ABSTRACT
Survey research is generally viewed as distinct from experimental-or quasi-experimentalresearch in which the researcher is interested in estimating the impact of a specific intervention or treatment on persons subjected to the intervention/treatment. 1 In practice, however, it is difficult to draw such clear distinctions, as when random sampIes 01' a population are surveyed (questioned) before and after implementation of a particular public policy to estimate the impact of that policy. Survey research is also most often associated with the questioning of cross-sectional sampIes of the population as in political polling or market research. However, the decennial census that attempts to question oll U.S. citizens. not a sampIe, is generally considered to be a survey as is the Current Population Survey that incorporates a longitudinal sample using a panel design with continuous refrcshment of ncw people. Surveys of voter behavior also are frequently longitudinal. involving data collection on the same sampIe before and after an election. Though so me methodologies for data collection-specifically in-person interviews and self-administered paper questionnaires-are most often associated with survey research, increasingly questionnaires administered over the Internet are also used and even individual tests and direct observations are sometimes utilized. EpidemiologicaI surveys frequently employ longitudinal designs as weil as individual medical tests and observations of rather large sampIes. Despite these differences, the most distinguishing common feature of what we will call scient(fic survey research is its concern with being able to generalize research findings to a specific population or group.
