FITTING RESEARCH DESIGN TO RESEARCH
PURPOSE: INTERNAL AND EXTERNAL VALIDITY
Although a wide range of potential research methods is covered in the following chapters, research strategies in the social and behavioral sciences can be divided roughly into two general types— surveys and experiments. The former include all observations that occur in “natural” (i.e., nonlaboratory) settings and which involve a minimum of interference over people's normal behavior or choices. Experiments (which can be conducted either in laboratory or field settings) include those observational studies in which data are collected under conditions where behavioral choices are limited or in some way constrained by the controlled manipulation of variables and measures selected by the researcher. The advantages and limitations of these two types of research strategies tend to be complementary, so an effective program makes use of both in different states of the research process.1 Experimental methods are particularly advantageous for determining causal relationships. They also are ideally suited for specifying systematic relationships among sets of isolated and rigidly controlled variables. However, for research in human behavior, the very control that marks the advantage of experimental techniques places limitations on the representativeness of the phenomena they are used to study.
Surveys, on the other hand, have the value of “real world” context and the availability of mass data in developing information about human actions. However, these advantages are bought at the cost of a lack of control over nonsystematic variation in the variables of interest. The inability to exert control over critical variables can result in interesting, but scientifically inconclusive, findings. The relative value of experimental versus survey research methods depends to a large extent on the importance of making inferences about the causal relationships among the variables being studied (Brewer, 2000).____________________