Another source of information about human rights events is documents of any kind. One example of documents that played a key role in a human rights campaign were the files on the Salvadoran military kept by the U.S. Defense Intelligence Agency. Most of the people who reported human rights abuses to Salvadoran NGOs identified the perpetrators of Salvadoran human rights abuses by naming the military or security force units responsible. Under Salvadoran military law, the commanding officers of units that commit abuses are responsible for the abuses. Thus by linking the violations committed by a particular unit with the officer who commanded the perpetrating unit during the time the violation was committed, an organization can build individual human rights dossiers on each officer. This kind of information proved to be a powerful presentation of how the perpetrators of human rights abuses are organized.
The information about the personnel in the Salvadoran military in this case came from documentary research. U.S.-based researchers at the National Security Archives (NSA) filed Freedom of Information Act suits against U.S. intelligence agencies. Ultimately the information was processed and represented in a database by a human rights and refugee services NGO in Los Angeles called El Rescate. The El Rescate team designed databases and programs that linked the career information from NSA to violation information coded from the annual reports of the Roman Catholic Archbishop's human rights office, Tutela Legal del Arzobispado (TLA). El Rescate presented their reports to the Ad-Hoc and Truth Commissions in 1992, where the reports had very substantial influence on the Commissions' findings.
Using the career information given to us by El Rescate, I worked with a team at the non-government Human Rights Commission of El Salvador (CDHES) during May-November, 1992. We did data processing and programming to calculate human rights violation dossiers based on some 6,359 CDHES interviews conducted 1979-1992. The CDHES materials were also presented to the Ad-Hoc and Truth Commissions. Some of the CDHES tables are presented in the Section 3.5.4 below.
In another example, researchers from the Haitian National Com-mission for Truth and Justice (CNVJ) and the Argentine Forensic Anthropology Team (EAAF) worked through morgue records at a large hospital in Port-au-Prince looking for the names of people who died violently before, during, and after the period of the de facto military regime (September, 1991 - October, 1994; Doretti and Cano, 1995). This information becomes very useful as a complement to the interview-based data in an organization's analysis. Jumping off from the morgue work, at the CNVJ, we compared the monthly trends of numbers of violent deaths found at the morgue with monthly trends in numbers of murders found in the CNVJ interviews of some 5,500 Haitians. The two series correlated quite strongly. The closeness of the two series implies that they measure the same phenomenon, probably political violence. The strong correlation also validates both series' representativeness in time, since biases affecting two quite different measures are unlikely to affect data in the same periods. Thus each series is more credible because of the other. See Section 3.5.5 for further discussion of this method.
In the first example, the documentary evidence (from NSA) was used (by El Rescate and the CDHES) in combination with data gathered directly from victim interviews (done by TLA and CDHES). In the second example, the documentary work was done in one project (the forensic research in the CNVJ), and a second project in the same organization (the interview analysis team in the CNVJ) used the documentary work for a validation of both projects. Other documentary research projects may be in parallel to or less closely related to other data collection efforts (e.g., Snow and Bihurriet 1992), or presented entirely on their own (Pion-Berlin and Lopez, 1991).
If the organization wants to integrate information found in documentary sources with either information coming from interviews or with information based on physical evidence, or if the organization wants to provide some sort of central indexing of the ideas, events, or people covered in particular documents, then coded representations of these documents should be included in the information management system. As with all data collection, the organization must decide what information they want to extract from the documents in question. By developing a standard set of "questions," the organization has defined what is potentially relevant in each document. In this case, researchers record the answers to the questions on the protocol as they read each document, rather than as they interview an individual.