Commentary on Rice's "Statistical Studies of Social Attitudes and Public Opinion"
Louis L. Thurstone
The galley proof of the preceding article was submitted to Professor Thurstone for comment and criticism. He has been so kind as to prepare a statement which the editor takes much pleasure in appending at this point.
Professor Rice has raised a number of questions that are crucial in the problem of measuring social attitudes. Since several experimental studies have been made with some of these questions particularly in mind, and since these experimental studies have not all been published, a brief statement of the findings might here be in order.
One of the crucial questions that bothered me for some time was to ascertain whether the attitudes of the judges who help to construct the scale affect the scale values of the statements. If so, then the universality of the scale as a measuring instrument would, of course, be very seriously limited. Professor Rice raises this question in several forms. This problem was selected by Mr. E. D. Hinckley for his Doctor's thesis which has not yet been published. He prepared
( 193) a list of one hundred statements about the Negro, varying from those attitudes which gave him all the social privileges of the white man to those that restricted his social rights. He selected one hundred white college students in Chicago who expressed attitudes favorable to the Negro on five special statements, one hundred white college students in Florida who expressed attitudes restricting the social rights of the Negro, and one hundred educated Negroes. The same set of one hundred statements was sorted by the three groups independently and three scales were constructed. I was very much pleased to see that the three scales so constructed with the same statements were practically identical. This seemed to demonstrate quite effectively that the scale does transcend the attitudes of the judges who sort the statements.
Professor Rice raises the question, "Where is the distinction to be drawn between factual judgment opinions and attitude-representation opinions?" It is, of course, clear that a simple statement of fact which is well known cannot serve as an opinion by which to measure attitude. There are two statistical criteria by which the distinction is rather easily made, namely, the criterion of ambiguity and the criterion of irrelevance. Consider the statement, " Many churches are now being built." What would the judges do with it in sorting? They would have difficulty in sorting it, and its spread or ambiguity would probably be large. That would eliminate it from the scale because it does not convey with sufficient uniformity an attitude favorable or unfavorable to the church. It is a simple statement of fact. Consider the statement, "I don't believe church-going will do anyone any harm." This was statement No. 9 in our first experiment and it was sorted with small ambiguity.
( 194) But it was eliminated by the criterion of irrelevance because the pious people as well as those who are antagonistic to the church endorsed it. These two criteria of ambiguity and irrelevance constitute objective methods of drawing the distinction in Professor Rice's question.
The same criteria would eliminate the illustrative statement of Professor Rice about the Negro, namely, "I am impatient of the Negro's dependence upon sympathetic white men. Force him to work and take care of himself." Such a statement would be interpreted by some judges as reflecting an attitude favor-able toward the Negro and by other judges in the opposite manner. Hence, its ambiguity would be large and the statement would not be retained in the scale.
The term attitude has caused some trouble because its meaning has been rather hazy in current writing. I have redefined it recently for the purposes of these experimental studies as the degree of affect about a psychological object. Affect has two directions in its primitive sense, appetition and aversion. It is here used to designate potential action which is favorable toward a psychological object or unfavorable and against it. The strength of this affect which may be positive or negative is the trait measured by the attitude scale. It is just as hypothetical as the concept of intelligence which is measured by what it supposedly does. But these concepts are hypothetical in the same sense that the concepts force, momentum and volume are hypothetical in physical science. No one has ever seen or touched a force. It is measured by what it supposedly does. The legitimacy of these abstractions can be tested only in the consistency by which they operate in experience.
That a common core of affect about a psychological object can be postulated which is more or less independent of its detailed rational and verbal expression is indicated by a recent Master's thesis by Eggan. He submitted a list of twenty nationalities for preferential judgment by the method of paired comparison. Six different questions were submitted as, for example, "Which nationality would you prefer to associate with?" and "Which nationality would you rather have your sister marry?" The scale values of the twenty nationalities were practically unaffected by the six different forms of the question. In fact, a subject can proceed to make these preferential judgments about nationalities even after he has forgotten the exact question at the beginning of the schedule. This is as it should be because the preferential judgments are intended to be affective indicators and for this purpose the details of the rational question are really distractors. Affect cannot be conveyed except in some particularized and overt form.
Professor Rice raises the question whether the attitude scales would be affected by the degree of expertness of the group of judges. The statements should be such as occur in the actual population whose attitudes are to be measured and in this sense the scale is limited to one culture. People who cannot read hand-writing probably could not judge whether or not it is beautiful. If the statements are long and involved, the scale would not be applicable to those who cannot read complex prose. An experiment should be set up in which a group of experts, a group of educated people, and a group of uneducated people serve as judges in the sorting process. I should venture the prediction that the scales will be identical except that the unit of measurement will be small for the experts
( 196) and large for the uneducated group. The discriminal error will be smallest for the experts. It is as though the distances between stations along a railroad were expressed in miles and in kilometers. The scales are identical except for the unit of measurement.
It is quite probable that the subject of psycho-physics which has hitherto been barren can be vitalized in its application to these psychologically and socially interesting problems. The problems raised by Professor Rice are fundamental. As he has indicated, many of these problems can be solved experimentally so that we may rid the subject of much dialectics.