Observation is one of the few options available for studying records, mechanical processes, small children, and complex interactive processes. We can gather data as the event occurs and can come closer to capturing the whole event than is possible with communication. On the other hand, we have to be present to catch the event or have some recording device on the scene to do the job.
Observation includes a variety of monitoring situations that cover nonbehavioral and behavioral activities.
The strengths of observation as a data collection method include: • Securing information about people or activities that cannot be derived from experiments or surveys. • Avoiding participant filtering and forgetting. • Securing environmental context information. • Optimizing the naturalness of the research setting. • Reducing obtrusiveness.
Observation may be limited by: • The difficulty of waiting for long periods to capture the relevant phenomena. • The expense of observer costs and equipment. • The reliability of inferences from surface indicators. • The problems of quantification and disproportionately large records. • The limitation on presenting activities and inferences about cognitive processes.
We can classify observation in terms of the observerparticipant relationship. This relationship may be viewed from three perspectives: (a) Is the observation direct or indirect? (b) Is the observer's presence known or unknown? (c) Is the observer a participant or nonparticipant?
The design of an observation study follows the same general pattern as other research. Observational studies fall into four general types based on the degree of structure and the nature of the observational environment. The researcher must define the content of the study; develop a data collection plan that identifies participants, sampling strategy, and "acts" (often operationalized as a checklist or coding scheme); secure and train observers; and launch the study. Unobtrusive measures offer an unusual and creative approach to reducing reactivity in observational research by indirect observation and other methods. Measures of erosion and accretion serve as ways to confirm the findings from other methods or operate as singular data sources.
Managers faced with current decisions requiring immediate attention often overlook internal data in a company's data warehouse. Data mining is the process of discovering knowledge from databases. Data mining technology provides two unique capabilities to the researcher or manager: pattern discovery and the prediction of trends and behaviors. Data mining tools perform exploratory and confirmatory statistical analyses to discover and validate relationships. These tools even extend confirmatory statistical approaches by allowing the automated examination of large numbers of hypotheses. Data mining involves a five-step process: sample, explore, modify, model, and assess.
