The Measurement of Social Attitudes
Louis L. Thurstone
University of Chicago
IT IS an honor and a privilege for me to have this opportunity of addressing the Midwestern Psychological Association. I wish that I could do justice to the occasion and express my appreciation by an address that is worth listening to. I have selected among the few subjects that are available to me one that may be of fairly general. interest while it still involves many theoretical and psychological problems. I shall discuss the measurement of social attitudes. In doing so I shall review the development of the measurement methods that are applicable to attitudes and I shall also discuss some of the critic isms and questions that have recently been raised about this subject.
Several years ago when I was teaching conventional psychophysics, it seemed to me that psychophysics was really a very dull subject in spite of the fact that it did offer the satisfaction of clean and quantitative logic. This type of satisfaction is rare in psychological investigation and consequently psychophysics has stood out as a very dignified topic in psychology in spite of the fact that its intrinsic subject matter has been, on the whole, rather trivial. These depreciative statements about psychophysics can be readily amplified by referring to the conventional publications in this subject. You will then find that one of the elaborate parts of the subject is the determination of limens. 'There is a, great deal of hairsplitting about just how a limen should be determined with the greatest possible precision. in determining a limen you fit a phi-gamma curve and. then there is more hairsplitting as to - whether you should adjust the errors of observation in the proportions or in the stimulus magnitudes, Then you will find. the several psychophysical methods compared as to which gives the most reliable limen determination. And then you can find short-cuts for these methods by which you can determine somebody's limen very quickly when you are in a hurry for a limen. Now, it seems strange that I have never seen a, psychologist who really cared much about any particular person's limen for anything!
I venture the guess that not more than perhaps half a dozen
( 250) psychologists in this room have ever needed or wanted somebody's limen for anything with a high degree of precision. And I don't believe that the proportion would be much higher in a group of Eastern psychologists.
Of course we are interested to know the order of magnitude of errors in visual discrimination as compared with those of various forms of auditory discrimination. Even in a rough pitch discrimination test we are determining a limen in a sense, but these problems never involve any profundities of curve fitting for the limen of an individual subject. Why then does psychophysics bother so much about methods, and short-cuts for these methods, which are never used on individual subjects except when an individual serves as a specimen for some type of situation in which the methods must be adapted to the conditions of each problem. anyway? This bothered me also in teaching the subject and I carne to share the distrust of my students in the significance of the whole subject.
One way in which to retain the satisfactions that can be found in the logic of this subject is to change its content. We have tried this and it has seemed to some of us that psychophysics thereby takes on an entirely new aspect. Instead of asking a person, "Which of these two little cylinders is the heavier?" (apologies to Mr. Boring for the stimulus error) instead of asking a person which of two cylinders is the heavier, we might as well ask him something interesting, such as, " Which of these two nationalities do you in. general prefer to associate with?" or, "Which of these two offenses do you. consider to be in general the more serious'?" or, "Which. of these two pictures or colored designs do you like better?" Questions of this sort of discrimination might be multi-plied indefinitely and if they could be handled with some sort of psychophysical logic it is clear that we should have here the possibilities of objective description of more psychological significance than the sensory limen.
The first objection that I encountered is that the very term psychophysics should be strictly limited to the field of sensory discrimination and that the questions that I prefer to ask should not be regarded as of the dignity of psychophysics—they should be relegated to the field of mental tests and questionnaires. I find justification for extending the use of psychophysics even to questions that are interesting by referring to Titchener. He says: "Fechner was chiefly interested in the intensive aspect of mental processes, and among mental processes in sensation.. His example has led other inquirers to give a disproportionate amount of
( 251) attention to the laws of sensation intensity." It is unfortunate that Mr. Titchener did not happen to ask his students to judge the relative merits of handwriting specimens or of English compositions or to make social judgments because then I am sure that we should now have the permission to use psychophysical methods iii a wide variety of problems.
Furthermore, if, instead of printing our questions on a piece of paper we should rig up an elaborate automatic contraption. for exposing these questions, running the said contraption by a electric motor, and spreading plenty of kymographs and telegraph keys and speech keys and time markets all around the room, then I am sure that our studies would qualify as experimental psychology. But since we ask the subject to indicate his response with a pencil instead of by a telegraph key, our investigations have been outside the pale of experimental psychology.
When the constant method is used in its complete form so that every stimulus serves in turn as a. standard, then it becomes the method of paired comparison. There was no quantitative logic for handling the method of paired comparison so as to obtain measurement which. satisfied the criterion of internal consistency. This difficulty was overcome by finding an equation that satisfied this criterion. It has been referred to as the law of comparative judgment. With this rational equation and the method of paired comparison we have made several studies involving social stimuli and in which the subjects were asked to express various kinds of judgments other than mere comparison of physical magnitude.
In one of these experiments a. list of twenty nationalities was presented to several hundred students (1). The nationalities were arranged in pairs so that every nationality was paired with every other one in the list. The .students were asked to underline one nationality of each pair to indicate which of the two nationalities they would rather associate with. The returns were tabulated in the form "proportion of the subjects who prefer nationality A to nationality B". With these experimental proportions and by the law of comparative judgment, the scale separation was calculated for each pair. By means of these data, it was possible to construct a linear scale of attitude to which each nationality was allocated. At the top of the list is the American., next come other English speaking countries, and at the bottom of the list are nationalities or races other than. our own. This order is what one should expect., but the scale values could not be predicted.
In general the scaling is accomplished on the principle that if the group of subjects very generally prefers A to B, then the
( 252) proportion of the subjects who vote for A will be high, perhaps close to unity (2). If, on the other hand, the two nationalities are about equally well liked by the group, then there will be about as many subjects who vote for A as there are subjects who vote for B. Hence, the proportion above described will be close to .50 and the two nationalities will have zero separation, that is, they will have the same scale value.
In an experiment of this sort the. criterion of internal consistency consists in the discovery that by assigning one scale value to each of the nationalities we can reconstruct all of the experimentally independent proportions. With twenty nationalities in the list, we have twenty scale values arid. these must be sufficient to lock the 190 experimentally independent proportions within the known probable errors of the given proportions.
One of the criticisms of this procedure has been that the en-tries in the list are not true nationalities. For example, in the list occur such entries as Jew, Negro, South American. It is not necessary to restrict ourselves to accepted anthropological classifications in these experiments. We are measuring the degree of affect for or against the social. objects listed. This is legitimate even if some of the classifications are races-rather than nationalities, or religious, or groups of nationalities. For example, it is conceivable that some of the students disliked -South Americans in general without knowing much about them and without stopping to debate whether these South Americans whom they disliked were one or twenty nationalities. The category is a conversational one which lends itself to the expression of affect, and it therefore serves our purposes even if the psychological or affective category does not fit the accepted anthropological or political classifications. There is of course nothing to prevent the use of other classifications so long as the categories lend themselves to the expression of the likes and dislikes of people generally.
Another criticism that has been offered against experiments of this type is that it would. make a difference if the question were worded differently. For example, in the form here described the question was " Which of these two nationalities or races would you rather associate with?" Now, so rims the objection, what would happen if the question were "Which would you rather have as a fellow student?", "Which world you rather have for a. neighbor'?" "Which world you rather have your sister marry??" and ''Which would you rather do business with?" (3). Fortunately, this question could be answered by an experiment. fifteen. hundred blanks were used in which there were three hundred blanks
( 253) with each of the five questions. These were arranged in random. order so that it was a. matter. of chance which of the five forms was given to each of the fifteen hundred subjects in the experiment. When the blanks had been. filled in, they were sorted out into five piles according to the question on the blank. The twenty nationalities were then allocated to a scale separately for each of the five questions. The twenty nationalities were found to be in the same rank order in the scales which were constructed on the basis of the five questions. This proves that the scale value of a nationality is determined primarily not by the detailed form of the question but rather by the general degree of like or dislike of the subjects for each nationality. This general like or dislike is what we have called the potential action toward the object or attitude.
On the other hand, if the question had been "Which nationality would. you. rather have as a servant?" it is quite conceivable that the scale values might have been different, because such a question is not calculated to bring out the attitude of the subject toward the nationality .with regard to social equality. If the question asked which. of the nationalities is the more intelligent, which is -physically superior, which is more emotional, which is the taller, it is quite certain that the scale values would have been different.
The generalization of the last experiment might be made by noting that a subject could be half -way through the blank, responding by underlining one of each pair, and still not remember what the question was that the was answering by his underlining. This means merely that the subject gets a set of checking preferences without recalling the cognitive detail of the question. The question has then served its usefulness by giving the subject a set of expressing his attitudes rather than some intellectual judgment about the nationalities in question. The scale value of each nationality measures the affective value of the nationality for the group of subjects.
It is obvious that the scale value of each nationality in these experiments is a description of the group of subjects as much as it is a description of the nationalities. If the same experiment were repeated with Italian students or with Russian students, the scale values of the nationalities would undoubtedly be radically different.
This suggests the possibility of measuring cultural similarities or dissimilarities. Suppose that these paired comparison. schedules were filled in by university students in ten different countries. The scale value of each of the twenty nationalities would be cal-
( 254) -culated separately for each of the ten countries. We. could then calculate the correlation coefficient for the scale values of. the twenty nationalities in two countries, for example, the German and the French students. If the correlation coefficient were high it. would. indicate a similarity in the nationality preferences of the two groups of students. If the correlation were low it would measure dissimilarity of the two groups of students as to their national preferences. If country C is very much hated by A and liked by B, then that difference would be measurable by the difference in the scale value for nationality C in the scales for groups A and B. In this way, international affiliation, and antipathies might be described in. a quantitative manner. Of course differences might appear in different occupational groups and in different regions of the sane country.
Suppose that one group of subjects likes all of. the twenty nationalities about equally well. Then the twenty nationalities would have the same scale value. Suppose that another group of subjects has very decided likes and dislikes among the twenty nationalities. Some of the nationalities would then have high scale values while others would have conspicuously low scale values. In other words, the spread in scale values would be much greater for the second. group than for the first group. The intolerance of a group is measured by the spread in scale values. In this manner it would be possible to measure the tolerance of different countries for other countries. These measurements of international tolerance might conceivably have considerable social interest.
Some of our experiments have been set. up so as to measure the effect of social stimuli on the international attitudes of high school children. We have worked with a. number of motion picture films as stimuli (4). In Genoa, Illinois, the film. "Four Sons" was shown in the local theater and 131 children iii grades 7 to 12 inclusive. were given free tickets. Several days before the performance they were asked to fill in a paired comparison schedule of nationality preferences. The Germans were included in this list. The morning after the performance the children again filled in the same schedule. It was assigned that the attitudes of the children did not change toward the twenty nationalities during the course of a week or ten clays between_ the two schedules except for the possible effect of the film on their attitude toward Germans. The film made the children much more friendly toward Germans. This experiment and others of the same general type demonstrate
( 255) that the effect of a single social stimulus on the international attitudes of the subjects can be measured.
The same general. technique has been used for measuring attitude toward crimes (5). In such experiments a list of crimes, was arranged in the same paired forms and the subjects were asked to indicate for each pair which of the two crimes they considered to he the more serious. In this manner it was found in one town that the attitudes of children toward gambling were considerably affected by seeing, the picture "Street of Chance". The film had the effect of making the children_ regard gambling as a more serious offense after seeing the picture. 'These experiments with motion pictures have been. effectively conducted by Miss Ruth. Peterson.
The question has been raised as to the degree of permanence of these effects. In order to answer this question we have repeated the schedules of comparison in several towns after an interval of four to five months. The attitudes have returned about half way, toward their original values in four months, but these effects. vary of course with the film used and the frequency of other social stimuli. In one town the effect lasted without diminution for five months.
When a series of social stimuli has been allocated to an affective continuum by the method of paired comparison we have the scale separations but we do not have a :rational origin for the affective continuum. This is a problem for which Mr. Horst has found a, very ingenuous solution. It is of some psychological interest to locate a series of stimuli to an affective continuum in such a manner that the measurements refer to a datum of affective neutrality. The solution. by Mr. Horst consists in asking a group of subjects to compare one stimulus A which is likely to be favorably regarded with another stimulus B which is likely to be considered unfavorable. The subject is asked this question, "Would you be willing to endure the disadvantage B in order to have the advantage A.?''' The proportion of subjects who are willing to accept B in order to have A locates the affective origin. between the two stimuli. The same procedure can of course be extended to a whole series of stimuli so that the location of an affective datum in the series can be tested by the criterion of internal consistency.
Another method of measuring attitude is to use a statement scale (6). This consists in a series of opinions which are submitted to the subject for endorsement or rejection. These statements or opinions have been so selected that they constitute an evenly
( 256) graduated series and so that a scale value can be given to each opinion. If the opinions A, B, C, D are four successive opinions such an evenly graduated scale about prohibition, for example, then: the following conditions would. be satisfied. If one person. endorses opinion. A and another person. endorses opinion B, then a group of observers should find. some difficulty in saying which of the two opinions is more favorable to prohibition. Let us suppose that three-fourths of the observers would say that opinion A is more favorable to prohibition than opinion. B. Then this degree of difficulty in judging .which of them is the more favorable to prohibition constitutes a measure of the .separation between the two statements of opinion on the attitude scale. Now, if opinion C is so chosen. that three-fourths of the observers say that B is more favorable to prohibition than C, then. the scale separation between A and B is the same as the separation between B and C.. In this manner a series of statements is selected from a large number so that the apparent increment in attitude from one statement to the next is the same for the whole series.
With a scale value assigned to each statement or opinion, it is of course easy to calculate the median scale value of all the statements that any given individual has endorsed. This median scale value is the score of that person on the attitude scale. The meaning of these scores can be illustrated further as follows.
Suppose that three individuals, X, Y, Z, have attitude scores on prohibition which are equally spaced. Then the difference or increment in attitude between X and Y would seem to be the same as the difference in attitude between Y and Z. In other words, it would be just as difficult to discriminate between and as to which is the more favorable. to prohibition as it is difficult to discriminate between Y and Z. This is the basis for the construction. of the attitude scales. The psychophysical experimental methods by win eh the attitude scales are constructed so as to satisfy these requirements are beyond the scope of this paper.
The statement scale enables one to make several types of measurement of which the following are examples.
The attitude of an individual subject can be measured by means of a, statement scale. The paired comparison procedure enables us to compare groups of subjects but the statement scale procedure is preferable for measuring the attitudes of individual subjects. The range of statements that the individual endorses gives some indication of his tolerance. It is possible to plot a frequency distribution of the attitudes of a. group of people toward labor unions, for example. This distribution has a central tend-
( 257) -ency or average and it has a measurable dispersion. Two groups of people may then be found to have the same average score on a disputed issue but one of the two groups may be more heterogeneous than the other. The degree of heterogeneity in attitude of a group of people is directly measured by the standard deviation of the frequency distribution of their attitude scores. This is an important aspect .of group comparisons which can be reduced easily to measurement in terms of the dispersion of the scores,
In two small towns, West Chicago and Geneva, Illinois, an experiment was arranged so that a film favorable to the Chinese was shown in one town and a film unfavorable to the Chinese was shown in the .other town (7). The two flints were "Welcome..Danger" which is thought to he unfriendly to the Chinese and which has been so criticized by the Chinese themselves, and "Son of the Gods" which is generally thought to be friendly in its interpretation of Chinese culture. The films were shown in the local theaters and the children were given free tickets to the performances. In each town the children were asked to fill in a statement scale about the Chinese several days before seeing the film and .also the morning after seeing it. The results show a very decided shift in favor of the Chinese in Geneva where "Son of the Gods was shown. In West Chicago there -was a small opposite effect where the children saw the film, "Welcome Danger." The effect of a single social stimulus, such. as a motion picture film, on the international attitudes of school children can be described by the statement scale as well as by the pen red comparison method.
The statement scale is constructed by asking a group of one hundred judges to sort out a list; of opinions into a, series of eleven successive piles to represent attitudes from one extreme to the other. The question was raised early in our experiments whether the attitudes of the judges themselves would influence the final scale values of the statements in the scale. For this reason Mr. Hinckley set up an experiment with a scale for measuring attitude toward the negro. He had 114 statements about the negro. At one extreme were the opinions that the negro is the equal of the white man and should have equal social privileges. At the other extreme were the opinions that the negro is inferior to the white man and should not have the same social privileges. Three groups of judges were used., namely, one group of white college students friendly toward the negro, one group of white college students who thought the negro was definitely inferior, and one group of educated negroes. The whole list of 114 statements was scaled separately for the three groups of judges. The result was that
( 258) the three scales so constructed were practically identical, thus proving that the attitudes of the judges have no serious effect on the measuring function of the statement scale.
It is possible to apply the statement scale method to the measurement of social trend. This will be illustrated in terms of attitude toward the Germans and the French. A collection. of quotations from newspaper editorials has been made by Mr. Russell. His quotations cover the twenty year period 1910 –1930. A group of judges sorted the editorial quotations into a series of eleven piles rouging from No. 1. expressing extreme admiration for the Germans to No. 11 expressing extreme contempt for them. The scale value of each quotation was calculated. Then the average scale value of all the quotations from the year 1910 was determined. It is the mean attitude toward the Germans for that year. The average scale value of all the quotations from year 1911 was determined and. it is the mean attitude toward the Germans for that year. In this manner the mean attitude toward the Germans was represented quantitatively so that this social trend for a period of 20 years could be inspected in a single graph.
For this particular issue the curve shows the expected depression in the average scale value for the Germans during the years of the war and a corresponding rise in average scale value for the French during the same period. Of course the curves so plotted represent only the altitudes of the editorials of one large newspaper. M. Russell is now making this type of inquiry for four newspapers, namely the New York Times, the World, the Chicago Daily News, and also the Chicago Tribune. The curves for these four newspapers are similar. They all show a return of attitudes toward the pre-war values with interesting deviations that correspond to popularly discussed issues at various times. The newspapers also show some differences in the rapidity with which their editorials return toward. pre-war attitudes for Ger-many. We have here the possibility of measuring the changes in. attitudes as represented in the press .during past times even though the attitude scales were not available for these periods. It will be interesting to study by these quantitative methods the rapidity with which international attitudes have changed before and after each of the recent modern wars by analyzing the foreign. press for a few years before and after each was. It will also be of interest to correlate these rates of change with other social facts such. as facility of communication, similarity of language and culture, and the like. Perhaps this will be a psychophysical contribution to the methods of history.
Many of the criticisms and questions that have appeared about attitude measurement concern the nature of the fundamental concepts involved and the logic by which the measurements are made. I shall consider a few of these questions briefly.
One of the most frequent questions is that a score on an attitude scale, let us say the scale of altitude toward. God, does not truly describe the person's attitude. There are so many complex factors involved in a person's attitude on any social issue that it cannot be adequately described by a simple number such as a score on some sort of test or scale, This is quite true but it is .also equally true of all measurement.
The measurement of any object or entity describes only one attribute of the object measured. This is a universal characteristic of all measurement. When the height, of a, table is measured, the whole table has not been described but only that attribute which was measured. Similarly, in the measurement of attitudes, only one char characteristic of the attitude is described by a measurement of it.
Further, only those characteristics of an object can be measured which can be described in terms of "more" or "less". Examples of such description are: one object is longer than another, one object is hotter than another, one is heavier than another, one person is more intelligent than another, more educated than another, more strongly favorable to prohibition, more religions, more strongly favorable to birth control. than another person. These are all traits by which two objects or two persons may be compared in terms of "more" or "less".
Only those characteristics can be described by measurement which can be thought of as linear magnitudes. In this context, linear magnitudes are weight, length., volume, temperature, amount of education, intelligence, and strength of feeling favorable to an object. Another way of saying the same thing is to note that the measurement of an object is, in effect, to allocate the object to a point on an abstract continuum. If the continuum is weight, then individuals may be allocated to an abstract continuum of weight, one direction of which represents small weight, while the opposite direction represents large weight. Each person might be allocated to a, point on this continuum with any suitable scale which requires some point at which counting begins, called the origin, and some unit of measurement in terms of which the counting is done.
The linear continuum which is implied in all measurement is always an abstraction. For example, when several people are
( 260) described as to their weight, each person is in effect allocated. to a point on an abstract continuum of weight. All measurement implies the reduction or restatement of the attribute measured to an abstract linear form. There is a popular fallacy that a. unit of measurement is a thing—such as a piece of yardstick. This not so. A unit of measurement is always a. process of some kind which can be repeated without modification in the different parts of the measurement continuum.
Not all of the characteristics which are conversationally described in terms of "more" or "less" can actually be measured.. But any char characteristic which lends itself to such description has the possibility of being reduced to measurement.
We admit that an attitude is a complex affair which cannot be wholly described by any single numerical index. For the problem of measurement this statement is analogous to the observation that an ordinary table is a complex affair which cannot described by any be wholly described by any single numerical index. So is a man such a complexity which cannot in wholly represented by a single index. Nevertheless we do not hesitate to say that we measure the table. The context usually implies what it is about the table that we propose to .measure. We say without. hesitation that we measure a man when we take some anthropometric measurement of him. The context may well imply without explicit declaration what aspect of the man. we are measuring, his cephalic index, his height, or weight, or blood. pressure, or what not. Just in the same sense we shall say here that we are measuring attitudes. We shall state or imply by the context the aspect of people's attitudes that we are measuring. The point is that it is just as legitimate to San that we are measuring attitudes as it is to say that we are measuring tables or men.
Whenever a common word is adopted for scientific use it nearly always suffers some restriction in its connotation in favor of greater precision of meaning. This has happened in many sciences so that it is by no means peculiar to psychological terms. Consider, for example, such words as elasticity, resilience, momentum, force, which are after all. common ordinary words but as they are used by the physicist they are very much restricted and. more precise while still retaining the essential ordinary idea. So it, is in psychology with terms like sensation, perception, illusion, meaning, idea, and concept. Now when we turn scientific logic and. experimental psychophysical procedures to the subject of attitudes we fired it necessary to restrict here also the rather
( 261) loose conversational meaning of this term in order to make it at all suitable for scientific discourse.
Our present definition of the term may be briefly stated us follows: Attitude is the affect for or against a psychological object. Affect in its primitive form is described as appetition or aversion. Appetition. is the positive form of affect which. in more sophisticated situations appears as liking the psychological object, defending it, favoring it in various ways. Aversion is the negative form of affect which is described as hating the psychological object, disliking it, destroying it, or otherwise reacting against it. Attitude is here used. to describe potential action toward the object with regard only to the question whether the potential action will be favorable or unfavorable toward the object. For example, if we say that a man's attitude toward prohibition is negative, we mean that his potential actions about prohibition may be expected. to be against it, barring compromises in particular cases. When we any that a man's attitude toward prohibition is negative, we have merely indicated the affective direction of his potential action toward the object. We have not said anything about the particular detailed manner in which he might act. In this sense the term attitude is an abstraction in that it cannot be described without inserting the cognitive details-that are irrelevant but this is also true of many of the simplest. concepts in daily use.
The affect about an object may be of strong intensity or it may be weak. The positive and negative affect therefore constitutes a linear continuum with a neutral point or zone and two opposite directions, one positive and the other negative. Measurement along this affective continuum is of a discriminatory character with the discriminal error as a unit of measurement.
Against this restricted definition of attitude as the affective-character of potential action about a psychological. object these have been raised several questions. It has been pointed out that the emotional experiences of the past constitute an integral part of a man's attitude. If we should use the term in that inclusive manner we should say that a man's attitude toward. religion consists in par of his childhood experience with Sunday Schools. The man might then say that his attitude toward religion is that he went to church when lie was a child. I should prefer to say that such a fact is really not a part of his attitude toward religion but that it may help to explain how he got that way. It is quite conceivable that two men may have the same degree or intensity of affect favorable toward a psychological object and that their
( 262) attitudes would be described in this sense as identical but that they arrived at their similar attitudes by entirely different routes. It is even possible that their factual associations about the psychological object might be entirely different and that their overt actions would take quite different forms which have one thing in common, namely that they are about equally favorable toward the object.
In these discussions the term psychological object has its customary meaning. It may refer to a physical object or it may refer to an idea, a plan of action, a form of conduct, an ideal, a :moral principle, a slogan or a symbol. In fact it may refer to any idea about which the subject may express positive or negative affect.
There comes to mind the uncertainty of using an opinion as an index of attitude. The man. may a liar. If he is not intentionally misrepresenting his real attitude on a disputed question, he may nevertheless modify the expression of it for reasons of courtesy, especially those situations in which frank expression of attitude may not be well received. This has led to the suggestion that a, man's action is a. safer index of his attitude than what be says. But his actions may also be distortions of his attitude. A politician extends friendship and hospitality in overt action toward a negro while hiding an attitude that he expresses more truthfully to an intimate friend. Neither his opinions nor his overt act constitute in airy sense an infallible guide to the subjective inclinations and preferences that constitute his attitude. Therefore we must remain content to use opinions or other forms of action merely as indices of attitude. It must be recognised that there is a discrepancy, some error of measurement as it were, between the opinion or overt action. that we use as an index and the attitude that we infer from such in index.
But this discrepancy between the index and "truth" is universal. When you want; to know the temperature of your room, you. look at the thermometer and use its reading as an index of temperature just as though there were a single temperature reading which is the "correct" one for the room. If it is desired to ascertain. the volume of a glass paperweight the volume is postulated as an attribute of the piece of glass, even though volume is an abstraction. The volume is measured indirectly by noting the dimensions of the glass or by immersing it in water to see how much .water it displaces. These two procedures give two indices which may not agree exactly. In every situation involving measurement there is postulated an alstract continuum such as volume or temperature, and the allocation of the thing measured
( 263) to that continuum is accomplished by indirect means through one or more indices.. Truth consists only in the relative consistency of the several indices, since it is never directly known. We are dealing with the same type of situation in attempting to measure attitude. We postulate an attitude variable which is like practically all other measurable attributes in the nature of an abstract continuum, and Ave must find one or more indices which will satisfy us to the extent that they are internally consistent. Only to the extent that different indices are consistent can we be justified in postulating the attitude as a trait for a trait is never directly measured.
If we should find that what a man says has absolutely no relation to what he does, then such inconsistency would constitute a serious limitation on the legitimacy of the abstraction of attitude. However, when. we actually carry out such comparisons we do find that the correlation is positive between verbally expressed attitudes and overt action. The correlation is not perfect but it is certainly positive. Discrepancies will arise as when a subject expresses himself as favorable to prohibition although he himself violates it. But if the correlation is tabulated for a large group of people, it is found that the attitude scores for those who vote for prohibition are markedly different from those who vote against it, and a similar positive correlation is found between attitude scores and overt action about drinking. The season for limiting ourselves to verbal expressions of attitude is that they can be evaluated with more certainty and they are much more available than a list of overt acts. This type of correlation between attitude and overt action was the subject of a recent study by Mr. Stouffer.
In order to deal with overt actions as expressive of attitudes in a feasible manner Mr. Rosander has prepared lists of situations with alternative overt acts. He asks the question, In the following situations, which of the given alternatives are you most likely to do?" Then follows description of a situation with a number of alternative overt acts. For example, in a situation scale on the negro occurs the following:
The congregation of the church you attend has always been white. One Sunday morning a, negro attends the services.
a. You do
nothing about it.
b. You complain to the minister.
c. You welcome the negro to the church.
d. You shake hands with him and ask him to come again.
e. You sound out opinion to fund how many want to keep out negroes.
f. You tell the fellow he had better move along.
g. You ask the minister to tell him that he is not wanted.
h. You tell the fellow he had better leave before you throw him out.
i. You defend the negro against some who complain of his presence.
j. You give the negro friendly warning that perhaps he had better not come back.
The various overt responses are to be scaled in a, manner analogous to the procedure for the statements of opinion. It is quite probable that these two types of scale, the opinion scale and the situation scale, will be highly correlated.
We take for granted that people's attitudes are subject to change. When we have measured a man's attitude on. any issue such as pacifism, we shall. not declare such a measurement to be in any sense an enduring or constitutional constant. His attitude may change, of course, from one day to the next, and it is our task to measure such changes, whether they be due to unknown causes or to the presence of some known persuasive factor, such as the reading of a discourse on the issue question. However, such fluctuations may also be attributed in part to error in the measurements themselves. In order to isolate the errors of the measurement instrument from actual fluctuations in attitude, we must; calculate the standard error of measurement of the scale itself, and this eau be accomplished by methods already well known in mental measurement.
We shall assume that an attitude scale is used only in those situations in which one may reasonably expect people to tell the truth about their convictions or opinions. If a denominational school were to submit to its students a scale of attitude about the church, one might find that some students would hesitate to make known their convictions if they deviate from the orthodox huh ifs of their school. At least, the findings could be challenged if the situation in which attitudes were expressed contained pressure or implied threat bearing directly on the attitude to be measured. Similarly, it is difficult to discover attitudes on sex liberty by a written questionnaire, because of the well-nigh universal pressure to conceal such attitudes when they deviate from supposed conventions. It is assumed that attitude scales will be used primarily in those situations that offer a minimum of pressure on the attitude to be measured. Such situations are common enough
However, it is sometimes of considerable interest to inquire what the distribution of attitude may be in a group which is known to be influenced by social. pressure or taboo. If, for example,
( 265) a group of college students are asked to express their attitude on the subject of sex liberty, the results might indicate conformity with conventional standards. Such a result might be interpreted to mean that the students agree with. conventional ideals in. regard to sex liberty or the results might, be challenged as reflecting only the social taboo against deviations from the conventional standards. lf, on the other hand, the results should be a distribution of unconventional attitudes, the interpretation would be more conclusive in that the expressed attitude appears in spite of the known. taboo. It goes without saying that the distribution of attitude on any social issue and with any particular group must be interpreted in terms of the known factors that may influence judgment.
The question has been raised whether the concept of' attitude as here used and as measured by an attitude scale is not hypothetical rather than "real."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, volume, are hypothetical in physical science. No one has ever seen or touched a force or a momentum or a volume. They are measured by what they supposedly do. The legitimacy of these abstractions can be tested only in the consistency by which they operate in experience. Not infrequently these. hypothetical entities are discarded either, first, because. they lead to inconsistencies in experience; or, second, because they have to be multiplied in number so that they become as numerous as the effects that they are intended to explain or facilitate in analysis. As long as biologists insisted on the definition of instincts in terms of overt acts, they found that the instincts had to be as numerous as the overt acts to be accounted for and then the instinct abstraction lost its usefulness. .The instincts probably could be defined more successfully in terms of the other end of the psychological act. The concept of intelligence is a useful though hypothetical entity. It is postulated that intelligence is that which is dynamically common to a large group of overt acts. The degree of this hypothetical power which we call intelligence is estimated in terms of overt performances and the term is successful to the extent that different forms of adaptive overt performance are positively correlated. The greater confidence with which we handle such a hypothetical entity as force is completely contained in the higher degree of consistency with winch the hypothetical force is measured in different forms of its expression. We are here dealing with a similar hypothetical
( 268) concerned merely with the description of the degree of affect for or against various social symbols by psychophysical methods. In giving each person a positive or negative score on a. disputed social issue, we do not, say anything whatever as to whether his attitude is good or bad, whether his attitude should be censured or encouraged. That is a matter of interpretation in each issue and it is not the scientific problem which we are concerned.
In closing I shall mention only one other question that has appeared on several occasions. It has been. suggested that the attitude scales might be used in order to eliminate undesirable students from colleges and. universities. They might he given an attitude scale on patriotism or on religion or on. something else that is supposed to tell whether a person is desirable or not from any particular point of view. In the first place, you would immediately make liars of many applicants who differ with you in their political or religious convictions on the issue. in question. We have not yet combined the attitude scales with the lie detector, although such experiments are contemplated. But even if it were possible to ascertain the political and religion attitudes of people under conditions winch would detect when they are lying, it would be a vicious policy for any educational institution to adopt.
It has been proposed that attitude scales might be used to determine whether a course of instruction in social science has had the desired effect. To be sure, one of the important results of social science instruction is change in social. attitude but to make the passing of a course contingent on taking the so-called right attitude on any particular social issue would be. preposterous. I am unalterably opposed to any such policy for judging progress in social science courses.
I have been informed that in the State of California several boys have been denied graduation from high school because or political beliefs. The stupid psychology of those school authorities can hardly be calculated to make those boys change their political convictions. If ignored they might change their minds many times before reaching mature years. But it is startling to hear that an adolescent can be refused graduation from a public educational institution because of his political beliefs. And this happens in one of our otherwise educationally most progressive states. I sincerely hope that none of the attitude scales that we have developed will he put to such vicious purposes.
I have reviewed some of our attempts to extend the experimental methods and the logic of psychophysics beyond the field of sensory discrimination to which it has been limited by psycho-
269) logical tradition. It has been stated by economists and by oilier
social scientists that affect cannot be measured, and some of the fundamental
theory of social science has been written with this explicit reservation. Our
studies have shown that affect can be measured. In extending the methods of
psychophysics to the measurement of affect we seem to see the possibility of a.
wide field of application by which it will be possible to apply the methods of
quantitative scientific thinking to the study of feeling and emotion, to
esthetics, and to social phenomena.
1. Thurstone, L. L., "Experimental Study of Nationality Preferences," Journal of General Psychology, July—Oct., 1928, pp. 405—425.
2. Thurstone, L. L., "A Law of Comparative Judgment," Psychological Review, July, 1927, pp. 273—286.
3. Eggan, Frederick ., "An Experimental Study of Attitude toward Races and Nationalities” ,Chicago, 1928, unpublished Master's Thesis.
4. Thurstone, L. L., "A Scale for Measuring Attitude toward the Movies," Journal of Educational Research, 1920, Volume 22, pp. 89-94.
Thurstone, L. L., "The Measurement of Change in Social Attitude," to he, published in the Journal of Social Psychology.
Thurstone, L. L., "The Influence of Motion Pictures on Children's Attitudes," to he published in the Journal of Social Psychology.
Thurstone, L. L., "The Effect of a Motion Picture Film on Children's Attitudes toward Germans," unpublished manuscript.
5. Thurstone, L. L., "The Influence of Motion Pictures on Children's Attitudes," to he published in the Journal of Social Psychology.
6. Thurstone, L. L., and Chave E. .J., "The Measurement of Attitude," Chicago, University of Chicago Press, 1929.
7. Thurstone L. L., "Measurement of Change in Social Attitude," to he published in the Journal of Social Psychology.