Observational techniques, a cornerstone of the qualitative research paradigm, can be divided into two main categories: participant and naturalistic observation. Naturalistic observation is a method of collecting information in a setting in which the behavior of interest occurs, typically unbeknownst to the targets of observation. Naturalistic observation is often used by ethnographers examining cultural behavior, organizational development researchers, and program evalu-ators. The hallmark of naturalistic observation is the lack of intrusion by the researcher into the setting and behavior of interest. An example of naturalistic observation would be a training program evaluator watching the content of the training and participant observations through closed-circuit television to assess comprehensiveness of the training program. In this example, the participants are not aware of the observer and, as such, do not shift their behavior to make a favorable impression.
Participant observation is the other broad category of observational techniques and can take one of three forms:
- Complete participant. The researcher conceals his or her role to more fully examine the issue of interest.
- Observer as participant. The role of the researcher is known to those being observed.
- Participant as observer. The research function performed by the observer is secondary to his or her role as a participant in the actions and behaviors.
In all of these cases, the researcher’s role as an observer is overt and may influence the behaviors of those being observed.
Key Elements for Conducting a Naturalistic Observation Study
As with any research design, decisions regarding the specific methods and scope of the study must be made. However, a number of decisions are unique to observational research:
- Level of involvement of the researcher/observer depends on the nature and sensitivity of behaviors to be observed. Naturalistic observation is better suited to settings in which a researcher’s presence might change the behavior (e.g., Western Electric employees in the Hawthorne studies). Participant observation may be more suited to those situations in which actively performing the behavior of interest lends a higher level of understanding to the researcher (e.g., understanding job content in the process of job analysis).
- Amount of collaboration in the coding process depends on the setting of the observation (e.g., is it feasible to involve multiple coders?). The inclusion of two or more coders allows the research team to assess the interobserver reliability of their observations.
- Length of time for observation of behaviors depends on the goal of the study. There are three general sampling frames for observational research: (a) Time sampling, in which the behavior is observed for a set period of time (e.g., 3 days); (b) point sampling, in which one individual or group is observed before moving on to the next individual or group of interest (e.g., observing the training department in one plant before moving to a different plant); and (c) event sampling, in which an event is observed every time it occurs (e.g., observing the administration of annual performance appraisals).
- Focus and nature of the observation depends on the specific research question. The narrower the focus, the easier it will likely be to code the behaviors reliably and efficiently using a structured approach. Broader, unstructured context examinations (e.g., narratives) may be better suited for descriptive studies in the nascent stage of a research area.
The design of a naturalistic observation study relies on clarity in goals and measurement to target the behaviors of interest to adequately address the research questions. Because the behaviors of interest are occurring in their natural context, the researcher must be able to adeptly classify and interpret only relevant behaviors. For this reason, operationalization of behaviors is critical. The creation of a comprehensive coding sheet and training on the coding of behaviors is critical to ensure that behaviors are accurately captured.
Depending on the type of data collected during the observation period, analysis remains qualitative or turns to a more quantitative approach. If structured observation was used with behavioral checklists, descriptive statistical information may be compiled (e.g., frequencies). If the focus was broader and unstructured, narrative reviews may tell the story of the behavior in context.
Reliability and Validity of Naturalistic Observation
Inter observer reliability is the only way to assess the consistency of behavioral coding in a naturalistic observation study. For this reason, it is critical to include multiple coders in an effort to demonstrate a lack of observer bias. However, calculation of inter-observer reliability is only feasible when using structured and systematic observation. If an unstructured method is used, demonstration of consistency in observation is very difficult.
In terms of validity, naturalistic observation sacrifices internal for ecological validity. Because the behaviors observed occur in context with no interference from researchers, the extent to which the behaviors observed in this naturalistic setting mimic the “real world” is very high. However, because the observer does not interact with or participate in the context of behavior, there is no control of potentially influencing variables. To link internal and ecological validity, an observational study may be used to replicate and/or extend findings from a laboratory study on the same topic.
Advantages and Disadvantages of Naturalistic Observation
The benefits of naturalistic observation research are numerous. Specifically, naturalistic observation allows researchers to examine behaviors directly in context without interference, thus providing a foundation for understanding the environmental conditions associated with the issues of interest. This is especially true when the topic under investigation is very sensitive or the presence of the researcher would likely influence behavior.
Beyond the time commitment required to conduct this type of research, the disadvantages of naturalistic observation center on three main issues: reliability, validity, and ethics. Because observation is an inherently perceptual process, bias can be introduced in the coding and interpretation of observational results in a number of ways (e.g., coding the behavior, interpreting results). Therefore, some question the accuracy of observation records (e.g., checklists, narrative reviews) and disregard their value as contextual markers.
As previously discussed, because the observer does not influence the situation, the influence of extraneous variables on the behaviors of interest cannot be assessed (e.g., no internal validity). The real-world aspect of the data may be high, but without the knowledge and/or control of influential variables affecting these behaviors, the value of the research may be limited.
Because of the unobtrusiveness of the researcher in naturalistic observation, informed consent cannot be given by participants. Under American Psychological Association (APA) guidelines, informed consent procedures do not need to be initiated when there is no expectation of harm to the participant as a result of the research.
However, the application of this standard has been controversial (e.g., R. D. Middlemist and colleagues’ study of personal space invasions in the lavatory). Institutional review boards should provide guidance in the interpretation of waived informed consent for naturalistic studies.
- Creswell, J. W. (1994). Research design: Qualitative and quantitative approaches. Thousand Oaks, CA: Sage.
- Denzin, N. K., & Lincoln, Y. S. (2000). Handbook of qualitative research (2nd ed.). Thousand Oaks, CA: Sage.
- Middlemist, R. D., Knowles, E. S., & Matter, C. F. (1977). Personal space invasions in the lavatory: Suggestive evidence for arousal. Journal of Personality and Social Psychology, 35(2), 122-124.
- Patton, M. Q. (2002). Qualitative research and evaluation methods (3rd ed.). Thousand Oaks, CA: Sage.