3.1.2 The concept of representation

In the discussion, I'll use the term "representation" to focus attention on the effects of changing the form of the information as it flows from step to step in the information management system. For example, an interview with a deponent is conducted in more or less normal conversation. What the organization receives from the interview, however, is quite different. Information flows into the interview in human conversation, but information flows out of the interview into the next step on the completed questionnaires. Thus the form and quality of the questionnaires determine what the organization learns from the interview. The interviewer herself may learn a very great deal from the deponent, but if the questionnaire fails to capture the richness of the deponent's testimony, then the organization cannot use it.

Before I begin the discussion of the steps in an information management system, a brief example may help to highlight the importance of representation. Imagine a human rights organization to be conducting an analysis of arbitrary execution, arbitrary detention, and torture. Unfortunately, the questionnaire designers neglected to leave space on the questionnaire for the possibility that all three violations happened to the victim being discussed in the interview. When the deponent (male) says to the interviewer (female) that he has come to tell of the murder of his son, the interviewer faces a conundrum. Because the interviewer understands the purpose of the investigation, she asks if the son was also detained and/or tortured, and the deponent answers yes, both violations occurred. However, the interviewer has a faulty questionnaire and therefore has no way to represent the information about the detention and the torture that has emerged in the conversation. Of course the interviewer may write this additional information in a notes area, or scribble it on the back of the page. However, information gathered so arbitrarily cannot be comparable from one interviewer to the next. The data processors who come after the interviewers in the information management process will have no way of knowing if all the interviewers asked about the two violation types that are not mentioned in the single spot available. It is therefore impossible to compare this interview's rich information against the sparse information in the more formally conducted interviews. Information jotted in the page margins is essentially useless for anything except anecdotal remarks about this particular case.

Thus the form of the questionnaire itself determines what the organization can know. The example in the previous paragraph began with a problem with completeness: the interviewer could not represent on the questionnaire all the information that the organization will need for their analysis. As the interviewer tried to solve the problem in an ad-hoc manner, she created a problem of comparability because the organization cannot be sure that each interviewer created the same ad-hoc solution. Each step of the information management system has analogous pitfalls. Data processors only see the part of the deponent's testimony represented on the questionnaire. Data entry staff only see the part of the questionnaire represented in the controlled vocabulary coding. And the data analysts who interpret reports only see the parts of the database which are collapsed into tables, graphs, and case summaries, thus they are depending on the representation made by the report designer. In the discussion of the parts of the system, it is worth keeping in mind how sensitive each step is to the quality of the work at all previous steps.

| go to next page, Chapter 3.1.3: Parts of an information management system | jump ahead to Chapter 4: Planning the project |

| go to top of this page | go to previous page, Chapter 3.1.1: Overview of information management system |

| return to Table of Contents |