

The protocol of this research was approved by the Human Research Ethics Committee of the Australian National University, and the research was carried out strictly in accordance with the ethical guidelines set out by the National Health and Medical Research Council. Use of the telephone directory for the purpose of research is permitted under the Copyright Act 1968. Use of all instruments (and their Japanese translation) in the study were permitted by copyright holders.

A full paper in Japanese with English abstract was published on the Japanese Bulletin of Social Psychiatry.

Takada K. Sampling and identification in the general population of the Japanese community overseas: use of electronic residential telephone directory and surname statistics. Jpn Bull Soc Psychiat, 11; 1-10, 2002.

Go to Top

Return to Questionnire (Information Sheets in English)

Return to About the Study (English)

Return to About the Study (Japanese)