Observational Data

Methodological Questions Related to Observation

Researchers who seek to evaluate ongoing human behavior are confronted with a series of methodological questions, namely, whom to observe, when to observe, where to observe, what to observe, and how to observe. The question of whom to observe involves participant sampling and is closely tied to [Page 1534]what population findings are intended to generalize. The questions of when and where to observe have to do with time and place. Recording behaviors at inopportune or unrepresentative times challenges the whole nature of observation, while observing participants in surroundings where they are normally found often yields ecologically valid results. Less common or realistic situations, such as the laboratory, may reflect less valid representations of participant behavior or less valid representations of behavior in those situations. In naturalistic observation, participants are asked to behave as they usually do and to disregard the observer or recording instrument. Determining answers to the questions of what and how to observe varies because they are governed by conceptual, theoretical, practical, and ethical considerations.

Recording Technology

Observational coding can be done in the real time, or audiovisual records can be made using digital recording technology, which is now widely available and increasingly user-friendly. However, digital recording is sometimes unnecessarily detailed and requires a significant investment of time and resources. The presence of an observer/recorder can be intrusive and may represent a kind of novelty that evokes atypical responses from those observed, a phenomenon termed reactivity. This simply means that observation may promote socially desirable or appropriate behaviors and suppress socially undesirable or inappropriate behaviors. For example, adults may display higher than normal rates of positive interactions with children.

Coding Systems

A coding system with formal rules must be developed in order to obtain information from the stream of behavior. Such rules typically specify a set of behaviors, how the information is to be recorded, and how the resulting data are to form scores that can be analyzed.

Development of a coding system normally involves extensive observations, collaborative discussion, narrative accounts made in the field, field testing, and refinement. Initial unstructured, descriptive, qualitative data are shaped into more structured observations and, ultimately, quantitative data. Next, formal operational definitions of behaviors are developed to facilitate coder accuracy and consistency. These definitions typically possess three main characteristics: (a) the definitions are objective and refer to directly observable behaviors; (b) the definitions are clear, unambiguous, and easily understood so that coders can accurately comprehend and follow them; and (c) the definitions require little or no inference. Some systems use mutually exclusive and exhaustive codes: Exclusivity applies when only one code can be recorded for each time unit of observation, and exhaustiveness means that some code must be utilized for each observation unit. Other rating systems indicate the presence or absence of the behavior or the intensity or frequency of the behavior; they may rate behaviors on a scale as well.

...