THE OPINION-POLLING TECHNIQUE AS USED IN A STUDY
OF THE EFFECTS OF BOMBING ON JAPANESE
The sensational failure of the public-opinion polls to predict the election of Harry Truman in 1948 brought the beanstalk growth of sampling techniques forcibly to public attention. During the period of 15 or 20 years before that date, the notion that one could tell what all of the people thought by asking some of them what they thought had developed into a flourishing and lucrative enterprise. Entrepreneurs of the most varied sorts came to rely upon public-opinion polls to determine what the country would buy from them. Candidates for political office wrote speeches, manufacturers changed the shape of ironing boards, movie and radio magnates discarded productions, and public relations experts mounted million-dollar campaigns, all in accordance with the decision of "the sample."
Basically, the method of the sample is one which is made use of in almost every branch of scientific investigation. In order to find out the effect of a newly compounded drug on diabetes the physiologist tries it on a limited number of those afflicted with the disease. The artillery officer fires a few shells from a new shipment. The chemist tests several fragments of a vein of ore-bearing rock. In each case, the assumption is that if the sample is a true sample--that is, chosen without bias, completely at random--what is found out about the sample will be true of the whole also.
The novelty of the sampling technique is therefore not in its principle, but in its specific application to social situations. In this application, as the interpretation of the preelection polls of 1948 made clear, the technique is open to serious error. How the numerous possibilities of error and distortion can be eliminated or minimized, and in general, the nature of the limitations of sampling methods in social