Problems in the Use of Survey Questions to Measure Public Opinion

Science  22 May 1987:
Vol. 236, Issue 4804, pp. 957-959
DOI: 10.1126/science.236.4804.957


Sample interview surveys are frequently proposed and sometimes used as a way of studying public choices among alternatives. Questions in such surveys may be either "open" or "closed." Two experiments are reported that demonstrate the difficulty of inferring not only absolute levels but even relative orderings of public choices from either type of question, although such questions can be used more successfully to study temporal change or variations across social categories.