Log in / Register
Home arrow Environment arrow Research Methods in Anthropology: Qualitative and Quantitative Approaches

Coding and Recording Time Allocation Data

Sampling is one of two problems in TA research. The other is measurement. How do we know that when Oboler recorded that someone was ‘‘working,’’ we would have recorded the same thing? If you were with Johnson when he recorded that someone was engaged in ‘‘hygiene behavior,’’ would you have agreed with his assessment? Every time? You see the problem.

It gets even more thorny. Suppose you work out a coding scheme that everyone agrees with. And suppose you train other observers to see just what you see. (Rogoff [1978] achieved a phenomenal 98% interobserver agreement in her study of 9 year olds in Guatemala.) Or, if you are doing the research all by yourself, suppose you are absolutely consistent in recording behaviors (i.e., you never code someone lying in a hammock as sleeping when they’re just lounging around awake).

Even if all these reliability problems are taken care of, what about observation validity? What do you do, for example, when you see people engaged in multiple behaviors? A woman might be holding a baby and stirring a pot at the same time. Do you code her as engaged in child care or in cooking? (Gross 1984:542). If someone saw that you were lying down reading and you were studying for an exam, should they record that you were working or relaxing?

Do you record all behaviors? Do you mark one behavior as primary? This last question has important implications for data analysis. There are only so many minutes in a day, and the percentage of people’s time that they allocate to activities has to add up to 100%. If you code multiple activities as equally important, then there will be more than 100% of the day accounted for. Most TA researchers use their intuition, based on participant observation, to decide which of the multiple simultaneous activities they witness to record as the primary one and which as secondary.

The best solution is to record all possible behaviors you observe in the order of their primacy, according to your best judgment at the time of observation. Use a check sheet to record behaviors. Use a separate check sheet for each observation you make. This can mean printing up 1,000 sheets for a TA study, and hauling them home later. You can save yourself a lot of work by using a hand-held computer with the equivalent of an electronic check sheet installed that lets you watch behavior and code it on the spot (see appendix E) (box 14.4).

BOX 14.4


Saving work is one of two good reasons to use hand-held computers to record observational data. The other good reason to give up paper check sheets if you can is that, eventually, you have to type in the data anyway, transferring them from paper to computer so you can run statistical analyses. Each time you record behavioral data—as you watch it and as you transfer it from paper to computer—there is a chance of error. That's just how it is. So eliminating one step in the data management process cuts down on errors in data.

If you must hand-code your original observations, enter the data into a laptop while you're still in the field as a precaution against loss of the original data sheets. Be paranoid about data. Those horror stories you've heard about lost data? They're true. (Read M. N. Srinivas's account [1979:xiii] of how he lost all three copies of his field notes, compiled over a period of 18 years, in a fire at Stanford.)

Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
Business & Finance
Computer Science
Language & Literature
Political science