STUDIES IN SECRET-BALLOT TECHNIQUE

THE SCIENCE of public opinion measurement is still very young, as has been pointed out many times by Dr. George Gallup, Director of the American Institute of Public Opinion. It is true that this scientific research has shown much progress in recent years, as evidenced by the record. In the case of the Institute, an average state-by-state error of 6 per cent in the 1936 Presidential election survey was reduced to an average state error of only 2.4 per cent in the 1940 Presidential election survey. This progress toward an increasingly accurate measurement of public opinion has come about largely because of experimentation which has resulted in improved techniquesexperiments which found the answers to some of the unsolved problems of opinion research. Further experimentation will play an important r6le in improving opinion research methods so that a high degree of accuracy can be maintained. Last fall, during the Presidential election surveys, the Institute conducted a number of experimental studies, several of which are briefly reported here.