Application of Experimental Designs in Survey Settings.

The article informs that under certain conditions experimental and sample survey designs can be combined within a single study in order to obtain the advantages of both. Authors hope to open systematic considerations of what these conditions are. In discussing some advantages and disadvantages of experiments and surveys, sociologist Leslie Kish noted three important assets of experiments: randomization controls the effects of confounding variables; systematic introduction and variation of independent variables clarifies the direction of causation; and use of these designs permits more flexible, efficient, and powerful statistical manipulation. At the same time, Kish argued, experiments have the following three important disadvantages, the last two of which constitute the chief merits of sample surveys: it is often difficult to specify independent variables with sufficient precision so as to exclude all confounding extraneous variables; samples used in experiments often do not systematically represent an important universe; and measurements in experimental situations often cannot be made under the same conditions as in the natural settings of real populations. Kish concludes the following from this discussion: In any specific situation one method may be better or more practical than the other; but there is no overall superiority in all situations for either method. Understanding the advantages and weaknesses of both methods should lead to better choices. While authors wish to emphasize their agreement with this conclusion, people would also like to suggest that there are circumstances in which it is appropriate and desirable to combine both methods, thereby gaining some advantages of each.