Observational research involves descriptive and analytical studies. First, participant observation (in which the observer is obvious to and involved with the subjects) is less valid than a questionnaire would be for sensitive data. Second, the observer's expectations affect what he or she sees and reports, reducing the validity of the data. Validity is thus diminished when the observer reports seeing either "everything" or "nothing." In its turn, observational epidemiology involves the following types of research: descriptive studies, ecological studies fallacy, cross sectional studies and cohort studies (Oakes and Kaufman 2006).
In participant observation, the researcher is a regular participant in the activities being observed; while in non-participant observations, the researcher is not a participant in the ongoing activities being observed. Second, structuring the observation increases validity by focusing the attention of the observers on certain characteristics and events. Third, placing the observation on a scientific foundation by stating a hypothesis up front increases validity by avoiding distortion. Using observation only for studying those epidemiological phenomena that are appropriate to this method (e.g., nonverbal behaviors and social interactions) increases validity (Beaglehole et al 2007).
In contrast to observational epidemiology, experimental design involves ra