It is not uncommon to find people who think that being present when a behavior of interest takes place provides adequate information for an analyst to understand that behavior. For example, many developers of computer systems feel that watching people using computer applications is the ultimate engagement with users.

Because human activity usually happens more quickly than can analytic understanding of it, a single encounter cannot provide a basis for deep understanding. Also, memory limitations will not permit recalling all that happens in even a short spate of activity, and notes cannot be made fast enough to record observations. Further, any notes taken must be made using some analytic perspective; such perspectives can comprehend only a part of what was understood at the time of recording, and even less of what might be understood upon reflection.

So ethnographers use videotape to capture a record of activity. Such records, although limited in physical and temporal scope, provide a record of activity that is not filtered through a human perspective and understanding of the activity. These 'unprocessed' records provide a more neutral basis for analysis of the recorded activity.

Further, videotape records can be viewed repeatedly, providing multiple exposures to the activity. They may also be viewed at different speeds (slow and stop-action for dense activity; fast to obtain patterns). As a result a much richer analysis of activity can be achieved.

This analysis can then be used to identify other events in the activity that the analysis should account for. The analysis can be checked against the events for correspondence to the predictions of the analysis. This validation viewing, usually again entailing repeated viewing, is a second mode of viewing the videotape records.

Any analysis is obtained under a perspective that the analyst brings to the analysis. This raises the concern that the analyst's perspective may significantly affect the outcome. There are two ways (at least) for addressing this concern:

First, other analysts can be invited to also view the videotape records, and asked to either critique the analysis in hand or produce their own analyses for comparison. Such communal viewing and group analyses are at the heart of the Interactions Analysis Laboratory (IAL) analytical method.

Second, the videotape records may be viewed in a third viewing mode: evaluating the videotape records to determine where the proposed analysis does not account for all the material on the videotape record. Such unexplained activity can suggest ways in which the perspective used to create the analysis is limited or flawed, and can lead to proposals of alternative perspectives under which to analyze the tape.

Either of these methods for improving analytical perspective can lead to a further round of analysis, validation, and evaluation. This analytical cycle is Interaction Analysis (IA).

This talk illustrates these ideas with videotape (made in 1981) of people making copies with a Xerox copier. It shows how successive rounds of analysis/evaluation work lead to successively better understandings of the activity, and raises questions about convergence and completeness.

© 2002 Noldus Information Technology bv