Research methods

The research methods used in this investigation were chosen to optimise the utility of the data collected, and in particular to mix methods to increase the possibility of effective triangulation, given the various challenges to reliability and validity that may be present. As a result a range of different methods were used and these are described here. All these methods describe data collection from families of the Missing, since most data was in this category. The final section discusses data collection from elites. Details of interviews and focus groups are tabulated in Appendix III.

Semi-structured interviews

Interviews provide the opportunity for subjects to present information on their own terms: an interview combines structure with flexibility. For a study such as this one it is an ideal technique to allow families of the disappeared to tell their stories and articulate their needs on their own terms. The interview is semi-structured, following a format prepared iteratively with family associations during the first weeks of the collaboration. The iterative process is crucial to ensuring that the priorities of families are prominent in the interview and that culturally specific elements, often initially invisible to outsiders, are included. The „script‟ for this interview was used as a guideline: the course of questioning was determined by the responses of the interviewee, and this used as a framework upon which a discussion with the families was hung, with the family determining the issues of greatest interest to them. A typical interview lasted around 90 minutes. The interview began with a general discussion of the circumstances of the family, the role of the disappeared person within it, and the nature of the disappearance. Families were then asked an open question: “What action would you like to see taken in response to the disappearance?” This allowed families to identify what they saw as their priorities, whether it be an answer concerning the fate, economic support, prosecution or something else. More detailed questioning concerning the various potential needs of families then followed this. Interviews often became a discussion with the entire extended family. This was positive, not only for the support it offers during what might be

141

an upsetting discussion, but also because it gives an insight into family dynamics, much like a focus group. The perception of needs has gender dependence: women will very often be most sensitive to economic needs, and needs of children, while men will be more likely to articulate political needs. Within a „family focus group‟ these dynamics can be explicitly probed. Traditional hierarchies would often mean that a certain member of the family (typically the father or the eldest son) would be presented as the principal interviewee. Since wives, particularly younger ones, were most likely to be impacted by social stigma where possible they were spoken to in private or with other wives of the disappeared, so as to best understand the social and family pressures to which they may be subject.

Focus group discussions

A focus group is essentially a group interview, with each participant given the chance to express himself or herself, but with the additional dynamic of inter-group discussion. The questions used to initiate discussion were very open, inviting participants to choose, and then discuss, the greatest problems they were facing as a result of the disappearance of their relative, with the ensuing discussion permitting detailed attitudes to emerge. The same script as used for the interviews drove such discussions. For individuals who may feel vulnerable a focus group can create an environment that is more secure for the expression of feelings, particularly where all members feel some solidarity. The most striking success of this technique was when wives of the disappeared in Nepal were invited to discuss their problems, and chose issues in the family and community that have not previously been widely articulated by conflict victims, and that were not heard in mixed or family groups (Section 6.2.3).

Participant observation

The traditional ethnographic method of participant observation was also used throughout the contact the researcher had with families of the disappeared. Given that in the two contexts the researcher met more than 300 families over a period of 2 years, as well as leaders of family associations repeatedly, there was an opportunity to collect a large volume of data. In particular, participant observation was an additional tool for triangulation, since it allowed the possibility to confirm or refute the verbal data gathered in interviews and focus groups discussions. During interviews, focus groups and throughout field visits field notes were taken of observations of participants and their environment that formed part of the data analysed.

142

Understanding elite attitudes

In both contexts, elite attitudes to transitional justice processes are a matter of public record. In Timor-Leste mechanisms are largely complete and there is both extensive documentation of those processes through the CAVR and CTF reports and trial records as well as an extensive commentary from the media, academics and concerned national and international agencies. The attitudes of public figures, such as the Prime Minister and the President, are also widely recorded. As such, in Timor-Leste there was little need to meet such figures to understand motivations and perceptions in relation to transitional justice process, and as a result few interviews were made. Most references in this thesis will thus be to publicly available documents.

In Nepal the situation is more complex since there has been no process of substance. Whilst political parties have committed themselves to the documents of the peace process (that include explicit transitional justice mechanisms) in practice all political actors are choosing either to distance themselves from those commitments or decline to discuss them publicly. As such, there was some value in meeting senior political actors to understand their perspectives. It was also necessary to understand what drives advocates of the rights discourse whose emphasis was hugely divergent from the priorities of the victim families met in this research. As a result senior figures from national and international rights agencies were met in Kathmandu and asked to comment on the results of this study in relation to their agency‟s priorities. This drove the analysis of Section 9.1. Political figures were also met to understand their rationale and perspectives concerning transitional justice processes. The Action Research outlook in both contexts also demanded that contact with political and rights actors was used as an opportunity to disseminate the results of this study (even while it was ongoing) and to encourage these decision makers to consider the attitudes of victims.

Validity

Whilst validity has traditionally been discussed with reference to quantitative analyses, it remains a relevant measure with which to understand the extent to which qualitative research is plausible, credible and trustworthy (Johnson, 1997). A significant literature has discussed the controversies arising (e.g. Whittemore, Chase & Mandle, 2001) even though many qualitative researchers perceive understandings of validity to be inherently positivist (Golafshani, 2003). In this study a range of approaches are taken to ensure validity and these include the use of low-inference descriptors (most notably direct

143

quotes from participants), both data and method triangulation, participatory methods and participant feedback after initial analysis. Underlying all of these approaches is a desire to maintain a commitment to reflexivity, “an awareness of the researcher's contribution to the construction of meanings throughout the research process” (Nightingale and Cromby, 1999: 228). Issues of both personal and epistemological reflexivity are eased considerably by the methodology foregrounding the views of victims, which encourages the researcher to background his own personal views. This approach appears vindicated since, despite the researcher‟s long engagement with both contexts, in both case studies extremely important new priorities were revealed by the methodology here that had remained unexposed by the less rigorous approaches that defined his previous work with families of the Missing.

The descriptive validity (Maxwell, 1992) of the data, the extent to which it truly represents what was said by participants, is confirmed here through the extensive use of verbatim quotations from the data. Interpretive validity (Johnson, 1997), in which one seeks to demonstrate that the meanings assigned to the words of participants are accurate, is confirmed in this study through not just the feedback of participants, but through their active participation in the presentation of results. When the data were published in the two contexts, families of the Missing were involved and enthusiastically welcomed the published reports as representative of their own needs.

Issues of internal validity refer to the degree to which any observed relationship is causal. In this study such issues are minimised since it implicitly assumes a causal relationship between the transitional process and the addressing of needs: everything that impacts upon the addressing of family needs is considered part of the „transitional process‟, and so the causal relationship is necessarily true. In practice, data collection will demand a study of all factors and if one that is not traditionally part of transitional mechanisms becomes important this can be flagged. This will be particularly relevant where there are variables present in one context that impact on the analysis that are absent in the other.

External validity refers to the extent to which studies in any one case study are

generalisable (or transferable) to others. Because the generalisation here will be analytical (i.e. through theory) rather than statistical, the limits of this will to some extent become apparent in the data. For example, the needs of families (theoretical proposition one, Section 1.4) will never be identical in different contexts, due to cultural artefacts, but

144

if they are very similar, then the generalisation of the proposition is favoured. Similarly, in the more difficult generalisation concerning proposition three, given that the transitional mechanisms concerned will be very different in each individual case study, the replication logic will be less compelling. Again however, the data should be able to reveal how and why needs were addressed in one case and not the other: this should inform the extent to which what we are seeing is an intrinsic consequence of the mechanisms concerned, or deriving from other variables.