Method Guide for Life History Interviews, Peter Davis

Peter Davis developed this life-history method while he was conducting research in Kushtia district in Bangladesh starting from 1999. The 2007 approach has moved the methods further on, spanned eight districts, and had the advantage of being nested as a subsample of a quantitative panel survey, allowing an integrated qual/quant analysis of findings. SDRI fellows Peter Davis and Bob Baulch are working together with Agnes Quisumbing from IFPRI on integrating these findings, some of which have been published (see publications).

For the life history interviews we decided from the outset to interview, whenever possible, one man and one woman separately in each household. There were four main advantages from interviewing two respondents per household:

1. It allowed immediate cross-checking (triangulation) on memories of key events. From the differences between the two accounts we got an idea of the accuracy of the information. Some kinds of events were recalled more accurately by both respondents, other types of information were more reliably recalled by one person but not the other. One of the key problems was quality of recall, and this helped us to judge recall accuracy of details, particularly of dates and family events.
2. It gave a gendered perspective on events, episodes, and processes. We found differences in emphasis between what affects women's well-being and what affects men's, and perceptions of these, and it was worth picking these differences up. Women tended to remember more about children, revealed more about domestic tension and family relationships, while men gave more accurate information on land purchase, sale, land areas, business and events where women were excluded.
3. A life history is a case study of an individual within a household or households. The unit of analysis (or case) is an individual and provides a useful contrast with a household survey where the household is the case. This can help to correct some of the biases introduced by using households as cases - e.g. towards seeing households as stable as long as the household head stays the same. It gives us an alternative view of instability, of household formation, growth, decline, splitting/dissolution, when we study individuals within households and especially if we interview more than one individual per household. Some of the life history is about life within a different household - for example when talking about a woman's life before marriage, or a man's life when he was single and possibly somewhere else.
4. Having two interviews per household was not necessarily twice the effort because a large part of the cost and time involved is in travelling, finding households, and arranging times to interview. If a team can interview two individuals in one household using one vehicle, it is quicker and cheaper than interviewing two individuals in different households in different locations. In Bangladesh it was better for men to interview men and women to interview women so a mixed-sex team could work well interviewing two individuals in one household at one time before moving to the next household. The mixed-sex team was also good for the community focus group discussions with women as well as men.

After the life-history interviews we allowed ample time (preferably on the day of interview) for the interviewers to write up the interview (in a format that had been formulated in the initial workshop and refined in the field in discussions with the analyst). In addition to a standard approach to writing up the interview we also encouraged free-writing of what was learned in the interview and this was kept a fieldwork diary (in addition to the more formally-agreed write-up structure) with reflective impressions and lessons-learned about methods included. These diaries were also translated and became a part of the qualitative data set.

During interviews we used small unobtrusive digital voice recorders. We used Olympus WS320 digital voice recorders which could be placed between the interview participants and were small enough not to ‘get in the way’ during interviews. These could be plugged straight into a laptop without a lead and ran from locally purchasable batteries - better in many ways than rechargeable batteries, as they can be quickly changed in the field. This type of digital recorder had more than enough memory for a study like this.We were conscious of avoiding 'wowing' or intimidating the participants with gadgets during interviews as we felt it would distract and hinder fostering a relaxed natural conversational style.

We did not attempt to write full transcripts of interviews because full discourse analysis was never our intention. The digital recordings were therefore useful to keep with the other case files on computer and were used for checking back on interviews for the initial same-day write-up in Bangla, for extracting quotations and for the final anonymised write-up in English. It was therefore important to decide in advance how the interview should be written up as this was the main data to be analysed from each interview. We found that a chronological account of the events was best, given that we were most interested in understanding events and trajectories, including the respondent's ideas of the significance of the events in their life, identifying causal mechanisms and encouraging counterfactual thinking. It was essential to give a lot of guidance and coaching to interviewers, including debriefing feedback at the start of the field work on how interviews should be written up using notes, diagrams and recordings taken during the interview - so as to 'create' (rather than 'collect') the kind of data that can then be used to help explore processes and answer research questions.

We spent about two weeks in each site in the life-history phase of the research. We located households first and checked availability of respondents. We then started interviewing with households where both members were available. It made sense to interview husband and wife separately at the same time, where possible, to avoid multiple trips to the same household. This was less disruptive for households and reduced travel time since we only had one vehicle. It was also better to interview them in such a way that they were not distracted by the other’s interview. When only one member was available we left that household till later, finding out when both members would be available. Also during the initial days we arranged a time for ‘knowledgeable people’ to attend a discussion meeting on the last day in the village. This was usually held in a school building or in a near village leader’s house. We tried to have a Union Council (parishad) member and a number of elderly people in attendance. These focus group discussions tended to be dominated by men so in most sites the female researchers also facilitated separate focus groups with women. Towards the end of the time in each site the team ended up working in different households which became complicated for arranging transport. When the analyst was with the team he sat in on one of the male or female interviews.

It was initially envisaged that we conduct about 100 life histories in 100 households. However we subsequently decided to carry out two interviews per household and although this doubled the number of life histories, it seemed sensible to maintain a gender balance, and be able to triangulate and compare different versions of events from the perspective of two household members. When a household member wasn’t available – which was more often a man than a woman, especially during the main April-May rice harvest – or when only one member was an adult, we did one life history. In the end we did 293 life history interviews in 161 households.

We had a team of 4 interviewers, a supervisor and a driver. At night we stayed as close to the sites as possible, often in the accommodation rooms attached to NGO offices or government guest houses. Peter Davis supervised the team in the field, sitting in on interviews and providing on-the-job guidance for the first six weeks (April-May 2007) and six weeks towards the end of the fieldwork (July-August 2007). The fieldwork lasted from March 2007 to the end of October 2007.

Respondents were often husband and wife, but in some cases, such as when one partner had died, we interviewed one parent and their son or daughter. In general we tried to interview people who were older than 25 and used historical markers, such as the 1971 war of independence and 1988 floods, to determine the years particular events described by the respondents occurred. At the end of the life history interviews, the researcher who facilitated the interview drew a diagram of respondent’s life history from the time-line of events that he or she had drafted during the interview, with the help of the respondent. The other researcher wrote up the life history in written form based on the notes she had taken during the interview. Only two interviews were conducted per day to allow the diagram to be finalised and interview to be written up in Bangla on the same day.

Life history diagrams were then traced, translated and anonymised, using inkscape, so as to be made available in SVG format with the final anonymised and edited life histories. On these diagrams, the level of wellbeing at different points in the life trajectory is indicated using a scale of one to five using the categories described in Table 1 below, based on life-conditions described by the respondent. These levels were checked during a final village level discussion with people who knew the households well, and then were finally written onto the life history diagrams during a round-table discussion among the researchers who had carried out the life-history interviews and facilitated the village discussion groups. In round-table discussions all information about households and members was used, and levels of wellbeing were agreed by consensus after discussion. These discussions were also digitally recorded – creating another data source about the households which could be coded and analysed in nVivo8.

Table 1: Qualitative well-being levels for individuals

A cover sheet was used at the start of the interview for basic data. The coversheet included a consent form which was signed by the interviewer – we found it was not appropriate to ask participants to sign this form as it aroused suspicion, especially among those who could not read. At the start of the interview the purpose of the research and use of the data was clearly explained, and permission was sought to record the interview.

An interview guide/ checklist was developed during the initial training workshop and an English translation of it appears in Table 2 below.

Table 2: Interview guide/ checklist for the life history interviews

General topic area	Particular events and issues
Family life history	Marriage dates (dowry, wedding costs), births (children etc), deaths (parents, siblings, children, spouse), separation of property.
Education and training history	Own education, what level, reason for leaving education, sibling’s education, children’s education
Employment history	Dates of jobs, who helped to get a job, business starting-finishing, assets bought or sold, how were assets bought (loans, savings, mortgage, sale of land etc), promotions, loss of job, migration for work (national, international)
Asset and loans history	Land (bought, sold, lost, mortgage, leased); livestock; buildings (house, shop); pond; materials (bamboo, bricks, tin); trees (when sold and why); jewellery; furniture; cooking utensils; loans, savings.
Migration and place history	Reasons for moving (marriage, employment, security, other), family members moving
Health and illness history	Chronic illnesses, illness before death of relatives, accidents, medical expenses
Identity and membership history	NGO samiti, ROSCA, neighbourhood groups, kinship (bangsho), religious groups, labour union, political group, who do they celebrate Eid or other festivals with, samaj, who are ‘amader lok’?, who helps in times of crisis?
Crisis and coping history	Crisis: dowry, illness, flooding, crop loss, livestock losses, business loss, unemployment or job loss, divorce, court cases, land and property division, migration, death of family members, accidents, injuries, cheating, theft, violence, threats, intimidation, extortion, conflicts and disputes, loans In these crises who helped and why? Coping: Forms: sales, loans, savings, labour, business, mortgage, informal help, local collections, religious charity, begging, common property, divorce, migration, marriage, child labour, sending children away, crime Channels: own resources, kinship, friends, employers, neighbours, community groups, NGOs, public programmes, political leaders, mohajan, mastaans
Opportunities and improvements history	Job, land (bought, gift), dowry, remittances, loan, Govt. programme (VGD, VGF, boyoshko bhatta, bidhoba bhatta, mukti juddho bhatta) pension, provident fund, savings, son working, daughter working
Additional contextual information to look for	How social structures (roles, values, norms, sanctions) have constrained or enabled people’s agency (choices, options, opportunities) How endowments and circumstances (economic, health, status, education, memberships) have interacted with events and episodes (crises and opportunities) How sequences or combinations of events have combined How life cycle position is important
Historical markers to use	National events: 1947-partition, 1965-Indo-Pak war, 1971-Independence, 1975-Sheikh Mujib’s death, 1981-Zia Rahman’s death, 1988-flood, 1990-protest and end of Ershad era, 1998-flood Local events: floods, droughts, roads built, electrification, schools built etc.

During the life-history (phase 3) fieldwork in each village at least one (sometimes more) additional focus group discussion was carried out in order to map a history of the village. The ‘development’ of the village was mapped on a trajectory diagram similar to the individual life history diagram, recording important village events, periods of opportunity, perceived causes (new crops, irrigation, new roads, electricity etc.) and community level shocks (storms, floods, crop failure, fertiliser shortages etc.).

From the life history research the following types of data were created:

Table 3: Forms of data in the life history research