Samuel Stouffer and the GI Survey: Sociologists and Soldiers during the Second World War

5. The Research Branch Refined

Chapter 5 The Research Branch Refined In the fall of 1944, when it might be expected that a successful research organization connected to a large bureaucracy would have rested a bit on its laurels and settled into a routine, Stouffer was still looking for ways to refine and improve the Research Branch. He was looking ahead, not behind, and hoped his thoughts would help the Research Branch in “charting our future plans.” Although he acknowledged that his ideas were “pretty obvious,” he sought to classify the levels of research he was then conducting as “observation,” “controlled observation,” and “controlled experiment.” Stouffer was chary about observation. He warned his colleagues that simple observational reports lend themselves to “editorializing .” “Moreover,” he wrote, “if a report merely exposes defects without specifically suggesting corrective action, it can be punishing, and backfire on the researcher .” In addition to the “anxiety” such reports aroused, they could provide “ammunition which can be used by the devil as well as by the angels.” While not dismissing observation as a tool, Stouffer thought its major flaw was that it did not render a “prima facie case for making any given recommendation.” Observations had severe limitations—although Stouffer acknowledged that they were the starting point of all research. Controlled observations were of a higher order in Stouffer’s mind, as they limited the observation to two or more preselected groups. The results rendered “approximate those of a controlled experiment,” with the proviso that the controlled 66 The Research Branch Refined experiment offered greater reliability than controlled observations because variables were more easily controlled in the former. Therefore, Stouffer favored the controlled experiment because it “represents the only research method which embodies in the research the recommendations for action.” Stouffer was seeking to find ways of raising the level of the Research Branch’s labors to controlled experiment and of moving away from simple observation, while recognizing that the best he would often be able to get was controlled observation rather than controlled experiment. He was also, he wrote, somewhat frustrated with the difficulty of verifying findings even from controlled observations, and was keenly aware of the criticism that would be leveled at the Research Branch should he be unable to do so: If the Division is concerned with helping the Army discover what practices really will improve morale—both as a contribution to winning this war and erecting doctrines for leadership of the post-war army—some effort must be made to follow up the method of controlled observation by explicitly getting put into practice in certain units the presumptive lessons learned therefrom and then observing whether they really work. . . . If the war were to end today and if the Army should ask us what single practice General Osborn’s million-dollar research operation has proved to be helpful to morale, we honestly could not cite a scrap of scientific evidence. The curtain would go up on the stage and there we would stand —stark naked.1 Ever-scientific and ever-thoughtful, Stouffer was also frustrated by social science ’s lack of preparedness before the war, and by the Research Branch’s inability during the war to move beyond the practical into the theoretical—a problem of which he was well aware even while the war was underway. Shortly before the end of the war, he expressed his frustrations to the president of the University of Chicago: We here are forever haunted by the fact that what we are accomplishing is such a small fraction of what might have been done if the science of social psychology had been ready at Pearl Harbor with a body of tested theory. I believe that one of the great obligations of those of us in social sciences who go back to our uni- The Research Branch Refined 67 versities after the war will be to reexamine critically the shortcomings of the kind of crude pressure job we have done, and to help reformulate social psychological theory sharply so that hypotheses can be tested by crucial controlled experiments, with the aid of new quantitative tools.2 Stouffer was as good as his word, as The American Soldier proves. Although it is, in fact, primarily a work of data, The American Soldier is not a work of unprocessed data. It includes examples of discrete theory, such as that of relative deprivation —a concept of immense sociological importance.3 Still, Stouffer was all too...