Just as with attitude scales and surveys (in chapters 9 and 11), there’s no point in reinventing the wheel. Over the years, researchers have developed coding schemes for using direct observation in many different situations—in studies of interactions between married couples, in studies of teacher effectiveness, in worker-management negotiations, and so on. If others have developed and tested a good system for coding behaviors of interest to you, use it. Don’t feel that it’s somehow more prestigious or morally better for you to make up everything from scratch. Knowledge grows when researchers can compare their data to the data others have collected using the same or similar instruments.
Figure 14.3 shows the basic coding scheme for interaction process analysis, a system developed 60 years ago by Robert F. Bales in his research on communications in small groups (Bales 1950).
Categories for direct observation.
SOURCE: SOURCE: R. F. Bales, ''Some Uniformities of Behavior in Small Social Systems.'' In Social Interaction Systems: Theory and Measurement, p. 165. New Brunswick, NJ: Transaction Publishers, 2000.
Despite its age, the Bales coding scheme continues to be used in the study of classrooms (Koivusaari 2002) and work teams (Nam et al. 2009)—in fact, in any situation where people interact with one another.
Stewart (1984) audiotaped 140 doctor-patient interactions in the offices of 24 family physicians and assessed the interactions with Bales’s interaction process analysis. Ten days later, Stewart interviewed the patients at their homes to assess satisfaction and compliance. That is, were the patients satisfied with the care they’d gotten and were they taking the pills they’d been told to take? Sure enough, when physicians are coded as engaging in many patient-centered behaviors, patients report higher compliance and satisfaction.
One of the best things about the interaction process analysis system is that any act of communication can be identified as being one of those 12 categories in figure 14.3, and the 12 categories are recognizable in many cultures around the world. A more detailed outline for coding interpersonal relations was developed by Bales and Cohen (1979). A complete course on how to use their system is available in their book, aptly titled SYMLOG, which stands for “systematic multiple level observation of groups’’ (box 14.2).
DOING CONTINUOUS MONITORING IS HARDER THAN IT SOUNDS
Make no mistake about this: Continuous monitoring is tough to do. It takes several months of intensive training for observers to become adept at using complex coding schemes. In the 1920s, Rudolf von Laban developed a system of 114 signs with which to record dance or any other set of human movements (see Lange 1975). If you are trying to understand the meaning of complex human body movement, you've got to start with some way to record the basic data, like any other text. Brenda Farnell used Laban's script (known as labanota- tion) in her meticulous research on Plains Indian sign language (1995) and has written extensively about how the system can be used in the anthropological study of human movement in general (Farnell 1994, 1996; Farnell and Graham 1998).
These days, behavioral coding in psychology and ethology is being done with hand-held computers and software that lets you program any key to mean ''initiates conversation,'' ''reciprocates affect,'' or whatever. (See Ice  and Koster  for details about using this technology and see appendix E.) As it becomes easier for fieldworkers to observe and code behavior at the same time, I think we'll see renewed interest in continuous monitoring and in the use of complex coding schemes like labanotation.
One of the problems in the use of direct observation is the need for reliable coding by several researchers of the same data. We’ll take up measures of intercoder reliability in chapter 19 on text analysis. The problem of testing intercoder reliability is the same, whether you’re coding text or behavior.