5.2 Research Design
5.2.4 Material and Sources
The detailed evidence that is necessary to detect every step in a theorized causal mechanism and thus to conduct within-case analysis through process-tracing is often difficult to obtain. This is especially so in conflict settings where warring parties may refrain from stating their true intentions in order to keep ahead of an adversary; where victims may be too frightened to speak the truth due to fears of reprisal or revenge; or where official documents may get lost or are purposefully destroyed (cf. Bakke, 2014). UNESCO, for instance, lists deliberate war-related violence as one of the key threats to “world documentary heritage” (Hoeven and Albada, 1996). Therefore, and similar to the importance of being aware of validity issues and selection bias when gathering statistical conflict data through the evaluation of news reports (cf. Chapter 1 and Kreutz, 2015; Öberg and Sollenberg, 2011; Themnér and Wallensteen, 2014), it is vital to discuss possible weaknesses in the data underlying my qualitative, comparative analysis (George and Bennett, 2005).
In a first step, and in order to arrive at detailed day-by-day information ac- count on events leading up to, during, and after the respective interim periods in Nepal, Angola, and Cambodia, I conducted an initial search on news reports in the LexisNexis database on the situation in the three countries. Due to the overwhelming amount of material, I concentrated my search on news reports published by the two largest global news agencies that hold offices in the major- ity of countries around the world – the Associated Press (AP) and the Agence France Press (AFP). I also included articles published by a few internationally renowned newspapers that had sent their own foreign correspondents to the respective conflict zones, such as Victoria Brittain reporting for The Guardian in Angola (e.g. Brittain, 1992a; Brittain, 1992c), Somini Sengupta reporting for
The New York Times from Nepal (e.g. Sengupta, 2006a; Sengupta, 2006b), or
Terry McCarthy (1991) reporting for The Independent from Cambodia. Con- fining my search that way resulted in 2892 documents published for the search term “Nepal” between 01 January 2006 and 31 December 2008; 2258 docu- ments published for “Angola” between 01 January 1991 and 31 December 1992; and 2673 documents published for “Cambodia” between 01 January 1991 and 31 December 1993. For Nepal, I additionally relied on three English language newspapers published by outlets in Kathmandu – the Nepali Times, The Hi-
malayan Times, and The Kathmandu Post (cf. section 5.3) – and for Cambodia,
I made use of the online archives of The Phnom Penh Post.
Figure 5.2: Associated Press Reporting on Angola 0 10 20 30 1991−01 1991−07 1992−01 1992−07 1993−01
Notes: Figure based on a LexisNexis search of AP publications on the search term “An- gola.” I excluded all articles that contained the term “Olympics,” because Angola’s par- ticipation in the 1992 Summer Olympic Games in Barcelona resulted in a large number of hits for the search term in July and August 1992; hits that were unrelated to the situation in Angola at that time. The interim period is shaded in gray.
tative conflict data through reporting as discussed in Chapter 1, not least in that foreign correspondents are likely to be based in the national capital. This bias is also detrimental for the purpose of this dissertation, because it could result in under-reporting of events in the periphery, in particular events and actions concerning parallel political rebel government structures (although UNITA, for instance, regularly invited journalists to its parallel-held territories and the rival capital of Jamba, cf. Chapter 7 and AFP, 1991c). Furthermore, news reports come with a temporal bias in that they tend to focus on “big events” of interest for readers in Western societies. This includes in particular the signing of peace agreements and the holding of elections that terminate an interim government. I illustrate this temporal bias using the example of AP reporting on Angola between January 1991 and December 1992, as visualized in Figure 5.2. The graph displays the number of AP articles containing the search term “Angola” aggregated by week, with the interim period shaded in gray. The graph shows that the largest number of articles was published in the last week of May 1991 (33 articles in total), which is when the warring parties signed the Bicesse Peace Agreement in Portugal. The second largest spike in Figure 5.2 is displayed in the last two weeks of September and the first two weeks of October 1992 (be- tween 19 and 26 articles per week), which is when the elections terminating the interim period were held.
I also found that news reports were more likely in providing information on aspects relating to power-sharing or international interim government, or
with regard to the disarmament and demobilization of warring parties, while they rarely report details on the role of civil society in interim decision-making. Finally, news reports can of course be biased in how they provide information on the respective warring parties engaged in a civil war, and some journalists may be more supportive of one side to the conflict than of the other. For instance, Heywood (2000) and Windrich (1992) discuss how UNITA leader Jonas Savimbi managed to gather a significant amount of overly sympathetic support for his strive against Angola’s MPLA government, as reflected in the style of reporting by Western journalists. Becker (1998) similarly reflects upon her own interpretations in reporting as a correspondent on Cambodia’s civil war.
In order to deal with these issues, I rely in my case studies on a broad mix of sources to assess the statements and actions of parties in each interim gov- ernment under analysis; and while I expect this triangulation of sources to be a remedy for some of the problems of news reports, each of the additional types of sources of course suffer from their own limitations. In particular, I complement news reports with secondary political science, historical, ethnographic, and area studies academic literature (that is however often not detailed enough to trace very particular mechanisms). I also consult a variety of primary sources, such as policy reports (e.g. Human Rights Watch, 2015b; International Crisis Group, 2005) and official documents (e.g. UN Security Council Resolutions). Policy re- ports are insofar often biased in that they tend to focus on issues of interest to Western policy makers, while official documents may be biased in that confiden- tial documents with particular important insights may be impossible to obtain. I also rely on personal memoirs (e.g. Anstee, 1993), transcripts of speeches and radio broadcasts (e.g. in Weimer and Fandrych, 1995), blog posts and websites (e.g. Nepali Congress, 2015; United We Blog, 2006), as well as “gray literature” like party manifestos (e.g. Communist Party of Nepal (Maoist), 1993; Embassy of Democratic Kampuchea, 1976) or letters to newspapers by warring party rep- resentatives (Primo, 1992; Savimbi, 1989). Particular sensitive aspects of gray literature published by the parties may thereby get lost throughout the course of the war, especially so if there is one clear winner or loser. Finally, my collec- tion of information on the qualitative case studies is limited because I have to rely on sources published or translated into English or German – UNITA, for instance, published German versions of their party programs (National Union for the Total Independence of Angola, 1988; National Union for the Total Inde- pendence of Angola, 1991) – and I thus cannot reconstruct events only reported in Portuguese, Khmer, or Nepali.