The Measurement of Value
Part III Attitude Measurement: Introduction
Louis L. Thurstone
In attempting to adapt and extend the psychophysical methods to interesting stimuli, I found another opportunity when I learned about the work of Floyd Allport in arranging a series of attitude statements to represent gradations from one extreme to the other on some debatable issue. To make such a list with a defensible unit of measurement was then the problem. The paper "Attitudes Can Be Measured" was written after we had done some experimental work with the problem of scaling such material. The paper was criticized by some colleagues in the social studies, but it had some defenders. This paper was the start of a lot of work in the psychometric laboratory on the construction of attitude scales. That subject has developed into a large field, so that it is now a major branch of social psychology.
The second paper in this section describes the scaling of the list of statements that were used by Allport.
The third paper on the study of nationality preferences was an application of paired comparisons and the law of comparative judgment to attitude measurement. As judged by the number of requests for reprints, it is perhaps one of the best-accepted papers that I have written. It was a transition from psychophysics to attitude measurement in our laboratory. Paired comparisons is a far more sensitive method for appraising attitudes than the statement scales, which are always rather crude, but it is not so generally applicable to social issues.
The sixth paper in this section was my presidential address for the Mid-western Psychological Association. It was a review of the problem of attitude measurement and closely related problems. This subject had become quite popular in two or three years.
The seventh and eighth are concerned with experiments in measuring changes in attitudes of school children induced by motion pictures. The Payne Fund studies consisted of about thirty experiments on this problem. In a number of experiments we found a very marked effect on the attitudes of school children from a single motion picture. The most striking effect was found with the film, The Birth of a Nation. These studies were described in a lithoprinted report that was available for some years.
The question was frequently discussed whether there was a cumulative effect of several motion pictures on the same issue. Our experiments demonstrated such an effect rather clearly.
Another question often discussed was whether a summation effect could be demonstrated for a number of films each of which had only slight propaganda effect. For this purpose we used three films that had only slight effect in changing the attitudes of children as regards the severity or leniency of punishment of criminals. Seven comparable groups at Moosehart were used in this experiment. Different groups, each of about 150 children, saw one film, two films, three films, and no film, respectively. The summation effect on their attitudes was clearly shown.
I used to have the hypothesis that the effect of propaganda films had differential effects, depending on the intelligence and education of the audience, but we did not continue attitude experiments to investigate such problems. The hypothesis was that when propaganda is spread on thick, it might be effective for an unintelligent audience but that it might backfire with a more sophisticated audience. With an educated audience the best propaganda might be a gentle plug inserted in material that was ostensibly concerned with other things, the final result being attained by the summation effect. Events of the last two decades have made me question this hypothesis. It seems that any propaganda can be put over on any audience if it is cleverly planned.
In a small 1929 monograph I described two types of attitude scales (pp. 93-96). These were called the maximum probability type and the increasing probability type. All of our work was with the first type. Recently there has been interest in the second type of scaling, which lends itself to certain types of attitude problems. The scaling of attitude statements can be accomplished directly from the records of acceptance and rejection for a group of subjects and without the sorting procedure that we used, but, as far as I know, such a scaling procedure has not yet been developed. It can also be done by factor analysis.
When I was working on attitude measurement, I found great interest in the
application of attitude scales to all sorts of groups, but I was disappointed in
the relative lack of interest in the methodological problems which seemed to be
more important for the development of social science. I had only scratched the
surface of an important field that justified more fundamental methodological
study. In the early thirties we prepared quite a number of attitude scales. When
I realized that the psychometric laboratory at the University of Chicago might
be swamped with such an enterprise, I decided to stop it. All of the incomplete
work on a number of attitude scales was abandoned to make time and room for the
development of multiple factor analysis which was already well under way.