2.2 Frequently seen errors in human rights information management system design

2.2.1 Overview

When an organization begins planning an information management system, there are two overwhelming temptations which must be avoided: the first is deciding only to represent the simplest imaginable form of a given case, the second is trying to represent every imaginable detail of every aspect of each case. Both extremes must be avoided as the organization decides how extensive their information management system will be. The dangers of oversimplification are presented below in 2.2.2 and 2.2.3. If the organization becomes obsessed with recording every possible detail, organization members will be overly burdened by the effort required to record all the information required by this level of detail. Each additional detail means another question in the interview, another chunk of information the data processors need to code, and another section in the database. The organization must clearly define what information is really important to its mission, and the information management system should reflect those definitions.

Consider a trivial example of a problem that comes of too little detail: the organization is deciding what information is needed from witnesses (deponents, complainants, etc.), and someone asks if the witness' date and place of birth is needed. One person points out that this information might help the organization later to distinguish between this witness and another individual with the same name. This second individual is very unlikely to have been born at the same time and in the same place: if the organization has recorded date and place of birth for all people, it will be easy to distinguish between people even if they have the same name.

Later, however, someone asks if the organization should query witnesses about number and size of the rooms in their houses. This information might be used to analyze the wealth of the witnesses relative to the wealth of the country (or province) more generally in order to estimate the representativity of the witnesses with whom the organization has spoken. Put differently, are the witnesses the organization has seen wealthier, roughly the same wealth, or poorer than the community in which they live? If the witnesses are system-atically different from their community, this might tell us something about either our technique for finding witnesses, or about the people who suffered in this community (and thus about the repression itself). This information might be more difficult to obtain (many questions may be required for interviewers to elicit the information necessary), and once obtained, difficult to represent usefully (how many rooms? how large are they? what covers the floors, walls, etc.?). This information may be more difficult to collect than will be worthwhile when the organization creates a report, or it may not be. I do not mean to argue against demographic or wealth analyses of deponents versus the community more generally, instead I want to raise the issue of the relative cost of information versus the relative worth of that information to the organization. Each organization has to decide for themselves what information is really important to them, and how much time they are willing to commit to gathering and processing it.

The following sub-sections will present two extended examples of errors that derive from oversimplification. These examples are not hypothetical -- I've worked with several human rights organizations that have had several variations of these problems. Although these examples are presented as database issues, problems created by over-simplification can occur in the data collection, data coding, or data-base steps of the information management system.