Elderly survivors of the 1995 Great Hanshin Earthquake in Japan

Author: Junko Otani  

Setting up the Project

My focus was research on older people in urban areas who are poor and have no functioning family. This is a group that will be of increasing concern for the future in Japan and many other countries (Wilson, 2000: 117). My study population lost their homes in the 1995 Great Hanshin Earthquake in Kobe, Japan, and were repeatedly relocated to various types of housing schemes in the following years. The primary research methods were media analysis of TV programmes and broadsheet newspapers, a reanalysis of the Hyogo Health Surveys, and ethnographic research at elected Kasetsu (temporary shelter housing: TSH) and Fukkō Jutaku (public reconstruction housing: PRH) compounds in central and suburban Kobe. The focus of the research is the processes of reconstruction for older people after the earthquake, with special reference to housing and community work.

For more information on the earthquake, go to http://www.kanadas.com/kobe-quake/  , and http://en.wikipedia.org/wiki/Kobe_earthquake.  

The above not only set out the research but also gave grounds to support it, saying that this research population is a model of the future ageing society in Japan. It is an increasingly important group of people, and is relatively new on this mass scale.

By looking at the highly age biased community of TSH created after the Kobe Earthquake and the following stage of PRH, this research follows the processes of reconstruction for older people after the earthquake with special reference to housing and community work.

My fieldwork at TSH was carried out in 1998 to early 1999. The first set of media data was collected in January 1999 at the week of the fourth anniversary of the Earthquake. Then my fieldwork at PRH was carried out in the following months. Then the second set of media data was collected in January 2000 at the week of the fifth anniversary of the Earthquake.  This allowed me to focus first on media, then ethnographic fieldwork, and then back to media. Also this allowed me to observe the communities in my ethnographic fieldwork while the media was filming to prepare for the 2000 wave broadcasting.

Reanalysis of survey data provided the background context of my study population and also indicated how government was seeing the needs of my study population. I was proposing this research as my PhD and at the time, the medical school (LSHTM) still tended not to accept research with small sample which is sufficient when design a qualitative approach. Using survey data was also a help to convince my committee who asked a question of generalization although I explained that my research was not aiming generalization.

Most interviewees were those who were not in my ethnography sites. A few interviewees were those in the sites such as the chief of the self-governing body. It was a courtesy to do an interview to them, recognising them as a leader of the community when entering there and they expected me to do an interview with them.

I used a mixed method of qualitative and quantitative approaches. By doing secondary analysis of the Hyogo survey data, I examined the changes that the different surveys show. By sampling the media, I showed the main foci of public attention, how their views changed and how what they emphasised or presented changed. The media is an important part of my research in the Japanese context. Older people, especially older people living alone, received considerable attention. I also sampled three sites in terms of what was happening on the ground and conducted discourse analysis. 


Public health survey

The Hyogo Health survey was conducted in October each year and the data were compiled as a report book in March. Although basic information, such as demographic and social indicators, was collected throughout the surveys, the questions more specific to some health issues changed over time. The second and third year questions were modified based on the answers to the first year open-ended questions and results of other questions and overall feedback from public health nurses working in the field.

Table 1 shows the response rate of study populations of the Hyogo Health Survey from year 1996 to year 1998 by the housing categories and shows the change of the surveyed housing categories over time.

Table 1: Response Rate of Study Populations of the Hyogo Health Survey 1996 - 1998 





Temporary Shelter

  Distributed/ Valid










Public Reconstruction Housing

  Distributed/ Valid










  Distributed/ Valid

902/ 2,714







Number of completed questionnaire/valid responses over the distributed/mailed questionnaire (%: Response rate).

Source: Hyogo Prefecture Department of Health Earthquake-affected household health surveys, 1996-1998.

Media data

Media data were collected, mainly on the fourth and fifth anniversaries of the Earthquake (17 January 1999 and 2000) for ten days each, in addition to the published reports and books produced by the media. Media coverage goes high for about one week before towards the anniversary date of the earthquake every year. I have classified data from late 1998 to January 1999 as the 1999 wave and those from late 1999 to January 2000 as the 2000 wave, following Altheide (1996) who refers to time based samples of media reporting as ‘waves’. Among the earthquake programmes, I decided to focus on those covering people’s life in the disaster-affected community and did not collect or use those reporting other subjects such as disaster prevention technology. Transcriptions were made of all the media data in the beginning but gradually focused to the coverage of older people, housing, and health by theoretical sampling as I started to do analysis while still collecting more data. Media data were collected in parallel with data collected in the ethnographic fieldwork and interviews.

I collected media data of both TV and newspapers. I videotaped TV coverage. As the total length of videotapes is 67 hours, and as NVivo could then not directly code such multimedia documents as videos, but only link to them, I prepared transcripts from the videos in English so that I could code them as documents. I interviewed a TV reporter and programme director, and a newspaper journalist.


I interviewed public officials at the Department of Health of Kobe city, officials at the Department of Health of the West Ward of Kobe City and an official of the Central Ward of Kobe city. Before interviewing the Director of the Department of Health of Kobe City, I interviewed the Director of Planning Division of Welfare for the Elderly of the Japanese Ministry of Health and Welfare in Tokyo. I interviewed other officials of the West Ward and the Central Ward of Kobe City when I visited their community activities.

Location of field sites for ethnographic work

After I visited several community locations, I selected three community locations by theoretical sampling. Location, conditions of facilities, and types of actors and activities were all considered in the selection process, as were the different types of housing schemes such as individual apartments, a group house, and Silver Housing Scheme because it would illustrate housing aspects of my research question of whether community generation differs by housing and how and if loneliness can be alleviated by housing. Table 2 summarizes the three research sites of the field work.

Table 2: Three research sites of the fieldwork


TSH: Haruyama

PRH 1: Natsu-Aki

PRH 2: Fuyuyama





Kobe city




Building type


Hutted apartments

High rise

High rise

Number of household units


700 (11 buildings) at Natsuyama PRH and

500 (3 buildings) at Akiyama PRH

550 (2 buildings) in total

Silver housing



200 (lower levels of one of the buildings)


Volunteer nurse

A woman who is a welfare commissioner and chief of the self-governing body and of volunteer group of neighbouring community

A male chief of the self-governing body


Participant observation

 I shadowed and worked with a public health nurse for a home visit.

 I worked as a volunteer waitress at a community tea shop.

Shadow a community leader

 I attended various community activities such as lunch programme, birthday party, day-service for the handicapped / older people.

 I accompanied a community leader for a home visit.

 I interviewed a chief of self-governing body.

Observation at events

 I attended several community activities such as a breakfast programme with local Catholic church,  health promotion programme with Health dept of municipal government.

I interviewed a chief of self-governing body and Life Support Advisors (LSAs).


Working with data

Media data were collected with theoretical sampling method and discourse analysis was done. For the analysis of fieldwork data, case study analysis and discourse analysis were employed. Interpretation and analysis of the field data texts were attempted. Secondary analysis was done on the public health surveys. Narratives were used as a preliminary way of dividing the data, then the narratives were analysed into discourses.

I organized the data according to type and research site. For media data, the data were analysed by channel/stations, i.e. public and private stations, and local and Japan-wide stations. This helped me to see that each station had its own mandate and that there were biases. Changes in focus and tone over time were analysed. And a gender perspective was included. Gender was important in media and my ethnographic fieldwork but it was lacking in the original analysis of the Public Health Survey done by the Hyogo Prefecture.

By coding and searching for a person’s name in all documents, I was able to compare how the tones of what was said and shown were different in various kinds of data. The same person appeared in TV video transcripts, newspaper articles, and personal interviews and observations in my ethnographic field notes.

In the course of data handling in this research, I created in the software several Document Sets (grouped items accessible for analysis as a group): Media , Haruyama TSH, Natsu-Aki PRH, and Fuyuyama PRH. The Media Document Set includes TV video transcripts, newspaper articles, and interview notes with a TV reporter and a newspaper journalist. This meant various types of data on the media were grouped into one set. I made Document sets for field notes and interviews at each research site. Since I was collecting data from various sources in parallel chronologically, creating such document sets helped me to organize vast pages of data, which might be stamped with the same date but should go to different folders. And later this set became a base idea of outline of chapters for my PhD thesis. With or without NVivo, still the outline could be the same eventually but NVivo helped me to see the chaos in more organized way at the time.

This research project was one of the first applications of NVivo in a non-English context, and is probably the first in Japanese. Japanese is a language that often skips subject and object, especially in conversational use. When a conversation dialogue sentence or phrase is cut out by coding, and separated from the paragraphs, it is clearer in the English translation than in Japanese what the word or sentence means in the context. NVivo enabled me to locate content not only in the same set of data, but in other types of data, i.e. TV, newspaper, ethnography and interviews, and so to make comparisons.

I started with the media data. I coded these documents by In-Vivo coding. The In-Vivo coding tool is to select a word from the text data and to apply it as a code. (In NVivo, this is done by making and naming a ‘node’.) It was a convenient way to start. Also it was a good way to learn the words used which reflected the new concepts born after the Earthquake, the new words created. I am a native speaker of Japanese, especially with dialects of the Osaka-Kobe area, but there were new words used that I did not know. If I use those words while talking with people from Tokyo or other outside areas in Japan, they do not know or asked if such a word exists! Prof Lyn Richards advised about coding, “Bother with the naming. Use terms that occur in the data (in-vivo codes) only if they accurately name the category.” But I got this advice after finished most of the primary coding. 

From the beginning, I used the In-Vivo coding tool mostly. This was convenient and helped me start thinking analytically. Yet one result was that some nodes, which could be grouped under one node at the first coding stage, were not combined. I was able to merge these nodes and rename them. While coding, I also used the software to link from the text to memos and started to write up memos.

I used colour coding to enjoy its fun functions.  I colour-coded words pointed out by my English supervisor in London, e.g. “What does “loneliness’ mean in the Japanese context?” It was used to develop a discourse of loneliness in the Japanese context. Looking at Japan with Western views is an important aspect in my research.

Different people code data differently. I would like to share with you an interesting experience. My supervisor also thought it was a good example and shared with her other students at the London School of Economics in her class. I participated in the qualitative research approach seminar by University of Western Sydney in Australia for one week in November, 1999. They kindly used my data at the seminar. Prof Lyn Richards made all the participants code my data. It was very interesting to see how differently the coding evolves. Prof Richards is a family sociologist. She picked up a word like “burden”. She also said “Kodokushi (Isolated death)? You have a word for this?” My answer was, “Yes, it is a very sad and miserable thing if one dies when no one is attending.” This is an interesting example of looking at Japanese values with Western views. She also stopped at the sentence by the Director of Health of Kobe city, “Women are doing OK. The problem is men. Imagine. What will become of me if my wife dies!” Dr Pat Bazeley is a community psychologist. She has done research on housing and care in Australia. She picked up and laughed to read “Community development through Origami class.” I did not realise that may be something weird until I heard her laugh. It is to me a rather common practice.

Analysis process

Triangulation (Robson, 2002: 174) was central to my research design. Data triangulation allowed a comparison of media data and field data when doing discourse analysis on loneliness. Media data was related to interviews with media people.  Methodological triangulation, a process of combining quantitative and qualitative approaches, was employed for analysis of surveys, media data and observation data in community research contexts.

Gender perspectives were employed in all analyses. A gender focus was lacking in public surveys, yet gender was important in qualitative analysis in the media and field sites.

For the Hyogo Health Survey, I compared the issues the survey was looking at over the years and analysed what was concerned by the health department, what was found out and concluded from the previous year’s survey, and how it reflected to the following years. I also made tables and graphs comparing the data over the time for the same indicators when comparable indicators and figures were available from different years, and analysed them.

Program logic models, a combination of pattern-matching and time-series analysis (Yin, 1994:118) was employed in my methods. These strategies are useful for explanatory, exploratory, and case studies. Pattern-matching analysis was conducted as follows. Having the Natsu-Aki PRH (Natsuyama PRH and Akiyama PRH) and Fuyuyama PRH enabled some pattern-matching to identify the commonality and variations, for example, the needs of leadership in community generation. At the same time it showed the variations in leadership to be seen by comparing community case studies.

I used NVivo  to aid discourse analysis of media data, and content analysis of all interviews, field notes, observational data and relevant documents and to identify the different discourses relevant to the aims and objectives of reconstruction, the definitions of outcomes and measures of success and failure, and the attitudes towards family and individual welfare, and Japanese values.

How did my software affect my analysis?  NVivo helped me to analyse the different types of leadership of key people by retrieving by category systematically to follow an individual case study with data scattered across field notes from various dates. For example, the programme assisted me in following a single case of a survivor who changed from a closed attitude to become socially outgoing over the time of his move from TSH to his new life at PRH. It helped me to search for the cases of people who appeared both in TV news and in my fieldwork, and helped me to search for media discourse on Kodokushi in the case of total lack of family. Having the whole set of data in one software package encouraged me to revisit the data over and over again while making an argument and helped me to locate the evidence in the systematic way, not simply by eyeballing. (Richards & Richards, 1991: 38 - 72)

I used Text Search on newly imported documents to locate ideas and get a quick overview of the whole in a short time before coding line by line. I also used Text search, for example, to gather at a node of everything about a key person to understand his/or her philosophy, to make comparisons, and to see consistencies or contradictions by looking at data from different times or different types of data. Some people I met during fieldwork also appeared in TV or newspaper coverage.

Content analysis helped me to see how media attention was shifted to a different focus and away from older people. Content analysis can be extremely laborious and time-consuming. It is a field where computerization has led to substantial benefits (Robson, 2002: 357).

As a part of the analysis process, I made a memo named “Diary, to Supervisor”, to record the analysis process and note issues to discuss with my supervisor. This was date/time stamped and included ideas to do next, before forgetting. 

Reporting the project

I conducted research as my PhD thesis and later I worked on it to publish as a book in Japanese, and further in Chinese. The recovery process after the Great Hanshin (Osaka-Kobe) Earthquake provided the opportunity for experimentation in community development for a highly aged society of older people with non-functioning families.  This is an increasingly important group in modern society, as the population ages. Kobe therefore received attention as a future model of Japanese society. This research looked at the elderly survivors of the Kobe Earthquake, especially in terms of community development where a large proportion of residents are older people and, furthermore, older people living alone in high-rise apartment buildings of Public Reconstruction Housing (PRH) and, in comparison, with the hutted apartments of Temporary Shelter Housing (TSH).  The thesis has not solved all the debates regarding community development for a highly aged society with a large proportion of older people living alone, for policy analysis or as an operational framework.  However, key lessons can be drawn from the exploratory micro-study research conducted in the context of Japan.

Quantitative analysis of media data provided evidence to show the actual length of the time spent on specific topics such as the focus on older survivors. It showed the changes and shift in focus over the years.

From the analysis of ethnographic field work data, my research supported the theory that local community support is needed for government officials to work in a community and my methods showed the variations in local support.

This study showed that one set of myths about temporary shelter housing was only partly true and that public reconstruction housing is far from a simple solution to the problem of rehousing survivors. Case studies of media data produced evidence of loneliness and Kodukushi and showed how these topics were built up from very little into new facts and new aspects of culture. In addition to the media discourse, analysis of my fieldwork observation data enabled me to produce an alternative discourse on these issues. As the analysis of media highlighted, the sense of isolation and loneliness was prevalent in my fieldwork sites, both at TSH and PRH, but it was partly because of the timing of TSH closing down after the peak of their activity, and of PRH starting up a new community. In PRH there was more of a forward looking discourse but only a longer time frame would have allowed any conclusions over whether loneliness would decrease over time.

The conclusion drawn from this evidence was that disasters are long drawn out events for vulnerable older people, especially those without money or families. Official statistics and the media make their own interpretations of what is going on, and the workers on the ground reproduce many of these views and some old prejudices of their own. Policy implications of this study’s findings are considered in my publications in Japanese, Chinese and English.


Junko Otani (2000), “Gendered Dynamics of Community Care Development: The Kobe experience”, Journal of Asian Women's Studies,Vol.9 pp. 14-29, 2000.12.


大谷順子(2001),「阪神大震災後 神戸の高齢者の経験 - ジェンダーの視点から」,『アジア女性研究』,第10号、32-39頁、2001.3



大谷順子(2006)『事例研究の革新的方法―阪神大震災被災高齢者の五年と高齢化社会の未来像―』、九州大学出版会 [Japanese]

Junko Otani (2006)“Jirei Kenkyu no Kakushinteki Houhou: Hanshin Awaji Daishinsai Hisai Koureisha no Gonen to Koureika Shakai no Miraizo”, Kyushu University Press, Japan, 2006




原著者 大谷順子、訳者 徐 涛、 監修 大谷順子   


Chugoku Shoten, Japan, and Nantian publishers:Taiwan,  (Oct. 2009).


Junko Otani (2010), “Older people in natural disasters” [English], Kyoto University Press, Japan, and Trans Pacific Press:Australia, (January 2010).

Looking back

More and more disasters have occurred next to another in the past years worldwide, including 2011 Great East Japan Earthquake and Tsunami disaster, 2008 and 2013 Sichuan Earthquakes in China, 2010 Haiti earthquake and 20102011 Canterbury Earthquakes in New Zealand.

This research is relevant. On the other hand each disaster has its own different context.

I have been looking at some other disaster cases but have not been able to do in the same or similar way, no-participant observation as deep as my Kobe field work, for example, due to money and time.  Meanwhile, reviews have appeared of my book (see References section above.)

I had a media data in my Kobe research.  I made a transcription text and imported into NVivo for qualitative analysis.  With a newer version, some could be done better and fancier by directly coding on the image.

Qualitative analysis software has not been implemented or widely used among Japanese domestic users despite my expectation that it would grow more popularly among researchers. 


Junko Otani

I initially trained as dentist in Osaka and in public health at Harvard where quantitative approaches such as epidemiology and biostatistics were traditional. I further received training in qualitative approaches since recognized its importance. I conducted my PhD research in London looking at elderly survivors of the 1995 Great Hanshin Earthquake in Kobe-Osaka area of Japan for their long-term life reconstruction. Whilst working with the World Bank and World Health Organization (WHO) in China and Geneva, I traveled internationally at my writing up stage but survived with NVivo data and writing management. I took my first academic post at Kyushu University in April 2005 where I experienced the Fukuoka West-Off Earthquake on my arrival. With University's grant, I invited Prof Lyn Richards to Kyushu University in September 2006, which was well-received and many researchers from all over Japan came to listen to the open lecture. I took my second academic post at Osaka University in my home town from October 2008. My current research focus is the longer-term consequences of the 2008 Wenchuan Earthquake in Sichuan, China. I was invited to deliver a key lecture at the launch of Japanese version of NVivo at the Australian Embassy in Tokyo in June 2007. I am a translator of the book, "Handling Qualitative Data: A practical guide", Lyn Richards, Sage, 2005, into Japanese edition, to be published from KitaOji Shobo in Kyoto in July 2009.

Contact details: Junko Otani, DDS, MPH, MS, PhD
Associate Professor, Graduate School of Human Sciences and Regional Director, East Asia Center for Academic Initiatives (Shanghai Office), Osaka University1-2 Yamadaoka, Suita, Osaka, 565-0871 Japan

Junko photo


Read selected reviews for Junko Otani’s book ‘Older People in Natural Disasters’, (2010):

Reviewed by Maggie Gibson, Canadian Journal on Aging, 2011, Vol.30, pp. 670-671.

Reviewed by THANG Leng Leng (NUS), Asian Anthropology, 2012,Vol.11, pp166-169.

Reviewed by Lori Peek and Lisa M. Brown, International Journal of Mass Emergencies & Disasters, 2012, vol. 11, pp. 166-169.