METHODOLOGICAL BIAS IN PUBLIC OPINION SURVEYS

Statistically designed sample surveys have enabled pollsters to gauge public opinion on a wide range of issues. In such surveys, selection of a data collection technique is generally based on four criteria: (1) cost; (2) completion time; (3) response rate; and (4) response bias. Typically, more weight is placed on the first three factors and, as a result, adequate attention has not been given to the latter consideration. The study described in this paper looks at one type of response bias-that which results from the use of a specific data collection method. More specifically, this research uses a controlled experimental design in order to determine whether responses given in a public opinion polling are influenced by the method used to collect the data. Three methods are investigated: (1) mail questionnaire; (2) telephone interview; and (3) personal interview.