Russel-Research-Method-in-Anthropology

Direct and Indirect Observation 419
a running interview with the shopper, probing for details. Murtagh was aware
of the potential for reactivity in his study. But he was interested in understanding
the way people thought through ordinary, everyday arithmetic problems,
and his experiment was a good way to generate those problems under natural
conditions.
Many CM researchers record their own observations. It’s less tedious than
writing; it lets you focus your eyes on what’s going on; it lets you record
details later that might be left out of an on-the-spot written description; it
avoids the limitations of a checklist; and it lets you get information about context
as well as about the behavior you’re studying. Moreover, you can easily
transcribe your recorded observations, once you’ve got your voice recognition
software trained (see above, chapter 9, and appendix F).
But there are trade-offs. If you want measurements from qualitative data
(like running commentaries on tape), you have to code them. That is, you have
to listen to the tapes, over and over again, and decide what behaviors to code
for each of the people you observe. Coding on the spot (by using a behavioral
checklist or by inputting codes into a handheld computer) produces immediate
quantitative data. You can’t code and talk into a recorder at the same time, so
you need to decide what kind of data you need and why you need them before
you choose a method.
If you are trying to understand a behavioral process, then focus on qualitative
data. If you need measurements of how much or how often people engage
in this or that behavior, then focus on quantitative data. And, as always, who
says you can’t do both?
Coding Continuous Monitoring Data
Go to a shopping mall and record the interaction behavior of 30 motherchild
pairs for 5 minutes each. Record carefully the number of children each
mother has and her interaction with each child. Try to find out whether interaction
patterns are predictable from: (1) the number of children a mother has to
cope with; (2) the ages of the children; (3) the socioeconomic class or ethnicity
of the family; or (4) some other factors.
This exercise is instructive, if not humbling. It’s a real challenge to code for
socioeconomic class and ethnicity when you can’t talk to the people you
observe. Do this with at least one colleague so you can both check the reliability
of your coding.
In hypothesis-testing research, where you already know a lot about the people
you are studying, you go out to observe armed with a coding scheme
worked out in advance. The idea is to record any instances of behavior that