University Microfilms, Inc., Ann Arbor, Michigan - American University
University Microfilms, Inc., Ann Arbor, Michigan
-
Upload
khangminh22 -
Category
Documents
-
view
0 -
download
0
Transcript of University Microfilms, Inc., Ann Arbor, Michigan
This d issertation has been 63—50 m icrofilm ed exactly as received
GIBSON, Jam es W illiam , 1932- DIRECT AND INDIRECT ATTITUDE SCALE MEASUREMENTS OF POSITIVE AND NEGATIVE ARGUMENTATIVE COMMUNICATIONS.
The Ohio State U niversity , P h .D ., 1962 Speech—Theater
University Microfilms, Inc., Ann Arbor, Michigan
5
probable if an individual initially agrees with the message or is the probability of reinforcement or change greater if the subject initially disagrees with the message? The implications for persuasion are im-
Oportant. Research reported by Brehm suggests that pressures will develop to reduce the state of dissonance. Evidence to support this. statement is based on subject action. This study will involve an examination of attitudinal changes taking place in consonant and dissonant subjects. The direct and indirect attitude scales will be utilized to measure the extent of attitude change as a result of the communication stimuli.
I. Experimental QuestionsThe experimental questions to be answered in this study are
these:1. What relationship exists between attitude scores toward
censorship obtained with a Thurstone attitude scale and attitude scores toward censorship obtained with a forced-choice attitude instrument?
2. Do positive type communication stimuli induce greater attitude changes than communication stimuli which are negative in
structure?3. Are changes in attitude by homogeneously structured audi
ences as a result of a communication stimulus different from changes in attitude by heterogeneously structured audiences?
O Jack W. Brehm and others, Attitude Organization and Change.New Haven: Yale University Press, 1960.
109
POSITIVE STIMULUS
Throughout history man has made his greatest accomplishments when his creative mind has been free to roam and develop ideas.In ancient Greece, Plato, Socrates and Aristotle developed philosophical theories which remain today as some of man's major contributions to ethical conduct. Homer's Iliad and Odyssey written by the blind poet who sang his works in verse bear witness to the achievements possible by man when he is left free to develop his thoughts.
Michaelangelo, Raphael and DaVinci are outstanding examples of the genius of man and the extents to which he can contribute to art forms with native talent and freedom of expression.
The great universities, whose purposes are to provide a climate where scholars can freely exchange ideas and attempt to arrive at truth or answers to problems facing man, are another product of a free society. From the time when the first university was established at Bologna, free men in their education have constantly re-evaluated society and formulated new theories for social and scientific action. No rigid boundaries are established for the containment of thought and the student is encouraged to challenge existing beliefs with the aim of discovering new truths. This atmosphere of academic freedom has stimulated such men as Hegel, Kant, Spinoza, Dewey and a host of others to develop revolutionary approaches to philosophy and to experiment and apply them to citizens of their age. These same universities have trained scientists whose inventions such as the diesel engine and the telephone and whose theories such as the theory of relativity have been invaluable contributions to man's efforts to solve the secrets of the universe.
A free society does not serve as the contributor of knowledge for men must have in their minds certain capacities and bents for creativity, but a free society provides for talented men an unparalleled opportunity for experimentation and examination.
A free society affords the author and the playwright a place to formulate impressions of society as he views it and to transmit to other human beings, through the medium of the stage or the printed page, those feelings and beliefs. Great authors like Shakespeare, Dante, Hawthorne and Whitman flourished in an environment which permitted them to express freely their impression of society and its institutions.
110
Critics of men and their organizations could gain.prominence only when a society encourages the expression of all points of view as a means of determining desirable courses of action. From the beginning of times, man has been, in some way, dissatisfied with his environment. Plato, Thomas Paine, Jonathan Swift, Eugene O'Neill and Valter Lippman have used the tools of satire and sarcasm as weapons to criticize and assist other citizens in a re- evaluation and redirection of their own and other nations' efforts.
Our nation is a tribute to freedom. Just as ancient Greece, Great Britian and France experienced their greatest social and intellectual achievement when freedom prevailed, so too our nation since its liberation from Great Britian and its Declaration of Independence has moved forward to become leader of the world. Our free citizens, aware that their freedom offers them the opportunity to determine their own course and success in life, have been spurred on to accomplishments previously considered impossible. Great inventors like Thomas Edison, Henry Ford and Charles Kettering have literally raised themselves by their bootstraps from poverty to wealth. The story of America is a rags to riches adventure in ideas as well as in worldly goods. Here was a nation, small but potentially wealthy, which achieved greatness by granting its citizens freedom of speech and self determination. Many other nations have had the same potential but only in America with its manifold freedoms has such a measure of success been realized.
We pride ourselves in living under a government where the individuals or political party out of power can freely criticize the party in charge of the government. The Republicans and their leader, former President Eisenhower, are now raising their voices in opposition to the policies of President Kennedy. At all levels, the freedom indict individuals, parties or policies is accepted and practiced as an inherent right of our form of government.
The moving force of each generation is its people; their abilities and desires. Freedom so often is the catalytic agent which hastens developments that might be delayed. But freedom, like any catalyst, cannot be a substitute for forces. If these abilities and desires exist, and they have existed in virtually every society since ancient Greece, then freedom will hasten the progress and achievements of mankind.
112
This communication was (check one)
_____________ Positive_____________ Negative
In evaluating the communications which you have just read, please make a check mark at the appropriate location on each of the following continua.
logical :____ :___:___:___: illogicalspecific____ :___:___:___:___:___ : generalclear_______ :___:___:___:___:___ : vagueinsufficient too muchmaterial____ :___:___:___:___:___ : materialstructurally structurallysound_______ :___:___:___:___:___ : weakinteresting :__:___:___:___:___: boringaccurate :__:___:___:___:___: inaccurate
Do you have any suggestions which might improve this communication?
If you feel there are specific sections which need improvement, please indicate in the margin beside the deficient section.
114
BIBLIOGRAPHY
Allport, Gordon W., and Postman, Leo, The Psychology of Rumor.New York: Henry Holt and Company, 1947.
Anastasl, Anne, Psychological Testing. New York: The Macmillan Company, 1954.
Bartlett, Claude, Quay, Lorene and Wrightsman, Lawrence, "A Comparison of Attitude Measurement: Likert Type and Forced-Choice," Educational and Psychological Measurement.” 1960, pp. 699-704.
Baruch, Dorothy W., "Aggression During Doll Play in a Preschool," American Journal of Orthopsychiatry. 11, 1941, pp. 252-259.
Bogardus, E. S., "Measuring Social Distance." Journal of Applied Sociology. 9, 1925, pp. 299-308.
Bray, Douglas W., "The Prediction of Behavior from Two Attitude Scales," Journal of Abnormal and Social Psychology. 45,1950, pp. 64-84.
Brooks, Keith, "The Construction and Testing of a Forced-Choice Scale for Measuring Speaking Achievement," Ph.D. dissertation. The Ohio State University, 1955.
Campbell, Donald, "The Indirect Assessment of Social Attitudes," Psychological Bulletin. 47, 1950, pp. 15-38.
Cronbach, Lee, Essentials of Psychological Testing. New York: Harper and Brothers, I960.
Cromwell, H., "The Relative Effect on Audience Attitude of theFirst Versus the Second Argumentative Speech of a Series," Speech Monographs. 17, 1950, pp. 105-122.
Dubin, Sanford, "Verbal Attitude Scores from Responses Obtained in the Projective Technique," Sociometry. 3, 1940, pp. 24-28.
Edwards, Allen L., Experimental Design in Psychological Research. New York: Rinehart and Company, 1950.
115
________________ . Statistical Analysis. New York: Rinehart andCompany, 1958.
_______________ and Kenney, Kathryn, "A Comparison of the Thurstoneand Likert Techniques of Attitude Scale Construction,"Journal of Applied Psychology. 30, 1946, pp. 72-83.
Ferguson, George A., Statistical Analysis in Psychology. New York: McGraw-Hill Book Company, 1959.
Festinger, Leon, "A Theory of Cognitive Dissonance. Evanston, 111. Row Peterson, 1957.
Fromme, Allan, "On the Use of Certain Qualitative Methods of Attitude Research: A Study of Opinions on the Methods of Preventing War," Journal of Social Psychology. 13, 1941, pp. 429-459.
Garrett, Henry E., Statistics in Psychology and Education. New York: Longmans, Green and Company, 1953.
Gordon, Leonard V., "A Comparison of the Validities of the Forced- Choice and Questionnaire Methods in Personality Measurement," Ph.D. dissertation. The Ohio State University, 1950.
Green, Bert, "Attitude Measurement," Volume I, Handbook of Social Psychology. Edited by Gardner Lindzey, Reading, Mass.: Addison-Wesley Publishing Company, 1954.
Guilford, J.P., Fundamental Statistics in Psychology and Education. New York: McGraw-Hill Book Company, 1956.
.________ , Psychometric Methods. 2d Edition, New York: McGraw-HillBook Company, 1954.
Guttman, Louis, "The Basis for Scalogram Analysis," Measurementand Prediction. Princeton: Princeton University Press, 1950.
Hammond, Kenneth R., "Measuring Attitudes by Error Choice: AnIndirect Method," Journal of Abnormal and Social Psychology. 1948, 43, pp. 38-48.
Hartmann, G. W., "A Field Experiment on the Comparative Effectiveness of Emotional and Rational Political Leaflets in Determining Election Results," Journal of Abnormal and Social Psychology. 48, 1953, pp. 78-92.
116
Hinckley, E. D., "The Influence of Individual Opinion on Construction of an Attitude Scale," Journal of Social Psychology.3, 1932, pp. 283-296.
Horst, Paul, The Prediction of Personal Adjustment. Bulletin No.48, Social Science Research Council, 1941.
Hovland, Carl, Janis, Irving and Kelley, Harold, Communication and Persuasion. New Haven: Yale University Press, 1954.
Knower, Franklin H., "Experimental Studies of Changes in Attitudes: 1. A Study of the Effect of Oral Argument on Changes of Attitude." Journal of Social Psychology. 6, 1935, pp. 315-347.
Krech, David and Crutchfield, Richard, Theory and Problems of Social Psychology. New York: McGraw-Hill Book Company,1948.
Levitt, Eugene E., "The Development of Prejudice: A Critique of the Horowitz Faces Test," Child Development. 27, No. 2,June, 1956, pp. 155-171.
Likert, Renesis, "A Technique for the Measurement of Attitudes," Archives of Psychology. No. 140, 1932.
Lovell, George and Haner, Charles, "Forced-Choice Applied to College Faculty Rating," Educational and Psychological Measurement.15, 1955, pp. 291-304.
Lund, Frederick H., "The Psychology of Belief," Journal of Abnormal and Social Psychology, 20, 1925, pp. 63-81.
May, Mark and Hartshorne, Hugh, "First Steps Toward a Scale forMeasuring Attitudes," Journal of Educational Psychology. 17, 1926, pp. 145-162.
McGregor, Douglas, "The Major Determinants of the Prediction of Social Events," Journal of Abnormal and Social Psychology,33, 1938, pp. 179-204.
McNemar, Quinn, "Opinion-Attitude Methodology," Psychological Bulletin, 43, 1946, pp. 289-375.
Morgan, John J. B., and Morton, James T., "The Distortion ofSyllogistic Reasoning Produced by Personal Convictions," Journal of Social Psychology, 20, 1944, pp. 39-59.
117
Mussen, Paul H., "The Reliability and Validity of the Horowitz Faces Test." Journal of Abnormal and Social Psychology. 45, 1950, pp. 504-506.
Osgood, Charles, Suci, George and Tannebaum, Percy, The Measurement of Meaning. Urbana: University of Illinois Press, 1957.
Parrish, Jack Albert, "The Direct and Indirect Assessment of Attitudes as Influenced by Propagandized Radio Transcriptions," M.A. thesis. The Ohio State University, 1948.
Proshansky, Harold M., "A Projective Method for the Study of Attitudes ," Journal o^Abnormal and £5ocial Ps^cholog^* 38, 1943, pp. 393-395.
Remmers, H. H., Introduction to Opinion and Attitude Measurement.New York: Harper and Brothers, 1954.
Rosenberg, Milton; Hovland, Carl; McGuire, William; Abelson,Robert and Brehm, Jack, Attitude Organization and Change New Haven, Yale University Press, 1960.
Ross, Paul, "A Comparison of Two Methods of Matching in Forced-Choice Rating," Ph.D. dissertation. The Ohio State University, 1955.
Sanford, Fillmore H., "The Use of a Projective Device in Attitude Surveying," Public Opinion Quarterly. 14, 1950, pp. 697-709.
Senders, Virginia, Measurement and Statistics. New York: Oxford University Press, 1958,
Sherif, Muzafer and Cantril, Hadley, "The Psychology of Attitudes,” The Psychological Review. 52, 1945, pp. 295-319.
Siegel, Sidney, Nonparametric Statistics. New York: McGraw-Hill Book Company, 1956.
Sisson, E. D., "Forced-Choice, The New Army Rating," Personnel Psychology. 1, 1948, pp. 365-381.
Sponberg, H. A., "A Study of the Relative Effectiveness of Climax and Anti Climax Order in an Argumentative Speech," Speech Monographs. 13, 1946, pp. 35-44.
118
Tendler, A. D., "A Preliminary Report on a Test for EmotionalInsight,1' Journal of Applied Psychology, 14, 1930, pp. 122- 136.
Thurstone, L. L., "Attitudes Can Be Measured," American Journal of Sociology, 33, 1928, pp. 529-554.
____________ and Chave, E. J., The Measurement of Attitude. Chicago;University of Chicago Press, 1929.
Wechsler, Irving R., and Bernberg, Raymond E., "Indirect Methods of Attitude Measurement," International Journal of Opinion and Attitude Research, 4, 1950, pp. 209-228.
Woodward, Harold S., "Measurement and Analysis of Audience Opinion," The Quarterly Journal of Speech, 14, 1928, pp. 94-111.
6
4. What relationship exists between changes in attitude by persons already in agreement with the stimulus and person initially
in disagreement with the stimulus?
AUTOBIOGRAPHY
I, James William Gibson, was born in Marysville, Ohio, July•'•v
15, 1932. I received my Secondary education in the public schools of Marysville, Ohio. My undergraduate training was obtained at Otterbein College, Westerville, Ohio, from which I received the degree Bachelor of Arts in June 1954. From July 1954 to June 1956 I served in the U. S. Army. From September 1956 to May 1958 I was a member of the teaching staff at Marysville High School, Marysville, Ohio; and froi* September 1958 to May 1959 I taught speech at Rushville High School, Rushville, Indiana. In September 1959 I received an appointment as Director of the Ohio High School Speech League in the Department of Speech, The Ohio State University.I received the degree Master of Arts from The Ohio State University in March 1960. In September 1960 I was appointed an Assistant Instructor. All requirements for the degree Doctor of Philosophy except the dissertation and the final oral examination were completed by mid-Spring 1962, and an appointment as Assistant Professor to the faculty of Butler University, Indianapolis, Indiana,was
received, effective September 1962.
119
CHAPTER II
REVIEW OF THE LITERATURE
In order to discuss attitude measurement, its philosophy and operation, a description of the underlying assumptions and a definition of an attitude should be developed.
An attitude can be viewed as a latent variable instead of being considered directly observable. In describing an attitude, the reactions of an individual to a variety of situations involving the same underlying continuum are involved. The description then is ofvariables of the same general classification, involving both overt
8and covert responses. We are measuring what Guttman calls "the universe of attributes," and such measurement involves investigation of all classes of behavior within the universe. Green mentions that there are three types of attitude universes: (1) elicited verbalattitudes derived through the asking of questions regarding the attitude, (2) spontaneous verbal attitudes such as the expression of feeling in normal conversation, and, (3) action attitudes or the
actual activity exhibited toward the person or object in the referent oclass. A description of the universe must necessarily precede
®LoUis Guttman, "The Basis for Scalogram Analysis," Measure- . ment and Prediction (Princeton, N. J.: Princeton University Press,1950), p. 80. qBert Green, "Attitude Measurement,11 Handbook of Social Psychology, Vol. 1, Gardner Lindzey, editor (Reading, Mass.: Addison- Wesley Publishing Co., 1954), p. 340.
7
8
the development of items which attempt to tap various portions of the universe.
The difficulty encountered in defining the term "attitude" is emphasized by Sherif and Cantril^ when they mention that the confusion regarding attitudes and their measurement is highlighted by the existence of some 23 different definitions of attitudes. To formulate a single definition constituting an acceptable approach for all 23 wordings may be impossible, but an eclectic attempt will be made. Possibly Quinn McNemar in his comprehensive review of opinion- attitude methodology has succeeded in fusing the best from all the definitions of attitude into a single statement. According to McNemar, "The common element of most definitions of social attitudes is such that an attitude is a readiness or tendency to act or react in a certain manner."**- Inherent within this definition are the implications that attitudes are (1) drive producing, (2) learned, (3) formed in relation to persons, values and objects which originally may not have had motivational appeal, and, (4) enduring states of readiness.
Since attitude measurement often is referred to as opinion research, definition of an opinion as, " . . . a verbal expression of
12attitude," would include opinion research within the constellation
*®Muzafer Sherif and Hadley Cantril, "The Psychology of Attitudes," The Psychological Review. 52, 1945, pp. 295-319.
■^Quinn McNemar, "Opinion-Attitude Methodology," Psycho1ogica1 Bulletin. 43 (July 1946), p. 289.
*2l . L. Thurstone, "Attitudes Can Be Measured," American Journal of Sociology. 33 (January 1928), p. 531.
9
of attitude measurement. However, because what we frequently hear termed opinion surveys are not conducted with instruments of the types developed by Likert, Thurstone, Remmers, Guttman et al., but rather rely upon a single structured question, this discussion will not include within its purvey this type of opinion measurement because of its questionable value.
I. Criteria for Evaluation of Attitude Scales A necessary prerequisite to the development of any measuring
instrument, be it achievement test, diagnostic test, physical examina tion or attitude measure, is the determination of criteria for its evaluation. The reasons for these criteria are obvious: unless the instrument performs the task it is supposed to perform and can successfully replicate the performance at a later date provided the variables remain constant, quantification of any usefulness will not be achieved. McNemar points out that the fundamental requirementsfor an adequate attitude measure are (1) Reliability, (2) Validity,
13and, (3) Unidimensionality.Reliability
According to Anastasi, "The reliability of a test refers to the consistency of scores obtained by the same individuals on different
^McNemar, o j j . cit., p. 312.
sets of equivalent items. If an attitude instrument consistentlymeasures the same universe of attitudes, we could expect then, thatif the same attitude instrument were administered to a group of personson two separate occasions there would be a high correlation betweenthe two scores. Application of the statistical measure of correlationfor determination of reliability of scores on a measuring instrumentcan be accomplished by three methods, according to Ferguson.'*'*’
Test-retest method. A test, or in this case an attitude measuringinstrument, is administered twice to the same sample of individualsand the scores are correlated. Under this method of determining thereliability coefficient and through use of the parallel forms methodthere may be something less than /l.00 correlation realized. Greenmakes this observation and reasons:
If the questionnaire is administered to the same group of respondents on two separate occasions there will be some shifts in the scores of individuals. . . . Memory and familiarity with the scale will tend to increase stability; extending the time interval between administrations will tend to decrease stability.^
^Anne Anastasi, Psychological Testing (New York: The Macmillan Co., 1954), p. 94.
15George A. Ferguson, Statistical Analysis In Psychology And Education (New York: McGraw-Hill Book Co., 1959), p. 279.
^Green, o j j . cit., p. 338.
11
Because attitudes are dynamic it is not expected that scores will remain completely stable but neither can attitudinal measuring instruments be defended on the premise that instability of the score and the resulting low correlation of results are indicative of sensitivity to attitudinal sets.
Parallel forms method. Parallel forms or equivalent forms of the test are given to the same subjects and the observations or scores are correlated. With this method of determining reliability coefficients it is also expected that some instability of scores will result but the reliability of the change in attitude can be better determined through this method To be assured of parallelism it is imperative that the criteria for determination of equivalency be formulated. One method is to construct two tests to meet the same specifications but containing different attitude statements or questions to which the subjects will react. The presence of differing amounts of motivation and knowledge of the test even in the parallel forms is an admitted difficulty and will affect the reliability coefficient, but this method is considered superior to the test-retest method
17Ibid.. p. 339.18Anastasi, oj>. cit.. p. 106.
12
Split-half or odd-even method. Scores on items are halved and
the two halves are correlated. This method is satisfactory if the spiral technique of increasing difficulty or dimensionality is not followed in constructing the test. A more satisfactory method used in psychological testing and particularly in attitude measurement is the odd-even method. Scores or marks on odd numbered items are correlated with scores on even numbered items. Because of the problems inherent in a test-retest or equivalent forms determination of the reliability coefficient, it has generally been the practice in attitude measurement to use the odd even or split half method.
ValidityThe question of validity could be stated, "Does the instrument
measure what it purports to measure?" In the case of the attitude instrument, it should sample the attitude universe. Psychologists, in referring to the validity of measures mention three types.
Face validity. This type of validity which is merely a question of whether or not a test appears to measure the given material or attitude universe is of little use to the attitudinal researcher. In fact, the disguised appearance of the instrument may often give
it the opportunity to make a more accurate recording of the subjects attitude than if the subject could determine the purpose of.the test
upon first viewing it.
13
Concurrent validity. As discussed by Cronbach, concurrentvalidity pertains to the statistical correlation between two measureswhich attempt to measure the same attitudinal universe and which are
19administered at very nearly the same time. Thus, if a new non- disguised structured attitude test were developed and its results could be quantified so as to be comparable to an already validated Likert type scale, both scales could"be administered to the subjects and a statistically significant validity coefficient between the scores on the two tests would indicate that both tests were sampling the same attitude universe.
Predictive validity. The prediction of action attitudes is the goal of many attitude researchers. If an attitudinal instrument can predict the activity of certain subjects subsequent to the taking of the test, then the attitude instrument has truly measured the subjects tendency to1 act. An example of this type of validity is seen in the Horowitz Faces Test, a well-known projective measuring device.Mussen^® reports that the validity of the test was established because subjects showing a great deal of prejudice on the faces test also demonstrated considerable prejudice in a behavioral situation involving stimuli similar to those on the test. Because of
19 " -Lee Cronbach, Essentials of Psychological Testing (New York:Harper and Brothers, 1960), p. 104.
^®Paul H. Mussen, "The Reliability and Validity of the Horowitz Faces Test," Journal of Abnormal and Social Psych. 45 p . 505.
14
Inconsistencies between elicited verbal attitudes and action atti-21tudes, as mentioned by Green, the use of indirect measures often
is a more valid predictor of action attitudes than direct instruments.The problem of determining the validity of many attitude measur
ing instruments is pointed up by Bray, "Attitude testers have not paidmuch attention to the validity problem. They have not compared re-
22suits on tests with subsequent behaviors." This problem may be one explanation for the use of the indirect devices which are more capable of accurately measuring the presence of the drive producing state.
Construct validity. This type of validity is concerned with an analysis of scores on a test in terms of psychological concepts Most frequently construct validity is examined when the tester has no available criterion of the value which he is exploring and he finds it necessary to employ indirect measures for validation. Construct validity is established by lengthy interaction between observations, reasoning and imagination. The experimenter may have in mind an incomplete theory and attempt to find evidence for it in certain behavioral activities. It may be possible only to hypothesize that a certain relationship should exist and the presence of these behav- iorally relevant constructs does not require that they be identified with the criterion measure. A briefer description of construct validity might involve three steps, (1) The mental development of
^Douglas W. Bray, "The Prediction of Behavior From Two Attitude Scales," Jnl. of Abnormal and Social Psvch .. 45, 1950.
^Cronbach, oj). cit.. p. 120.
15
constructs which might account for given performance on a task,(2) Development of testable hypotheses from this theory which sur
rounds the construct, (3) Conduct of a study to test the hypotheses.
Construct validity strives, then, to validate the theory which underlies the test.
Unidimens ionality
Attitude toward a social issue could be viewed as including a constellation of behaviors ranging from a favorable attitude toward
the issue to a highly unfavorable attitude toward the particular issue. A unidimensional scale would tap portions of the constella
tion ranging from one extreme to the other but not restricting it
self to an examination of any particular phase of the possible
reactions. The measurement of a single variable is actually the essence of a scale which is unidimensional. The items on the scale
are interdependent and measure different portions of the same latent
variable. Unldlmensionality can also refer to the measurement of a
single attitude rather than being a combination of measurements from
scales measuring attitudes toward different social issues or objects.
Although it is said a scale which samples more than a single dimen
sion of attitude has such inherent problems in its structure as to
make it invalid and unreliable, this depends upon traditional con
cepts of dimensionality.
16
II. Types of Attitude Scales
When attempting to measure social attitudes we are confronted
with an interesting dilemma. Basically, the question is whether the experimenter desires to indirectly measure the individuals social attitude toward the object, event or movement and thus derive a picture of the attitude without destroying the attitude in the process, or if the experimenter wishes to utilize the more direct attitude measures and trust that the subject will give an unbiased
picture of his social attitude. The selection of the type of instrument to be used has become an increasingly important consideration with rapid developments having been made in the area of indirect attitude measurement. Many of these instruments are a result of the "projective" personality measures developed by psychologists.
There are numerous advantages and disadvantages inherent in both the direct and indirect methods. In discussing indirect methods of measuring attitudes, Wechsler and Bernberg comment that this type of test conceals the intent of the measure from the subject and thus permits him to make responses without becoming personally involved.Additionally they believe that this method does not produce any effect
24 25on the attitude itself. Krech and Crutchfield assert that atti
tudes cannot be measured directly but must be measured on the basis
^Irving R. Wechsler and Raymond E. Bernberg, "indirect Methods of Attitude Measurement," International Journal of Opinion and Attitude Research 4, 1950, p. 211.
25David Krech and Richard Crutchfield, Theory and Problems of Social Psychology (New York: McGraw Hill, 1948), p. 207.
17
of Inferences drawn from the individuals experience and behavior.The value of the indirect projective type instrument is mentioned by
26Sanford but the time involved and the subjectivity of interpreta- tibn necessary in the evaluation step are stressed as disadvantages of this type of indirect measure. This criticism is applicable to the projective type of indirect measure but has little if any relevancy to the recently developed forced-choice and semantic differen
tial scales .The scales which attempt to measure social attitudes directly
are not open to the criticism of subjectivity of interpretation because of their objective method of scoring. In many instances, also, they can be constructed more rapidly and scored more quickly. How
ever , Bartlett, Quay and Wrightsman mention that the direct method ismore open to bias in the form of, ". . .deliberate faking, response
27set, or simple inaccurate estimation of one's own opinions."A useful classification of type of attitude scales or tests is
presented by Campbell who uses four categories:
26Fillmore H. Sanford, "The Use of a Projective Device in Attitude Surveying," Public Opinion Quarterly 14, 1950, pp. 697-698.
9 7 .Claude Bartlett, Lorene Quay and Lawrence Wrightsman, "A Comparison of Attitude Measurement: Likert Type and Forced-Choice," Educational and Psychological Measurement 20, 1960 , p . 699 .
18
1. Non-disguised-structured: the classic direct attitude tests of Thurstone, Likert, et al.
2. Non-disgulsed-structured: the free-response Interview and questionnaire approaches, the biographical and essay studies.
3. Disguised-non-structured: the typical projective techniques.
4. Disguised-structured: tests which approximate the objective testing of attitudes.***
In reviewing the work that has been done in the development of attitude scales since 1925, it would be useful to refer occasionally to these classifications of attitudinal measures and to evaluate contributions of different experimenters in light of the type of measure developed.
Bogardus social distance scale. This pioneer scale in the29measurement of attitudes was designed by E. S. Bpgardus in 1925 to
determine the social acceptability of given nationality or racial groups. Persons using the scale were instructed to check whether the individuals would be accepted as (1) visitors to the country,(2) citizens of the nation, (3) fellow employees, (4) neighbors,(5) personal friends, or (6) relatives through marriage. If a respondent were to check number five it is evident that he considers the person or group as acceptable in all of the preceding categories.
^Donald Campbell, "The Indirect Assessment of Social Attitudes ," Psychological Bulletin 1950, 47, p . 15.
2%!.. S. Bogardus, "Measuring Social Distance," Journal of Applied Sociology 9 , 1925 , pp . 299-308 .
19
This type of analysis makes use of the cumulative technique later modified by Guttman. The Bogardus scale is a type of non-disguised- structured scale, as were most of the attitude measures developed
prior to 1940.30Hartshorne and Mav. These men developed a scale for measur
ing dishonesty based upon the supposition that previous experiences and/or heredity will create a certain "predisposition" or attitude toward dishonesty or cheating. They utilized seven tests — an information test, disarranged sentences test, reading scale, sentence completion test, original spelling test, word knowledge test, and an arithmetic test — arranged in order from the one on which most children cheated to the one on which the fewest children cheated. Pupils who would cheat on the least tempting test would also cheat on the other tests. This type of test, when categorized according to Campbell's classification, would be considered a disguised- structured test of attitudes toward cheating.
Lund. In an attempt to determine if belief has emotional content and to examine the relation of belief to concepts of power and knowledge, a type of simple graphic rating was developed. Thirty
(ft
propositions from the areas of politics, religion, ethics and science were prepared such as the following:
Hark May and Hugh Hartshorne, "First Steps Toward A Scale for Measuring Attitudes," Journal of Educational Psychology 17, 1926, pp. 145-162.
20
1. Has Lincoln an honest and upright man?2. Did the whale swallow Jonah?’!
Subjects then reacted to these items on three scales: belief, certainty and desire. Persons completing the scales were instructed to place a check mark beside the statement on the scale which was most descriptive of their feelings. The scales ranged from a maximum of /10 which denoted strong belief, certainty or desire to a minimum of -10 expressing disbelief, uncertainty or lack of desire. After completing each of the three preceding scales, subjects were asked to determine to what extent, (1) indisputable evidence, and (2) opinion, conditioned their reactions to the statement. Each subject assigned a percentage figure to each of the categories. Lund found a correlation of / .42 between belief and evidence and a correlation of / .64 between belief and knowledge. There was virtually no relationship between desire and evidence.
Thurstone. With the development of their scale for the measure-32ment of attitude toward the Church, Thurstone and Chave gave impetus
and direction to the attitude measuring movement. The method that they used is termed the equal-appearing interval approach to scale construction. In developing the scale, several groups of people were
^Frederick H. Lund, "The Psychology of Belief," Journal of Abnormal and Social Psychology 20, 1925, pp. 63-81.
^2L. L. Thurstone and E. J. Chave, The Measurement of Attitude (Chicago: University of Chicago Press, 1929), 96 pp.
21
asked to write their opinions about the church. A survey of literature yielded additional material. A list of some 130 statements was then prepared, the statements expressing attitudes ranging from strongly opposed to highly favorable toward the Church. These statements were then placed on slips of paper and subjects were asked to sort them into 11 piles. The piles were lettered from A to K with A being most favorable and the subjects placed the statements in intervals ranging from high appreciation of the Church to strong depreciation. Thus the subjects placed the statements in equal appearing intervals. Item values were determined by computing the mean score which was achieved through sorting the statements into piles. Forty-five items were retained for use in the experimental scale and they had mean scale values ranging from near zero to 11. The experimenters attempted to distribute the items evenly so that they were uniformly graduated. When the test was administered, subjects were instructed to check those items which were applicable to them. In initial application, the test was administered to 200 Freshmen at the Universit/ of Chicago. Reliability of the test, as computed by the split-half method was /.89 and validity, as determined by the relation of scores to information listed by the testee in blanks at the top of the scale, was satisfactory.
Because of the acceptance of this new method of equal appearing intervals and its apparent usefulness, Thurstone constructed additional scales on subh subjects as prohibition, communism, patriotism,
22
God, capital punishment, birth control, Chinese, Germans and censorship. Thurstone'8 method has been criticized for involving too much time in developing items which can be used and because judges might affect the mean score of the item. However, the first criticism seems slightly irrelevant because preparation of any valid and reliablescale demands considerable time in item development and validation.
33The second criticism has been refuted by L. W. Ferguson and Hinckley^ when they point out that judges effect upon the construction of attitude scales has appeared negligible.
Since the Thurstone instruments were the first serious attempts at measurement of social attitude, they have served as the criterion for validation of other types of attitude measures. Because they are on the non-disguised-structured type of scale, they are open to the criticism that rater bias and knowledge of the quality being measured
may have an adverse effect upon results.Woodward. An early attempt to measure shift of audience opin-
35ion as a result of communication stimuli utilized a debate ballot.
•^L. W. Ferguson, "The Influence of Individual Attitudes on Construction of an Attitude Scale," Journal of Social Psychology.6, 1935, pp. 115-117.
D. Hinckley, "The Influence of Individual Opinion on Construction of an Attitude Scale," Journal of Social Psychology.3, 1932, pp. 283-296.
"^Harold S. Woodward, "Measurement and Analysis of Audience Opinion." The Quarterly Journal of Speech. 14, 1928, pp. 94-111.
23
Persons listening to a debate completed one-half of a form indicating whether they agreed with, were indifferent to, or disagreed with the proposition to be debated. Following the debate, subjects completed the second half of the form, thus indicating whether they had stronger feelings for the affirmative than before the debate, were indifferent as a result of the debate, or were more in agreement with the negative as a result of the debate. Changes in opinion were determined by comparing the pre-debate response with the post-debate reactions.
Likert. Renesio Likert developed his method of summated ratings in 1932. This approach used five categories of response which described the respondents reactions: strongly approve, approve, undecided, disapprove, or strongly disapprove. Categories were scored from fivefor strongly approve to one for strongly disapprove. When Likert
36developed his first scale he culled items from books, magazines, and other questionnaires used in attitude measurement. His materials were in the areas of international relations, race relations and economic conflict and items in all of these areas were combined into a single lengthy scale. This scale not only made use of items which were responded to by the summated ratings technique but it also had yes or no and multiple choice type questions.
36Renesis Likert, "A Technique for the Measurement of Attitudes," Archives of Psychology. No. 140, 1932.
24
When making use of the method of summated ratings, the. higher the individual score the more favorable the attitude. The discriminatory effect of individual items could be determined by statistically measuring the relationship of individual item scores to the individuals total score and the most discriminating items are retained for use in the final scale.
Likert compared the results obtained with his scale with thoseof Thurstone and found reliability coefficients ranging from /.90 onthe internationalism scale to /.79 in the imperialism scale. Thistype of scale is placed in the non-disguised-structured category and,
37as Edwards and Kenney mention, can be prepared in about one-half the time required to construct a Thurstone type scale. Additionally, Edwards and Kenney state that the Likert type scale yields reliability coefficients as high or higher than the Thurstone method, but they observe that Likert items tend to represent extremes of judgment.
Likert presents several criteria for the selection of statements for an attitude scale:
1. It is essential that all statements be expressions of desired behavior and not statements of fact.
2. . . .state each proposition in clear, concise, straight-forward statements.
3?Allen L.- Edwards and Kathryn Kenney, "A Comparison of the Thurstone and Likert Techniques of Attitude Scale Construction," Journal of Applied Psychology. 30, 1946, pp. 72-83.
DIRECT AND INDIRECT ATTITUDE SCALE MEASUREMENTS OF POSITIVE
AND NEGATIVE ARGUMENTATIVE COMMUNICATIONS
DISSERTATION
Presented in Partial Fulfillment of the Requirement for the Degree Doctor of Philosophy in the
Graduate School of The Ohio State University
By
James William Gibson, B.A., M.A.
The Ohio State University 1962
Approved by
Adviser Department of
25
3. In general It would seem desirable to have each statement so worded that the modal reaction to It is. approximately in the middle of possible responses.
4. If multiple choice statements are used, the different alternatives should involve only a single attitude variable and not several.
Guttman. The type of scaling making use of the cumulativetechnique was introduced by Guttman in 1944. This approach had beenutilized earlier in attitude measurement by Bogardus and Hartshorne
*and May, but it was Guttman who provided a formal structure for the technique. Based upon the concept of a universe of attributes and the postulate that the ordering of individuals based on a sample of items will be approximately the same as that based on the universe, the Guttman technique attempts to separate the universe into basic types. As Guttman points out,^ if the highest rank of categories one to five were five, that rank could be subdivided into categories as could the other ranks, but all categories of rank five would remain higher than all categories of rank four. It is also Important to note that if the subject marked yes to number three he would, of necessity, have marked yes to numbers one and two. An item which exemplified this approach asks, Are you taller than: (1) 5'0",(2) 5'2", (3) 5'4", (4) 5'9", (5) 6'2"? This technique gives
^®Likert, op. cit., pp. 44-46.Guttman, op. cit., p. 81.
26
particular emphasis to scale analysis and statistically assesses the relationship of variables appearing on the scale to actual intervals. It is stressed by Guttman that content alone defines the attitude universe and that scale analysis does not judge content but presumes the universe of content is already d e f i n e d . T h e Guttman scalogram analysis technique also is a non-disguised approach and makes more use of mathematical techniques than do the other direct attitude measuring techniques. McNemar mentions that the, " . . . chief merit of the Guttman method is that it leads to the elimination of items that are not on the principal continuum.
Renners. Generalized attitude scales were first developed by R e m m e r s ^ using a combination of the techniques of Thurstone and
Likert. The collection of opinion statements is first given to judges for sorting into equal appearing intervals. When the sorting is completed, distributions for each statement are computed with two statistics being given primary attention. The median and the distance between the twenty-fifth and seventy-fifth percentiles or Q are of importance. If the Q is small it indicates the statement was not
40Ibid.. p. 85.^McNemar, op. cit.. p. 311.^H. H. Remmers, Introduction to Opinion and Attitude Measure
ment (New York: Harper and Brothers, 1954).
27
considered ambiguous by the judges. In constructing the scale at least 20 items with low Q values and carying medians are selected.The scale, in its form for administration, would appear as follows: Directions:
Following is a list of statements about institutions.Place a plus sign (/) before each statement with which you agree with reference to the institution art institutions listed at the left of the statements. The person in charge will tell you the institutions or institutions to write in at the head of the columns to the left of the statements. Your score will in no way affect your grade in any course.
Institution
1. The world could not exist without this institution.
2. Is an ideal institution.3. Has done more for society than
any other institution.4. Benefits everybody.5. Has more good points than any
other institution.6. Appeals to man's highest
nature.
It can readily be noted that this type of scale could be used
for determination of attitude toward any institution. Additional generalized scales have been developed by Remmers to measure attitudes toward any: social institution, racial or national group, practice, vocation, school subject, teacher, proposed social action or home making activity. The aim of the Remmers generalized scales
43Ibid.. p. 91.
28
is ideal but its shortcoming is pointed up by McNemar who mentions that to make the Remmers scale applicable to phenomena in any class, the person is asked to keep a particular object in mind. However, because of the generalized nature of the instrument this procedure can become ridiculous in some cases.^
Indirect attitude measuresThe value and rationale of the indirect methods of attitude
45measurement is discussed by Wechsler and Bernberg when they mention that direct measurement of attitudes often deals merely with verbal content and fails to tap the comprehensive aspects of personality. Those persons who are requested, via interview or through the use of direct attitude measuring devices, to express their feelings mayattempt to conform to the socially accepted view or they may indicateneutrality so as to eliminate psychological conflicts. As a means of permitting accurate measurement of the attitude universe while not disturbing or destroying the implicit, drive producing response, the indirect methods of attitude measurement were developed. These instruments were of two types: (1) the disguised-non-structured or moreprojective techniques and (2) disguised-structured instruments. The projective techniques will be examined first and then the disguised- structured devices.
^McNemar. op. cit., p. 305.^Wechsler and Bernberg, op. cit., p. 210.
29
Horowitz faces test. This test^ uses pictures of eight Negro and four white boys ranging in age from five to 12 years. The faces
are identified by letters from A to L and the subjects are asked to perform two tasks: (1) Rank all faces for preference and (2) answer
sociometric questions such as, "Show me all those you want to be in your class at school." By summing the ranks from the first portion
of the test and adding the scores from the second section, a prejudice score can be determined. Since the validity of any attitude instrument can be determined by actions of testees in social situa^
tions similar to those measured by the instrument, such validation was attempted. Mussen^ states that validity of the test was es
tablished because the boys who showed a great degree of prejudice on the tests also demonstrated more prejudice in a situation in
volving a choice of cabin mates at a bi-racial camp. Reliability, as computed by the split-half method was /.88.
Proshansky. This projective technique^® used a series of pictures which were ambiguous with respect to outcome as far as labor
46Eugene E. Levitt, "The Development of Prejudice: A Critique of the Horowitz Faces Test," Child Development. 27 No. 2, (June1956), pp. 155-171.
^Mussen, loc. cit.HaroId M. Proshansky, "A Projective Method for the Study of
Attitudes," Journal of Abnormal and Social Psychology. 38, 1943, pp. 393-395.
30
and the laboring class was concerned. The pictures were then prepared as slides and each slide was then exposed for five seconds.The subjects were instructed to write for two and one-half minutes on what they thought the pictures represented. The responses to the pictures were evaluated by three judges and these results correlated from /.67 to /.87 with another attitude scale completed by the same subjects.
Fromme. A sample of subjects from the New York City area, representing different socioeconomic and educational levels was used. They first completed a questionnaire on methods of preventing war andthen they participated in an informal discussion of the items on thequestionnaire. The subjects were then shown a series of cartoons, such as the following:
1. A picture of Europe represented by a woman danglingfrom a rope, with the following captions beneath it:
(a) Feat: Escape from oneself.(b) I'll be all right as soon as Uncle Sam sees
my distress.(c) That rope tangles easily into a noose.(d) Acrobatics promoted by whom?^
The subjects were to select the caption they thought most appropriate. Following this the subjects were shown five pictures for thematic apperception and were asked to make up a story as an illustration. These procedures offered a projective method for study of the individuals attitude toward the prevention of war.
^Allan Fromme, "On the Use of Certain Qualitative Methods of Attitude Research: A Study of Opinions on the Methods of Preventing War," Journal of Social Psychology. 13, 1941, pp. 437-438.
31
Baruch. A technique for measurement of a child's aggression toward other members of his family and, hence, his attitude toward the family group used dolls which represented the parents and brothers or s i s t e r s . T h e child was given the dolls and was first asked to name them. If he could not do so the experimenter would name them for him. Some items of4toy furniture were presented to create a more
homelike atmosphere. In their play children exhibited various forms of aggression, such as beating the doll representing the father, and this aggression was usually confined to a single member of the family. The majority of children exhibited aggression but admission of aggressive feelings in post-play was greatly varied.
Dubin. Eighty toys representing war, public service, labor, and entertainment were presented to ten subjects. They were instructed to arrange the toys to represent the world as they perceived it and, in a second situation, as they would like to see it. The subjects were interviewed after they had arranged the toys and their arrangement was discussed. Graduate psychology students were then informed of the subjects "picture of the world" and were asked to predict, on a five point scale, the responses of the subjects to a Likert type scale on perception of the world. The coefficient of
^Dorothy W. Baruch, "Aggression During Doll Play in a Preschool," American Journal of Orthopsychiatry, 11, 1941, pp. 252-259.
32
correlation of /.49 between scores predicted for the subjects andscores actually made indicates this projective technique has con-
51siderable utility.Allport and Postman. In an experimental approach to the study
52of rumor, these men selected six or seven subjects from an audience and asked them to leave the room. The only instructions they received were to listen carefully to what they would hear when they returned to the room. After these subjects left the room, a slide showing a detailed situation was flashed on the screen and a member of the audience was selected to describe it to the first subject. The first
subject was recalled to the room and was placed in a position where he could not see the screen. He listened to the account of the audience member and was then instructed to repeat what he heard as accurately as possible to the second subject who would be seated beside him. This procedure was continued until all subjects reported the scene. .Allport and Postman found that distortions in transmission
51S. Sanford Dubin, "Verbal Attitude Scores from Responses Obtained in the Projective Technique," Sociometry. 3, 1940, pp.24-48.
^Gordon W. .Allport and Leo Postman, The Psychology of Rumor (New York: Henry Holt and Co., 1947),
33
were influenced by the attitudes operating in the situation and that these attitudes can be examined by means of the subjects expressed and repressed verbal behavior.
Tendler. An instrument utilizing the sentence completion technique was developed by A. D. T e n d l e r ^ in the belief that stimuli
will arouse a particular emotional set while still permitting free response. Subjects were first asked to write character sketches of themselves and then their responses to the stimulus stems were compared with attitudes and traits described in the autobiographical sketch. Tendler noted that the same stimuli, such as, "My heroi s 11 and "I get angry when ", evoked different responsesfrom different subjects and he stated that these responses may be the key to the individuals fears, interests and attitudes.
Hammond. A slightly different indirect technique which could be classified as a disguised-structured attitude measuring instrument was developed by Kenneth R. Hammond.^ The technique operates on the belief that error-choice will
. . . measure the effect of attitude, herein considered to be a (non-primary) source of energy, or affective state, capable of distorting perception and recall with
^A. D. Tendler, "A Preliminary Report on a Test for Emotional Insight." Journal of Applied Psychology. 14, 1930, pp. 122-136.
^Kenneth R. Hammond, "Measuring Attitudes by Error Choice:An Indirect Method," Journal of Abnormal and Social Psychology.43, 1948, pp. 38-48.
34
reference to an unstructured, ambiguous world of social events, by measuring the constancy of the direction of the error into which the respondent will be forced.$5Hammond presented several series of items to the subjects and
termed these tests information exams. In one series of eight questions, determinable answers which were equidistant from the truth were offered. In a second series, where the truth could not be determined, alternative answers were presented. A third series consisted of straight information questions. By making the respondent choose errors, especially in cases where the truth could not be determined, the direction of the bias and attitude of the subject could be sampled. Hammond found that responses differed when the test was presented as an attitude test instead of an information test and hence this instrument eliminates the factor of "attitude test set".
McGregor. .Another disguised-structured approach to attitude measurement asserts that an individuals wishes, attitudes and knowledge relative to a given social situation directly influence his
56predictions of situations based upon these attitudes. Douglas McGregor used questions in which subjects were asked to predict actions, group
55Ibid.. p. 39.“*̂ Douglas McGregor, "The Major Determinants of the Prediction
of Social Events," Journal of Abnormal and Social Psychology. 33,1938, pp. 179-204.
CONTENTS
Chapter Page
I. THE PROBLEM............... ‘....................... 1II. REVIEW OF THE LITERATURE......................... 7
III. PROCEDURE IN SCALE CONSTRUCTION.................. 41IV. PROCEDURE AND DATA ANALYSIS .................... 64V. SUMMARY AND CONCLUSIONS.......................... 86
AppendixA THURSTONE ATTITUDE SCALE ........................ 91B FORCED-CHOICE ATTITUDE SCALE .................... 96C SCORE CARD FOR FORCED-CHOICE S C A L E .............. 100D EXPERIMENTAL COMMUNICATION STIMULI .............. 105E CRITERIA FOR EVALUATION OF EXPERIMENTAL STIMULI . . Ill
BIBLIOGRAPHY ............................................. 113
ii
35
memberships, etc., with the variables of: (1) the degree of ambiguityin the stimulus situation, and (2) the importance attributed to the event by the respondent being determinants of the influence of the stimulus situation. For example, questions such as, "Hill the King of England announce plans for his marriage before May 1937?" and "Will Roosevelt be re-elected in November 1936?"^ were the stimuli. The amount of information possessed by the respondent appeared insignificant but the quality of the information appeared decisive. McGregor found that the attitude of subjects toward, for example, Roosevelt and the King of England, had an effect upon the types of predictions made.
Morgan and Morton. An attitude test using the syllogism and syllogistic reasoning as determinants of respondent attitude was developed at Northwestern University.5® A set of 15 syllogisms on current topics of national interest were used. The test permitted the subject to select one of five alternative conclusions since the major and minor premises were stated. The experimenters found that when issues relative to personal opinions, fears or convictions of the individual appeared in the syllogism, the distortion of the
57Ibid., pp. 183, 191.58John J. B. Morgan and James T. Morton, "The Distortion of
Syllogistic Reasoning Produced by Personal Convictions," The Journal of Social Psychology. 20, 1944, pp. 39-59.
36
conclusion or selection of answer was affected by the atmosphere effect of the syllogism. In essence, the subject felt that he was reasoning logically in selecting the conclusion but his choice was affected by his attitudes and fears.
Parrish. Two indirect attitude tests were developed byParrish, one being presented as an information test and the second
59ostensibly being an estimate of public opinion polls. Because both tests were of such difficulty and remoteness that subjects were unlikely to know the answers, they served as indicators of the respondents attitude. A third scale was used and this instrument was a Likert type scale. After dividing subjects into two groups and presenting one group with a "pro" type communication and the other group with an "anti" communication, the subjects again completed the scales. Results indicated that the communication did affect the respondents scores, i.e., they were higher or lower depending upon the type of communication heard, and hence the scores as determined by both the indirect and direct scales were measurements of the subjects attitudes.
Osgood. A very recent approach to the disguised-structured analysis of attitude measurement has been made aththe University of
CQJack Albert Parrish, "The Direct and Indirect Assessment of Attitudes as Influenced by Propagandized Radio Transcriptions," (Columbus, Ohio, M. A. thesis, The Ohio State University, 1948).
37(- " u
Illinois by Charles Osgood and others^ with the semantic differential . Basing the approach upon the definition of attitude as a learned, implicit process, potentially bi-polar, varying in intensity and mediating evaluative behavior, Osgood used bi-polar adjectives heavily loaded on the evaluative factor such as "good-bad; optimistic- pessimistic." Using six concepts (labor leaders, The Chicago Tribune, Senator Robert Taft, legalized gambling, abstract art and accelerated college programs) and six evaluative scales (good-bad, fair-unfair, valuable-worthless , tasty-distasteful, clean-dirty and pleasant-unpleasant) 135 subjects were tested. Scores were determined by summing scores on the six scales and these results were then compared with attitude scores of the same subjects on Thurstone and Guttman scales. The results of the semantic differential as compared to both scales were significant at .01 level. Osgood mentions that this type of instrument is capable of use as a generalized attitude scale which is not guilty of the same shortcomings as the Remmers scales.
Forced-choice. Recently a new form of the disguised-structured attitude measurement approach has been developed. The forced-choice technique was originated by Wherry®* and forces the respondent to
^Charles Osgood, George Suci and Percy Tannebaum, The Measurement of Meaning (Urbana, Illinois: University of Illinois Press ,1957), p. 189.
^Leonard V. Gordon, "A Comparison of the Validities of the Forced-Choice and Questionnaire Methods in Personality Measurements," (Columbus, PhD dissertation, The Ohio State University, 1951), p . 27.
38
describe rather than rate himself. By pairing items which appear equally desirable or undesirable but which actually differentiate individuals effectively, the placement of the individual or his attitudecan be somewhat objectively determined. The forced-choice technique
62has been used primarily as a rating instrument by educators andgo
management. Its popularity Atas increased because it reduces the effect of such rating problems as halo effect, personal bias and central tendency.
Bartlett. Quay and Wrightsman. These experimenters comparedthe forced-choice and Likert type scales as techniques for attitude
64measurement. They prepared statements which they believed reflected attitudes toward mentally retarded persons and statements having satisfactory factor loadings were used in both the Likert and forced- choice type scales. The scales were administered to 97 attendants at an institution for the mentally retarded. The pre-treatment scores on the Likert and forced-choice scales correlated / .22 which was significant at the .05 level. When the tests were administered following communications designed to develop more positive attitudes
62George Lovell and Charles Haner, "Forced-Choice Applied to College Faculty Rating." Educational and Psychological Measurement.15, p. 303.
63Paul Ross, ,"A Comparison of Two Methods of Matching in Forced-Choice Rating," (Columbus, PhD dissertation, The Ohio State University, 1955).
64Bartlett, Quay and Wrightsman, o p . cit.
39
toward the mentally retarded, scores on the Likert scale Indicated asignificantly positive change (p. .05) In attitude scores while theforced-choice scale showed no significant change. The authors feel,however, that the Likert scale had an Inadequate control for bias and
»that the change noted was merely a change In bias. They suggest that,11. . .one might speculate that the validity of attitude measurement might also be increased through the use of the forced-choice technique,"^ which appears to indicate that the forced-choice technique actually was a more valid indicator of the attitude of the respondents than was the Likert type scale.
Ill. Current Trends In Attitude Measurement
Among the current writers and experimenters in the field of attitude measurement, there seems to be somewhat general agreement that the indirect methods provide a more realistic means of examining social attitudes. The tremendous growth and development of interest in the construction and experimental use of indirect measures of attitude since World War II is indicative of the popularity and high regard in which this approach is held. Some feel, as does Dubin, that projective techniques examine layers of personality that could not be revealed by paper and pencil techniques. Wechsler and
~̂*Ibid., p . 704.66Dubin, ojj.. cit.., p. 27.
40
Bernberg, Rrech and Crutchfield and Donald Campbell agree that indirect measurement of attitudes is necessary in order to preserve the
natural form of the attitude.It would appear reasonable to assume, then, that such indirect
or disguised-structured techniques as the semantic differential, projective techniques and the forced-choice approach hold considerable promise as methods for the more objective measurement of attitudes.
CHAPTER III
PROCEDURE IN SCALE CONSTRUCTION
To develop the forced-choice scale for measurement of atti
tudes toward censorship, a search for descriptive words and criteria was undertaken. Books and periodicals containing articles discussing the theory or problems of censorship were examined and daily newspapers and published public addresses by authorities in the field were perused and pertinent terms extracted. Next, 122 students enrolled in a basic course were asked to write brief essays on censorship. One-half of this number wrote essays on why they thought censorship was undesirable and the other 61 wrote essays on why they felt censorship was desirable. From these papers words and phrases pertinent to the censorship question were extracted and considered for use in construction of the scale. The 226 phrases derived by the above procedure appear in Table 1.
These phrases were then examined for ambiguity and duplication and the list was reduced to 224 items. These phrases were then submitted to another group of 200 undergraduate students. Because the theory of forced-choice calls for description rather than rating, the subjects were first asked to react to the phrases while recalling a person who felt that censorship was desirable. Then the same subjects
42
reacted to the same phrases while recalling a person who considered Censorship undesirable. In describing the probable reactions of these persons to each descriptive phrase, the subjects were instructed to check boxes headed (1) dislike very much, (2) moderately dislike, (3) neutral, neither like or dislike, (4) like, (5) like very much.In preparing the directions for this preliminary operation, the suggestion of Hovland*^ was followed and the crucial portion of the instructions was underlined. The following directions appeared on the cover page of each form:
DIRECTIONS:1. Write your name in the blank provided in the left
comer of the answer sheet.2. Recall a person whom you have known or think of
a person you might have known who felt that censorship was (desirable or undesirable).
3. Keeping this person in mind, indicate his probablereactions to the phrases which appear on the attached sheets. You are to mark these probable reactions on the sheets.
4. If you mark space 1 you are indicating that hewould dislike this description very much. A mark in space 2 would mean that he would moderately dislike this item. Marking space 3 would mean this item was neutral and that he would neither like or dislike it. If you mark space 4 you are indicating he would like this description. A mark in space 5 indicates that the individual considered would like this description very much.
5. Remember that an individual might be favorable toone item and unfavorable toward another. Therefore, consider each item individually as it applies to the person you have in mind.
^Carl Hovland, "A New Method of Increasing the Reliability of the True-False Examination," Journal of Educational Psychology. XXVI (1935).
43
6. Be as accurate as possible and answer each item.7. Now react to the items on the following sheets.
44
TABLE 1
Preliminary Pbrasea Describing Censorship
Pro-Censorship Phrases Anti-Censorship Phrases
1« Assists parents2. Assists world peace3. Aids moral standing4. Arouses opinion5. Accepted procedure6. Builds good morals7. Checks ssait8. Classifies auterlals9. Challenges individuals10. Combats unhealth sex propaganda11. Control for the comun good12. Constructive criticism13. Clean entertainment14. Combats Juvenile delinquency15. Conservative16. Common security17. Christian18. Decisive19. De-emphaslces crime20. Discourages undesirable material21. Distinguishes right from wrong22. Democratic23. Educated24. Eliminates unwanted obscenity25. Eliminates false impressions26. Eliminates 1 moral pictures27. Establishes minimum standards28. Eliminates profanity29. Eliminates objectionable material30. Encourages morality31. Erudite32. Fairly liberal33. Fights communism34. Guardian of morality35. Good sense and moderation36. Good movies37. Good intentions38. Hides obscenity39. Helpful40. Helps society function acre smoothly41. Helps morality42. Helps eliminate impurity43. Historically proved44. Protects future generations45. Helpful to the uninformed46. High morals47. Improvement in the moral tone48. Inevitable49. Just50. Justified51. Keeps mind from being corrupted
1. Un-American2. Authoritarian3. Ill-defined action4. Ambiguous5. A1len to freedom6. Assumes infallible personal judgment7. Assumes lack of intelligence8. Absurd9. Antiquated10. Boycott11. Blacklisting12. Bigoted13. Biased14. Brain washing15. Creates false standards16. Coercive power17. Misguided criticism18. Untrained censora19. Campaign of Intimidation20. Converts adults to children21. Crooked politics22. Concentration of power23. Hinders comsunicatlon24. Stifles curiouslty25. Causes conformity26. Curbs learning27. Creates stereotypes28. Cnninistlc29. Controversial30. Confused31. Dangerous32. Violates democratic traditions33. Dictatorial34. Develops inhibitions35. Deceptive36. Dogmatic37. Distorts the truth38. Destroys freedom39. Discourages individualism40. Evil41. Encourages pornography42. Economic pressure43. Fascist activity44. Fanatical45. Fundamentally wrong46. Foreclosed judgment47. False standards48. Forsakes individual reason49. Fosters Ignorance50. Creates gullibility
.tables
Table Page
1. Preliminary Phrases Describing Censorship ........... 442. Item Indices, Means and Standard Deviations ......... 483. Item Preference and Discrimination I n d i c e s ......... 604. Pre- and Post-Stimulus Mean Scores and Standard
Deviations for Subjects on the Forced-Choice Scale. ...............,.......... 69
5. Pre- and Post-Stimulus Mean Scores and StandardDeviations for Subjects on the ThurstoneScale......................................... 71
6. Attitude Changes of the Control Group Comparedwith Attitude Changes of the ExperimentalG r o u p s ....................................... 74
7. Summary Table for Two-way ClassificationAnalysis of Variance of Forced-ChoiceAttitude Changes ............................. 76
8. Comparison of Attitude Scores as a Result ofExperimental Stimuli ........................ 77
9. Comparison of Attitude Changes of Heterogeneousand Homogeneous Audiences ................ . 78
10. Comparison of Attitude Changes of Heterogeneouswith Homogeneous Audiences.................. 80
iii
TABU 1
Pro-Censorship Phrases
51. Keeps alnd from being corrupted52. Logical53. Legally permissible54. Legal right55. Maintains surality56. Morally desirable57. Maintains nature standards56. Misunderstood59. Objective cossson good60. Opposes pornography and obscenity61. Prudent62. Practical63. Prevents teaptatlon64. Prevents Juvenile delinquency65. Prevents crlnes66. Protects the general public67. Protects children66. Profeselonel competence69. Purging of Communist literature70. Public protection71. Public Interest72. Protects the uneducated nlnd73. Promotes national security74. Prevents security leaks75. Public education76. Pure literature77. Prevents errors78. Public conscience79. Protects against false doctrine80. Protects against trickery81. Protects sural virtues82. Qualified censors83. Rational84. Rejects undesirable material85. Reorganises our sural status86. Right to protest and object87. Reduces antl-soclal development88. Removes false statesants89. Stifles pronography90. Stops sural decline91. Slows, spread of Cossaunlsm92. Self protection93. Social responsibility94. sound in principle95. Safeguards morals ^96. Sober97. Sound decisions98. Temperate99. Trustworthy -100. Vital to national security
45
(Continued)
Antl-Censorshlp Phrases
51. Hateful52. Historically unsound53. Hinders the discovery of new truth54. Hypocritical55. Hsphssard56. Has no criteria57. Intellectual Imprisonment58. Impractical59. Irritating60. Ineffective61. Illogical62. Inconsistent63. Illegal64. Inaccurate65. Intolerant66. Lack of trust67. Limits freedom of thought68. Lowers morale69. Lacks objectivity70. Monotonous71. Modern witchcraft72. Minority rule73. Halve74. Narrow minded75. Mot.essential76. Negative approach77. Not in good taste78. Outmoded79. Obscure80. Picketing81. Penalizes certain groups82. Prevents publication83. Political pressure84. Police authority85. PreJudgment86. Personal ldlosyncraciea87. Puritanical88. Prudish89. Prejudiced90. Prevents comparison91. Prohibits information92. Prevents freadom of choice93. Persecution94. Restrictive95. Rigid96. Radical97. Restricts voicing of opinion98. Social pressure99. Stereotypes material100. Suppression
46
Pro-Censorship Phrases101. Virtuous102. Wholesome entertainment103. Well grounded
TABLE 1 (Continued)
Antl-Censorshlp Phrases
101. Subjective102. Stops discussion103. Senseless104. Social Injustice105. Shelters people106. Sinful107. Thought control108. Tyrannical109. Totalitarian110. Unrepresentative opinion111. Undesirable112. Undemocratic113. Unwarranted114. Unrealistic115. Unsound116. Un-American117. Unproved118. Useloss119. Unjust120. Violates constitutional rights121. Vague standards122. Victorian concept of morality123. Witch hunt.
47
Their reactions to the items were then coded on IBM Data Processing Cards and the cards were sorted and counted by an IBM Card Sorter
and Counting Machine. In this way the responses to the items were determined. Reactions to the phrases were tabulated and preference and discrimination indices were computed. A sample of the statistical procedure followed in determining preference and discrimination indices appears below.Item 48 Subjective MoAlternatives 1 2 3 4 5 ResponseWeight (w) 0 1 2 3 4Frequency (f)
Favorable 14 45 60 56 13 12Unfavorable 20 41 57 50 21 11
I f 34 86 117 106 34 23if.w 0 86 234 318 126d ’F-U 6 4 3 6 8
Preference Index: f.w x 100 - 764.100 « 202~ 377.
Discrimination Index: 27
Mean: 3.053S tandard Deviat ion: 1.107 As an additional measure of variability, the standard devia
tion for each item was computed in addition to the mean. These data were computed for each of the 224 items. Table 2 contains item preference and discrimination indices, means and standard
deviations, arranged in descending order of preference indices.
48
TABLE 2
Item Indices, Means and Standard Deviations
Statements Favorable to CensorshipL
ItemPreferenceIndex
DiscriminationIndex Mean
StandardDeviation
118. Opposes pornography and obscenity 243 239 3.435 1.341
62. Clean entertainaient 239 270 3.394 ' 1.290
105. Good intentions 236 253 3.368 1.175130. Right to protest and object 233 50 3.338 1.223
136. Christian 231 274 3.339 1.344201. Protects children 231 333 3.310 1.39726. Good movies 228 182 3.287 1.330
128. Uholesome4K>tertainaent 227 257 3.272 1.30551. Eliminates unwanted
obscenity 224 - 327 3.240 1.46570. Objective cannon good 223 , 271 3.230 1.315
56. Qualified censors 222 286 3.225 1.280772. Stifles pornography 222 198 3.227 1.281202. Helps morality . 222 339 3.225 1.424
71. Morally desirable 222 319 3.220 1.48623. Combats Juvenile delinquency 220 322 3.207 1.40035. Social responsibility 220 242 3.204 1.134
174. High snrals 220 236 3.205 1.457210. Classifies materials 219 195 3.198 1.023187. Rejects undesirable material 219 279 3.198 1.20295. Encourages swrallty 218 320 3.180 1.39897. Good sense and moderation 218 287 3.181 1.266
«
49
TABLE 2 (Continued)
Statements Favorable to Censorship
ItemPreferenceIndex
DiscriminationIndex Mean
170. CcMibats unhealthy sex 218propaganda
6. Safeguards morals 217
16. Eliminates lassoral pictures 217IS. Aids moral standing 216
197. Assists psrents 216
171. Builds good morals 216
115. Promotes national security 216
91. Discourages undesirablematerial 215
87. Foreclosed Judgment 214
32. Improvement in the moral tone 213
39. Eliminates objectionablematerial 213
88. Prevents security leaks 213
114. Professional competence 213
217. Arouses opinion 213116. Sound lii principle ' 212
168. Protects the general'public 212160. Helpful 211
195. Constructive criticism 211
215. Common security 211
119. Stops sural decline 210
131, Public protection 210209. Maintains morality 210
317
331
307322
313
320
250
312
246
328
276
230
21091
312
326
317
257290
301
311
325
3.187
3.1723.1783.168
3.1673.168
3.164
3.154
3.146
3.138
3.138
3.133 3.135
3.131
3.125
3.125
3.117
3.117 3.116
3.105
3.103
3.105
StandardDeviation
1.475
1.494 1.336
1.399
1.240
1.494
1.321
1.386
1.228
1.334
1.4091.333
1.128
1.0201.422
1.394
1.248
1.2761.1111.421
1.3571.399
soTABLE 2 (Continued)
Statement! Favorable to Cenaorshlp
Item -PreferenceIndex
DiscriminationIndex Mean
StandardDeviation
141. Trustworthy 209 2B0 3.098 1.283
127. Educated 209 265 3.093 1.232122. Prevents juvenile delinquency 209 273 3.098 1.334
176. Protects moral virtues 208 337 3.087 1.312109. Assists world peace 207 237 3.075 1.298
181. Public education 207 192 3.071 1.138
24. Combats juvenile delinquency 206 268 3.061 1.31642. Control for the comaon good 206 329 3.068 1.444113. Protects future generations 206 222 3.067 1.406132. Virtuous 206 271 3.063 1.306
137. Guardian of siorallty 206 315 3.068 1.431182. De-emphasIres crime 206 214 3,063 1.275139. Sound decisions 205 295 3.057 1.29429. Helps eliminate Impurity 205 322 3.052 1.33379. Maintains mature standards 205 303 3.050 1.366164. Self protection ' 205 242 3.055 1.184135. Sober 204 131 3.043 .988178. Slows spread of Cousnunism 203 231 3.035 1.35750. Fights Communism 202 211 3.020 1.36585. Purging of Communist
1lterature 202 127 3.028 1.2039. Protects the uneducated mind 201 257 3.012 1.37640. Legally permissible 201 243 3.012 1.28547. Helps society function more
smoothly 201 300 3.017 1.3671. Prevents low morals 200 302 3.000 1,452
51
TABLE 2 (Continued
Statements Favorable to Censorship
ItemPreferenceIndex
DiscriminationIndex Mean
StandardDeviation
54. Well grounded 200
157. Logical 200189. Public conscience 200
206. Removes false statements 200
76. Protects against falsedoctrine 198
101. Justified 198
82. Practical 197
153. Just 197
219. Keeps mind from beingcorrupted 197
175. Establishes minimum standards 196
161. Pure literature 19757. Reorganizes our moral status 196
138. Rational 196
67. Reduces anti-socialdevelopment 195
125. Eliminates false impressions 193
93. Helpful to the uninformed 192
4. Distinguishes right fromwrong 191
74. Legal right 191
144. Restrictive 191
10B. Penalizes certain groups 191
155. Hides Obscenity 191
289
298150
274
273
317314
311
287
190
190
240
189
102
270
211
253
209
228
249
214
3.000
3.002
3.002
3.000
2.9B7
2.9892.979
2.977
2.979
2.9672.979
2.964
2.964
2.954
2.934
2.924
2.9162.193
2.914
2.914
2.919
1.347
1.3861.096
1.296
1.365
1.440
1.304
1.230
1.453
1.1491.221
1.294
1.246
1.200
1.292
1.200
1.3601.233
1.397
1.300
1.342
i
52
TABLE 2 (Continued)
Statements Favorable to Censorship
Preference Discrimination "StandardItem Index Index Mean Deviation
13. Vital to national security 190 231 2.907 1.347148. Democratic 189 271 2.891 1.425
65. Challenges individuals 189 88 2.893 1.196
30. Fairly liberal 189 189 2.899 1.08418. Decisive 189 136 2.896 .949222. Temperate 187 80 2.870 .995
37. Protects against trickery 186 240 2.864 1.156192. Prudent 184 121 2.847 1.24584. Social pressure 183 88 2.836 1.08794. Eliminates profanity 182 265 2.826 1.281
59. Historically proved 181 243 2.972 1.07944. Accepted procedure 174 227 2.750 1.16425. Erudite 174 53 2.808 .782147. Prevents errors 165 232 2.659 1.20112. Prevents crimes 109 41 2.096 .978180. .Rigid '206 105 3.060 1.13048. Subjective 202 27 3.053 1.107102. Puritanical 200 131 3.000 1.19341. Stifles curlouslty 199 148 2.797 1.20821. Prevents comparisons 198 237 2.982 1.214179. Lacks objectivity 195 283 2.952 1.2715. Historically unsound 194 215 2.948 1.230
111. Prejudgment 193 255 2.930 1.276
53
TABLE 2 (Continued)
Statements Unfavorable to Censorship
Preference Discrimination StandardItem Index Index Mean Deviation
123. Misunderstood 193 71 2.938 1.109
80. Antiquated 192 221 - 2.928 1.242
3. Bigoted 191 125 2.912 1.206
11. Unproved 191 226 2.919 1.239
61. Forsakes individual reason 189 252 2.893 1.267
124. Negative approach 189 232 2.890 1.200
107. Inconsistent 189 299 2.894 1.282
208. Irritating 188 287 2.883 1.25998. Dogmatic 188 144 2.884 1.129
183. Deceptive 188 251 2,544 1.342
86. Foreclosed Judgment 187 178 2.879 1.22345. Personal idlosyncracies 187 127 2.873 1.0008. Minority rule 187 217 2.872 1.372
190. Impractical 187 287 2.870 1.37728. Confuted 187 236 2.873 1.14838. Violates democratic traditions.. W 279 2.877 1.527
27. Unrealistic 186 279 2.862 1.29152. Not essential 186 309 2.868 1.4612. Discourages individualism 185 214 2.857 1.39614. Ineffective 185 284 2.850 1.35864. Biased 185 242 2.850 1.39873. Authoritarian 185 141 2.831 1.23231.. Createa false standards 185 254 2.850 1.40520. Shelters people 184 99 2.849 1.149
I
5ATABLE 2 (Continued)
Statements Unfavorable to Censorship
ItemPreferenceIndex
DiscriminationIndex Mean
StandardDeviation
221. Creates stereotypes ISA
120. Assumes lack of intelligence ISA
165. Prudish ISA
126. Coercive power 18A
66. Hypocritical ISA
207. Restricts voicing of opinion 183
172. Illogical 182
193. Violates constitutionalrights 182
191. Haphazard ** 182
110. Unwarranted 182
75. Concentration of power 182
100. Unrepresentative opinion 182
18A. Hinders communication 182
81. Ambiguous 182
58. Victorian concept of skorallty 181
216, Vague standards 181
69. Prevents publication 181
106. Social injustice 181
99. Stereotyped material 181
159. Ill-defined action 181
10. Outmoded 180
83. Creates gullibility 180
89. Assumes infallible personalJudgment 180
1A9. Naive 180
256
2A5
19A
79
227
283
265
276
297
277
139
266
261
2A0
89
250
121
290
206
258
2A2
2A6
113
209
2.8A9
2.8A3
2.8AI
2.8A6
2.8A9
2.830
2.823
2.826
2.826
2.821
2.670
2.823
2.852
2.829
2.810
2.812
2.817
2.811
2.818
2.813
2.800
2.800
2,80 A
2.800
1.281
1.673
1.2A2
1.1A0
1.320
1.A38
1.37A
1.52A
1.229
1.318
1.228
1.3A2
1.3A0
1.269
1.223
1.111
1.285
1.A19
1.2121.287
1.29A
1.32A
1.3AA
1.257
55TABLE 2 (Continued)
Statements Unfavorable to Censorship
ItemPreferenceIndex
DiscriminationIndex Hean
StandardDeviation
214. Unsound 179
133. Inaccurate 179
173. Undemocratic 179
17. Marrow minded 179
146. Has no criteria 178
33. Undesirable 177
204. Limits freedom of thought 177
213. Prevents freedom of choice 177
186. Totalitarian 177
77. Fundamentally wrong 176
140. False standards 176
198. Fosters Ignorance 176
188. Prohibits Information 176
205. Suppression 176
167. Obscure 176
220. Hinders the discovery ofnew truth 175
158. Campaign of Intimidation 17522, Intellectual imprisonment 174
136. Distorts the truth 174
53. Political pressure 173
68. Absurd 173
104. Monotonous 173
117. Blacklisting 173
301
290236
271
277 288
287
290
250
275
292
261
237
254
193
280
175
262
294
143
290
258
151
2.791
2.8932.790
2.790
2.786
2.778
2.776
2.778
2.771
2.767
2.767 2.764
2.768
2.767
2.760
2.7572.750
2.746
2.743
2.7042.736
2.731
2.737
1.261
1.325
1.543
1.4241.338
1.654
1.5171.552
1.516
1.520
1.434
1.445
1.197
1.433
1.138
1.469
1.219
1.455
1.503
1.063
1.532
1.215
1.244
CHAPTER I
THE PROBLEM
The measurement of attitudes toward social issues, individuals, and institutions has occupied the attention of many psychologists, sociologists, and other researchers in recent years. Their efforts have been directed toward the development of instruments which will reduce to quantifiable terms that latent variable called an attitude. Instruments of several types have appeared and each has professed to sample the universe under examination in either a direct or indirect manner. The classic, direct attitude-measuring instruments require the subject to check on linear continua the extent to which he either endorses or disagrees with the statement made, or subjects are instructed to indicate approval or disapproval of scaled statements. Indirect instruments utilize toys, pictures, and dolls in creative or manipulative situations, and predispositions are inferred from the actions of the subjects.
When the forced-choice approach was developed during World War II,1 it was hailed as the answer to existing rating limitations such
as halo effect, personal bias, and central tendency. This indirect
^Paul Horst, The Prediction of Personal Adjustment. Bulletin No. 48, Social Science Research Council, 1941.
1
56
TABLE 2 (Continued)
Statements Unfavorable to Censorship
Preference Discrimination StandardItem Index Index Mean Deviation
119. Boycott 173 150 2.732 1.105
145. Lack of trust 173 244 2.739 1.304
150. Senseless 172 291 2.722 1.359
7. Tyrannical 172 236 2.725 1.486
162. Fanatical 172 246 2.720 1.315
36. Intolerant 172 202 2.723 1.300
46. Radleal 171 233 . 2.718 1.344
60. Forsakes individual reason 171 218 2.713 1.240
103. Not in good taste 171 272 2.716 1.332
151. Useless 170 324 2.706 1.494
200. Dangerous 170 287 2.701 1.466
196. Un-American 169 277 2.691 1.526
212. Stops discussion 168 240 2.684 1.309
143. Dictatorial 167 278 2.679 1.558
92. Curbs learning 167 266 2.676 1.477
55. Unjust 166 268 2.665 1.479
165. Illegal 165 269 2.657 1.374
49. Prejudiced 164 204 2.649 1.436
121. Destroys freedom 164 274 2.645 1.615
134. Hitch hunt 164 210 2.643 1.297
169. Picketing 163 69 2.639 .975
19. Persecution 160 222 2.609 1.246
211. Modern witchcraft 158 240 2.587 1.330
142. Brainwashing 158 254 2.580 1.463
57
TABLE 2 (Continued)
Stateaenta Unfavorable to Cenaorahlp
Preference Dlacrlalnatlon StandardI ten Index Index Mean Deviation
218. Crooked politlce 154 224 2.543 1.316
112. Hateful 150 277 2.506 1.279
224. Paaclat activity 150 207 2.506 1.371
63. Converta adulta to children 149 204 2.494 1.467
90. Untrained cenaora 146 211 2.460 1.370
203. Evil 145 275 2.458 1.422
96. Comunlatlc 134 248 2.341 1.410
166. Sinful 133 213 2.334 1.273
34. Lovera ax>rale 133 165 2.330 1.234
223. Encouragea pornography 130 175 2.306 1.219
58
TABLE 2 (Continued)
Ambiguous Items
Preference Discrimination StandardItem Index Index Hean Deviation
154. Controversial 219 96 3.199 .925
152. Conservative 203 87 3.030 1.038
177. Police authority 191 58 2.919 1.251
194. Inevitable 167 238 2.677 1.224
163. Causes conformity 190 205 2.904 1.305
43. Develops inhibitions 183 122 2.834 1.093
78. Prevents temptation 182 293 2.827 1.314
129. Economic pressure 179 131 2.791 1.009
59
In constructing the tetrads for the forced-choice instrument,
items with similar preference indices and different discrimination68indices were paired. The recommendation of Brooks to avoid the
pairing of items with extreme differences in discrimination indices was followed in the preparation of the tetrads. Efforts were also taken to prevent inclusion of items with high standard deviations since usage of items of that type might result in the selection of items with unstable preference and discrimination indices. Most of the items used had standard deviations of less than 1.300. The establishment of this level for acceptance of.items resulted in the rejection of all but 70 items. The acceptable positive statements were then arranged in descending order of preference index and each item was tentatively paired with another item of approximately the same preference index and different discrimination index. The identical procedure was followed for the negative statements. This method yielded 24 pairs of favorable statements and 24 pairs of negative statements. The tetrads were then constructed utilizing favorable and unfavorable pairs of similar preference indexes. The 24 tetrads thus constructed constituted the forced-choice instrument. These tetrads and their preference and discrimination indices are shown in
Table 3.
Keith Brooks, "The Construction and Testing of a Forced-Choice Scale for Measuring Speaking Achievement," Unpub. Doctoral Dissertation, The Ohio State University, 1955.
60
TABLE 3
Item Preference and Discrimination Indices
Preference - DiscriminationIndex Index
Prudish 184 194Sober 204 131Confused 187 236Sound Decisions 205 295Inconsistent 189 299Negative Approach 189 232Assists World Peace 207 237Educated 209 265Classifies Materials 219 195Lacks Objectivity 195 283Prevents Comparisons 198 237Good Sense and Moderation 218 287Reduces Anti-Social Development 195 102Author itarian 185 141Pure Literature 197 190Biased 185 242
Vague Standards 181 250Decisive 189 136Stereotyped Material 181 206Distinguishes Right From Wrong 192 211
Practical 197 314Well Grounded 200 289Hypocritical 184 227Restricts Voicing of Opinion 183 283
Prohibits Information 176 237Rational 196 189Unsound 179 301Justified 198 317
De-emphasizes Crime 206 214Deceptive 188 251Public Education 207 192Unrealistic 186 279
61TABLE 3 (Continued)
Item Preference and Discrimination Indices
Preference DiscriminationIndex Index
9. Sinful 133 213Professional Competence 213 210Lowers Morale 133 165Constructive Criticism 211 257
10. Reorganizes our Moral Status 196 240Naive 180 209Establishes Minimum Standards 196 190Outmoded 180 242
11. Rigid 206 105Qualified Censors 222 286Social Responsibility ' 220 242Puritanical 200 131
12. Sound in Principle 212- 312Dangerous 170 287Intolerant 172 202Prevents Security Leaks 213 230
13. Concentration of Power 182- 139Shelters People 184 99Eliminates False Impressions 193 270Helpful to the Uninformed 192 211
14. Creates False Standards 185 254Discourages Individualism 185 214Protects Against False Doctrines 198 273Prevents Low Morals 200 302
15. Senseless 172 291Blacklisting 173 151Common Security 211 290Helpful 211 317
16. Aids Moral Standing 216 322Narrow Minded 179 271Promotes National Security 216 250Creates Gullibility 180 246
62TABLE 3 (Continued)
Item Preference and Discrimination Indices
Preference DiscriminationIndex Index
17. Legally Permissible 201 243Unrealistic 186 279Creates Stereotypes 184 256Removes False Statements 200 274
18. Irritating 188 287Self Protection 205 242Antiquated 192 221Helps Eliminate Impurity 205 322
19. Trustworthy 209 280Monotonous 173 258Forsakes Individual Reason 171 218Maintains Morality 210 325
20. Hides Obscenity 191 214Unjust 166 268Democratic 189 271Prejudiced 164 204
21. Purges Communist Literature 202 127Public Conscience 200 150Dogmatic 188 144Minority Rule 187 217
22. Fairly Liberal 189 189Legal Right 191 209Haphazard 182 297Ambiguous 182 240
23. Not essential 186 309Fights Communism 202 211Ineffective 185 284Maintains Mature Standards 205 303
24. Assists Parents 216 313Rejects Undesirable Material 219 279Prejudgment 193 255Historically Unsound 194 215
63
The final step in preparation of the forced-choice instrument was the preliminary administration of the scale to a small group of subjects to determine the reliability of the scale. The forced-choice instrument was completed by 13 subjects and an odd-even correlation undertaken. The correlation was /.91, corrected for split half.
When construction of the scale was completed an instrument to be used for comparison purposes was selected. Because of its established position in the field of attitude measurement, the Thurstone Attitude Scale for Measurement of Attitudes Toward Censorship was selected. When originally constructed by L. L. Thurstone in 1931 this scale had two forms. Although available material on the two forms of the test asserted that they were equivalent, it was decided to administer both forms to a sample group to ascertain if they were, in fact, equivalent. Fifty undergraduate students completed the two forms and a Pearson Product-Moment Coefficient of Correlation was computed between the scores. The _r corrected by the Spearman-Brown
formula for split half was /.90 for 20 items. This correction was made because it was felt that administration of both forms of the Thurstone scale as a single test, thus providing a scale of 40 items, would be a more valid and reliable measure of the subjects attitude
toward censorship.
CHAPTER IV
PROCEDURE AND DATA ANALYSIS
After the forced-choice Instrument was constructed and adminis
tered to a pilot group to determine Its Internal reliability, It was
used with the Thurstone Instrument for the purpose of measuring atti
tudes toward censorship before and after the presentation of a
communication stimulus. Since the Thurstone instrument is an open
faced-structured type of measuring device, it was expected that the
relationship between scores on the two tests would be positive but
not perfect. The correlation between pre-test scores of subjects on
the two tests was /.67, thus indicating the two instruments seem to
sample the same property.
The Thurstone scale (Appendix A) combined Forms A and B of the
original instrument for Scale of Attitude Toward Censorship and con
sisted of three pages. The cover page presented the instructions
for scale completion:
This is a study of attitudes toward censorship. On the following pages you will find 40 statements expressing different attitudes toward censorship.Put a check mark (V**) if you agree with the statement.Put a cross (X) if you disagree with the statement.If you cannot decide about a statement, you may mark it with a question mark.This is not an examination. People differ in their opinions about censorship.Please indicate your own attitude by a check mark' when you agree and by a cross when you disagree.
64
65
The responses to the statements were scored according to instructions
published by L. L. Thurstone in which he stated that a person's score
is the median scale value of all the statements he endorses. These
scale values had been previously determined by use of the equal-
appearing intervals method.
The forced-choice items were arranged on three sheets with two
columns of boxes after the items&r subject response. The forced-
choice scale is shown in Appendix B. The directions for completion
of the forced-choice scale appear below:
Directions: Check two items in each of the 24groups below as follows:
1. In the group headed MOST, check the one item in each group which is most descriptive of censorship.
2. In the column headed LEAST, check the one item in each group which is least descriptive of censorship.
Item responses were scored /l, 0 and -1, according to the discrimina
tion index of the item in the tetrad. If a statement favorable to
censorship were checked as most descriptive and the item, according
to the discrimination index, discriminated more than the other state
ment favorable to censorship it was scored /l. If the more discrim
inatory favorable item was checked as being least descriptive it was
scored -1. When a less discriminatory favorable item was checked
as most or least descriptive it was scored 0. If a more discrimina
tory unfavorable item was checked as least descriptive it was scored
/l, and if checked as most descriptive it received a value of -1.
66
Less discriminatory unfavorable items were assigned a value of 0 when
selected as most or least descriptive. Therefore, a positive score
on the total forced-choice test indicated a favorable attitude toward
censorship while a negative score meant that the subject felt censor
ship was undesirable. The score card used for marking the forced-
choice instrument appears in Appendix C.
The stimuli used in this study consisted of two communications
on the subject of censorship. One stimulus presented arguments for
freedom and is called the "positive" communication. The other stimu
lus which presented arguments against censorship is termed the
"negative" stimulus. These two communications were submitted in
written form to a group of eight graduate students in speech at The
Ohio State University for content validation. The first page con
tained instructions:
On the following pages you will read two communications designed to be used in an experimental study on censorship. One of these communications is of the "positive" type. That is, it speaks only of the favorable aspects of freedom, emphasizing the accomplishments of free people and the advantage of living in a society which permits its citizens to think and act freely. The other communication is "negative" in that it discusses the disadvantages of censorship and the problems inherent in restricting individual rights in a pluralistic society.
After each communication you will find a sheet with adjectives on bi-polar continue. Considering the communication which you have read, first check in space number one the type of communication.Then check on the continua your reactions to the communications. If you have any specific suggestions to make regarding revisions in the communications, please write in the margins beside the section you feel needs improvement.
67
The stimuli which the graduate students read and the bi-polar adjectives they reacted to appear in Appendix D. Revisions were suggested for eight passages. After these changes had been made, the stimuli were tape recorded for validation of the delivery. Each of the eight evaluators heard both stimuli and indicated their evaluation on the check list appearing in Appendix E. Suggested changes in delivery were made when the speaker re-recorded the validated stimuli for presentation to the experimental groups.
Two hundred and sixty-nine subjects from three Central Ohio *
colleges participated in the experimental procedure. During a regular classroom period, each subject completed the forced-choice instrument and the Thurstone scale. None of the subjects were informed
of the purpose of the study.After the completed tests had been scored, the subjects were
grouped according to their forced-choice scores. Twenty-nine subjects, representative of the pre-test distribution, were selected for
the control group.Three weeks after completing the pre-test, the subjects listened
in pre-determined groups to the stimuli. Each tape recorded communication was heard by three groups of subjects. One group was termed "favorable" because their pre-test scores indicated they would agree with the content of the stimulus. A second group which heard each communication was labelled "opposed" because their pre-test scores
suggested that they would be in disagreement with the point of view
expressed by the stimulus. A third group was termed heterogeneous
or mixed because it contained subjects, some of whose scores indi
cated agreement with the message and others whose pre-test scores
indicated they would disagree with the communication to be heard.
After hearing the stimulus, each subject again completed both the
Thurstone and forced-choice instrument. Approximately one-half of
the subjects listened to the "positive" stimulus and one-half heard
the "negative" communication. The control group did not listen to
either communication but completed the Thurstone and forced-choice
post-tests during the same class period.
Relationship between forced-choice attitude scores and Thurstone attitude scores
To answer the experimental question: What relationship exists
between attitude scores toward censorship obtained with a Thurstone
attitude scale and attitude scores toward censorship obtained with a
forced-choice attitude instrument, the following procedures were
undertaken.
Mean scores were determined for all groups for the pre- and
post-stimulus tests and the group standard deviations were computed.
Table 4 shows the pre- and post-stimulus mean scores and standard
deviations for all groups including the control group on the forced-
choice scale.
69
TABLE 6
Pre- and Post-Stimulus Moan Scores and Standard Deviations for Subjects on the
Forced-Choice Scale
Group
Numberof
Subjects
MeanPre-Test
Post-Test
StandardDeviation
Pre-Teat
Post-Test
All Groups 240 - 6.33 -12.43 17.34 17.17Control Group 29 - 4,28 - 5.90 16.47 19.10
Negative CoanunlcatlonAll Groups 115 - 7.08 -12.91 15.96 14.88
Hoaogeneoualy "Favor" Stimulus 49 -17.84 -21.7B 8,08 8.45Hoaogeneoualy "Oppose" Stimulus 25 tD.es 7 1.68 14.89 18.29Heterogeneous to Stimulus 41 - 6.37 •12.10 13.06 12.10
Positive Coaaunlcatlon
All Groups 125 - 4.95 - 7.40 18.20 18.61Homogeneously "Favor"Stimulus 54 -17.08 -19.78 8.77 10.03Homogeneously "Oppose"Stimulus 30 /13.50 713.53 13.95 10.28Heterogeneous to Stimulus 41 -12.47 -16.40 14.42 14.52
70
Forced-choice scores indicate that all groups exposed to the
stimuli changed in the direction of the stimulus excepting persons
in the "opposed" group who heard the "positive" stimulus. Those
individuals who heard the "negative" stimulus exhibited greater mean
attitudinal changes than persons listening to the "positive" message.
The greatest individual group change was shown by the "opposed" group
hearing the "negative" communication and the smallest change took
place in the "positive opposed" group.
Standard deviations of subjects hearing the "negative" stimulus
decreased, while standard deviations for persons listening to the
"positive" stimulus Increased.
The data for all experimental groups combined on the forced-
choice scale suggests that a much greater change in attitude occurred
as a result of the stimulus than took place in the control group. As
an apparent result of the message, the standard deviation of the sub
jects who heard the stimuli decreased while the control group standard
deviation increased over two and one-half points.
Table 5 shows the pre- and post-stimulus mean scores and stand-%
ard deviations for all groups, including the control group, on the
Thurstone scale.
The Thurstone instrument indicates, as did the forced-choice
scale, that all groups moved in the direction suggested by the
stimulus with the greatest changes occurring in groups hearing the
"negative" communication. The largest attitude changes took place in
71
TABLE S
Pre- and Post-Stimulus Mean Scores and Standard Deviations for Subjects
on the Thurstone Scale
GroupHwaberof
SubjectsMean
Pre-Test
Post-Test
StandardDeviation
Pre-Test
Post-Test
All Groups 240 4.90 4.46 1.42 1.44Control Group 29 4.82 4.66 1.44 1.49
Negative Communication All Groups 115 4.91 4.30 .94 1.26
Homogeneously "Favor" Stimulus 49 4.15 3.65 1.20 1.00Homogeneously "Oppose" Stimulus 25 5.80 5.52 1 .00 1.14Heterogeneous to Stimulus 41 5.03 4.33 1 .09 1.12
Positive Coimsunlcatlon All Groups 125 4.98 4.59 1 .57 1.60
Homogeneously "Favor" Stimulus 54 4.12 3.70 1 .19 1 .19Homogeneously "Oppose" Stlsnilus 30 6.39 6.14 1 .07 1 .03Heterogeneous to Stlsnilus 41 5.18 4.77 1 .04 1.09
72
the heterogeneous group hearing the "negative" communication while
the smallest change appeared in the "opposed" group which heard the
"positive" communication. The standard deviation for the subjects
listening to the "negative" stimulus decreased, indicating greater
group agreement. Greater dispersion of scores occurred in the groups
as an apparent result of the "positive" stimulus.
The control group showed a negative shift in score, but the
change in all experimental groups combined on the Thurstone scale
was greater than the control change. The standard deviation of all
groups and the control group increased slightly.
The data in Tables 4 and 5 on mean changes of subjects from
pre- to post-test on both the Thurstone and forced-choice scales
suggests that the two instruments reflect approximately the same
mean changes in attitude scores. Both scales showed the greatest
change taking place as a result of the "negative" stimulus and stand
ard deviations of scores on both instruments indicated slightly
greater dispersion as a result of the "positive" stimulus and
slightly less dispersion in groups hearing the "negative" stimulus.
To determine if the changes in attitude by the experimental
subjects could be attributed to the communications and not to the
operation of extraneous forces in the three week period between the
pre-test and the introduction of the stimulus, the attitude changes
of the control group were examined. Changes taking place within cer
tain experimental sub-groups as compared to the total control group
73
were also tested. The significance of changes was tested through the
The results of these comparisons appear in Table 6.
The data from the Thurstone and forced-choice instruments on
changes in the control group compared to the experimental groups
shows that the Thurstone instrument indicates a statistically signifi
cant change in the control group from pre- to post-test when no experi
mental stimulation was introduced. This finding of the Thurstone
scale may reflect individual respondent willingness and ability to
report changes in attitude that were induced by the extensive nation
wide discussion of military and civilian censorship during the conduct
of the study. The forced-choice scale indicating a significant differ
ence between the control group and those subjects receiving the
"negative" stimulation is consistent with earlier findings indicating
All £ tests for unrelated measures used the formulae of Allen Edwards..Statistical Analysis . New York: Rinehart and Company, 1958, pp. 131 and 142 which follow:
use of the t for related measures^ and the t.for unrelated measures
69The formula used for the t test for related measures was
70
6 -(fm, -
2
approach attempted to reduce the subjective aspect of earlier ratings and was utilized widely by management for personnel evaluation. Proponents of the forced-choice approach assert that its primary strength lies in its ability to be descriptive. An examination of the philosophy and operation of this approach reveals that it is based on the assumption that description leads to more effective evaluation than evaluation based upon rating alone. If applied to attitude measurement, a forced-choice scale seems to offer opportunities for examination of the latent variable without seriously disturbing or destroying its structure in the measurement process.
Statement of the problem
It is the purpose of this study to construct a forced-choice scale for the measurement of attitudes toward censorship and to determine changes in attitude toward censorship induced by positive and negative communication stimuli. Measurement of the direction and magnitude of the attitude shifts will be made through use of a Thurstone attitude scale and a forced-choice attitude instrument. To achieve this purpose, attention will be given to selection of descriptions of censorship; construction of scale items; validation of the stimuli; selection of criteria for validation of the forced-choice instrument; determination of the relationship between the indirect and direct attitude scales; the effect of audience composition on
attitude change; and differences between positive and negative communi
cation stimuli in inducing attitude change.
74
T A I 1 I . E 6
A t t i t u d e C h a n g e s o f t h e C o n t r o l C r o u p
C o m p a r e d w i t h A t t i t u d e C h a n g e s
o f t h e E x p e r i m e n t a l C r o u p s
Group Scores Compared Instrument Level of Slg.
Control pre-stimulus and Control post-stimulus Forced-choice 1.07 not slg.
Control pre-stimulus and Control post-stimulus Thurstone 2.68 .05
Control changes and "Positive" communication changes
Forced-choice .69 not sig.
Control changes and "Negative" communication changes
Forced-choice 2.08 .05
All Experimental changes and Control changes Forced-choice 1.61 not slg,
Control changes and "Positive" communication changes
Thurstone 1.08 not sig.
Control changes and "Negative" communication changes
Thurstone 1.83 not slg.
All Experimental changes and Control changes Thurstone 1.80 not slg.
I
75
the "negative" stimulus exerted a greater influence upon attitudes. When all experimental changes were combined and compared with the control group changes, the data indicated non-significant changes on both the Thurstone and forced-choice scales. A final conclusion as to the relationship between the Thurstone and forced-choice scale scores id dependent, however, upon data analysis relating to other experimental questions.
Effect of positive and neeative stimuli. To answer the experimental question: Do "positive" type communication stimuli inducegreater attitude changes than coiununication stimuli which are "negative" in structure, the following procedures were undertaken. Because two stimuli were presented, the data were treated with a Two-way, Classification Analysis of Variance to determine if changes as a result of the "positive" and "negative" stimuli were greater than those changes to be expected as a result of chance. The bases for classification of the variables were, (1) the "positive" and "negative" stimuliand, (2) the groups which were initially favorable to or opposed to
71the messages. The formulae of Guilford were used to compute the analysis of variance whose results are summarized in Table 7.
^J. P. Guilford, Fundamental Statistics in Psychology and Education. New York: McGraw-Hill Book Company, 1956, p. 277.
76
TABLE 7
Summary Table for Two-way Classification Analysis of Variance of Forced-Choice Attitude Changes
SourceSum of Squares df Variance
F.ratio
Attitude Scores (Rows) 5331 1 5331 22.95*Communication (columns) 18092 1 18092 77.89*Interaction (rxc) 4269 1 4269 18.38*Within Sets (error term) 35304 152 232 .26Total 155Significant at .01 level of confidence
Since the above findings suggested that the attitude score changeswere greater than those to be expected as a result of chance, the separate group changes were treated with £ tests to determine which communications exerted greater changes upon attitude. Table 8 presents the results of this procedure.
77
TABLE 8Comparison of Attitude Scores as a Result
of Experimental Stimuli
CommunicationsCompared Instrument t Level of Sig.
"Negative" pre-test and
"Negative" post-testForced-choice 5.20 .001
"Positive" pre-test and
"Positive" post-testForced-choice 3.24 .01
"Negative" pre-test and
"Negative" post-testThurstone 5.33 .001
"Positive" pre-test and
"Positive" post-testThurstone 7.H .001
Both instruments indicate statistically significant changes in
the sub-groups as a result of the "positive" and "negative" communi
cation stimuli.The attitude score changes of the persons hearing the "positive"
stimulus were then compared to changes of the groups listening to the negative stimulus. The results of this comparison appear in Table 9.
78
TABLE 9
Comparison of Attitude Changes as a Result of Experimental Stimuli
CommunicationsCompared Instrument t Level of Sig.
"Negative" Communication Changes
and"Positive" Communica
tion Changes
Forced-choice 2.20 .05
"Negative" Communication Changes
and"Positive" Communica
tion Changes
Thurstone 1.66 not sig.
The forced-choice changes as a result of the stimulus show that
the negative communication caused significantly greater attitudinal
changes than did the positive stimulus. However, the Thurstone scale
results do not indicate a significant difference between attitudinal
changes as a result of the "positive" or "negative" communication.
The findings for the forced-choice instrument appear consistent with
earlier results of both scales which indicated that greater attitu
dinal change was induced by the "negative" stimulus. Data from this
study indicates, therefore, that "negative" type communication stimuli
induce greater attitude changes than communication stimuli which are
"positive" in structure.
79
Homogeneous audiences compared with heterogeneous audiences.
An answer to the experimental question: Are changes in attitude by
homogeneously structured audiences as a result of a communication
stimulus different from changes in attitude by heterogeneously struc
tured audiences, was explored through the following procedures.
Since certain experimental groups were homogeneous in attitude and
other groups were mixed or heterogeneous in their attitudes toward
censorship, the j: test for unrelated measures was utilized to deter
mine if attitude change in heterogeneous groups was significantly
greater than changes by persons in homogeneous groups. Table 10r •• ,
presents the result of this examination.
80TABLE 10
Comparison of Attitude Changes of Heterogeneous with Homogeneous Audiences
CommunicationHeard
AudiencesCompared Instrument _t Level of Sig.
"Negative11 Homogeneou sly Favorable and Heterogeneous
Forced-choice 3.95 .001
"Negative" Homogeneously Opposed and Heterogeneous
Forced-choice
2.31 .05
"Positive" Homogeneously Favorable and Heterogeneous
Forced-choice •57 not sig.
"Positive" Homogeneously Opposed and Heterogeneous
Forced-choice
.70 not sig.
"Negative" Homogeneously Favorable and Heterogeneous
Thurstone .37 not sig.
"Negative" Homogeneously Opposed and Heterogeneous
Thurstone .79 not sig.
"Positive" Homogeneously Favorable and Heterogeneous
Thurstone .01 not sig.
"Positive" Homogeneou sly Opposed and Heterogeneous
Thurstone .67 not sig.
The forced-choice data indicate that homogeneously opposed subjects hearing the "negative" stimulus made significantly greater changes in the direction of the stimulus than heterogeneous groups hearing the same message. Heterogeneous groups listening to the "negative" stimulus made significantly greater changes in the direction of the communication than homogeneously favorable subjects.Both changes are in expected directions since persons having greater distance to move in terms of the scale and the message made the greater change. The homogeneity or heterogeneity of the groups had no significant effect on attitude changes for groups hearing the "positive" communication. When all homogeneous group changes were compared with total heterogeneous group changes, the t of .25 indicated there was no statistically significant difference between changes in the groups. Both the "positive" and "negative" stimuli caused attitude shifts in the direction of the communication.
Changes in attitude by homogeneous groups were not significantly different from changes by heterogeneous groups on the basis of Thurstone scores. The t of .37 resulting from a comparison of total heterogeneous and homogeneous changes was also non-significant.
The above results indicate that, although there was a slightly greater change for heterogeneous groups on the forced-choice scale, there is no statistically significant difference between attitude
changes by homogeneous and heterogeneous groups.
82
Absolute changes in attitude. The absolute changes in atti
tude, regardless of direction, for the control group and the experimental groups were compared. The only experimental group showing a change which was significantly greater than the change of the control group was the group which was favorable to censorship but exposed to the "negative" stimulus. The lack of significance for changes between groups may be a result of the fact the control group made small mean changes from pre- to post-test, but many of the individual changes which occurred were large. These changes corresponded in magnitude to attitude changes in the experimental groups. However, the experimental group changes were also rather large but generally in one direction. This evidence lends further weight to findings of the forced-choice instrument that the "negative" stimulus induced significant changes in experimental subjects.
Cognitive consonance and dissonance. To answer the experimental question: What relationship exists between changes in
attitude by persons already in agreement with the stimulus and persons initially in disagreement with the stimulus, the following procedure was utilized. Cognitive consonance involved the presentation of a communication stimulus to experimental subjects who
were initially in agreement with the message. Dissonance existed when the experimental stimulus presented to subjects was in disagreement with their initial attitude toward censorship. The
83
number of attitude changes of consonant and dissonant subjects were72placed in a 2x2 contingency table for a chi sqaure test. The re
sultant value of 1.97, significant at the .20 level, indicates that persons in states of cognitive dissonance approached significantly greater changes in attitude in the direction of the communication than did subjects in states of cognitive consonance. This finding is consistent with earlier data regarding changes of homogeneous and heterogeneous audiences which suggested that those subjects who were initially opposed to the stimulus changed more in the direction of the message than those persons in agreement with the communication. Such results suggest that subjects who are initially opposed to the message in spite of their dissonant state have more distance to move, and therefore change more than do persons who are initially in agreement with the message and can only slightly reinforce their
belief.Further analyses of the data on dissonance and consonance was
made to determine if homogeneously consonant or dissonant groups
72Chi Square was computed according to the formula of Sidney Siegel, Nonparametric Statistics New York: McGraw-Hill Book Company, 1956. The formula used was:
jrt Nfa D-gc/-g)* _____= 0W-8 )(C+D)(rh-C)(8*D)
84
changed attitude more than consonant or dissonant subjects in heterogeneous groups. Analysis by use of standard error of proportion and critical ratio indicated that homogeneously consonant subjects made proportionately greater changes in the direction of the message than consonant listeners in a heterogeneous group. Dissonant subjects in a heterogeneous audience made slightly greater changes in the desired direction than dissonant subjects in a homogeneous group. None of these changes were statistically significant. The indication here is that group reinforcement along with consonant and dissonant states apparently operate to influence individual change or inhibition of change.
Individual attitude changes. An analysis of individual as well as group changes in attitude was made. Significant changes by the experimental group were determined on the basis of control group shifts, as described earlier. Of those subjects who favored censorship prior to the stimulus, 26 or 30 per cent made changes in the direction of the stimulus. No subjects in this group moved in a direction opposite to the stimulus. Seventy subjects initially opposed to censorship m$de significant changes as a result of the stimulus. Sixty-four per cent or 45 of these changes were in the direction of the message and 25 or 36 per cent of the subjects made changes which were in opposition to the stimulus (boomerang
effect).
3
2Importance of the study. Attitude, as described by Green, has three levels: (1) elicited verbal attitudes, (2) spontaneous verbalattitudes, and, (3) action attitudes,, This study attempts to tap the elicited verbal attitudes by direct and indirect methods and thus determine whether the experimental stimuli have altered the subject's attitude toward the problem area. Both the forced-choice approach and the Thurstone instrument deal with the affective component of attitude and attempt to evaluate the respondents' feelings about the subject or issue. By considering attitudes as "predispositions to action," the importance of the consideration of these components is clear.
Communication types. The organization of communications has been examined by many experimenters, including Cromwell^ and Sponberg,
2Bert Green, "Attitude Measurement," Handbook of Social Psychology, ed. Gardner Lindzey, Vol. I, Reading, Mass.: Addison-Wesley Publishing Company, 1954, pp. 335-369.
^H. Cromwell, "The Relative Effect On Audience Attitude of the First Versus the Second Argumentative Speech of a Series,"Speech Monographs. 1950, 17, pp. 105-122.
^H. Sponberg, "A Study of the Relative Effectiveness of Climax and Anti-Climac Order in an Argumentative Speech," Speech Monographs. 1946, 13, pp. 35-44.
85
Readability of stimuli. As a follow-up to the earlier ana
lysis of data, the readability of the two experimental stimuli was73analyzed through use of the Flesch formula for testing readability.
According to the Flesch formula, higher scores indicate greater
readability. The reading ease of the "negative" stimulus was 29.46,
or, according to the reading ease norms provided by Flesch, very
difficult. The reading ease of the "positive" stimulus was 65.56
or standard, according to the norms.
Final considerations. The foregoing analyses of data by the
Thurstone and forced-choice scales leads to these conclusions re-■ • ° • <. y ■■garding the comparative effectiveness of the two instruments.
(1) Both instruments reflect approximately the same mean group
changes. (2) The Thurstone scale is sensitive to individual will
ingness to report changes that may be only a conscious desire to
make responses the subject feels are desired. (3) The forced-
choice results, particularly in regard to the effectiveness of
"positive" and negative" stimuli, are consistent with all pre
liminary data of both instruments. The forced-choice instrument
appears to be consistent in its description of the attitude state
and is more sensitive to actual attitudinal shifts.
^Rudolph Flesch, How to Test Readability New York: Harper and Company, 1951.
CHAPTER V
SUMMARY AND CONCLUSIONS
As a result of the procedures followed in the construction
and use of a forced-choice scale and Thurstone scale for measuring
attitudes toward censorship and the comparison of the effects of
"positive" and negative" argumentative stimuli upon experimental
subjects, the following conclusions appear justified:
1. The use of student essays on censorship as a supplement to
descriptive phrases from literature on censorship provided a list
of evaluative statements for the forced-choice scale which repre
sented favorable, neutral and opposed attitudes.
2. The procedure followed for computation of the preference
and discrimination indices was satisfactory. Two hundred students
recalled persons whom they knew or might have known who felt that
censorship was (1) desirable and (2) undesirable. They reacted to
each of the 224 items while considering each person separately.
3. The criteria followed for pairing of items in the tetrads
appeared satisfactory. Items with similar preference indices and
dissimilar discrimination indices were paired. Twenty-four tetrads
were constructed in this manner.
86
87
4. The selection procedure o£ items for the forced-choice instrument and the construction of the scale appears sound. The reliability coefficient of /.91 during preliminary administration indicates high internal consistency in item response.
5. Forms A and B of the Thurstone Scale for Measuring Attitude Toward Censorship are not truly equivalent. Administration of the two forms of the Thurstone scale to a group of 50 subjects resulted in a correlation between scale scores of* /.90. The two forms were'combined to form a test of 40 items and to increase the test
reliability. When using the Thurstone scale or any measuring instrument, preliminary tests should be made to determine if two so-
called equivalent forms of a test are truly equivalent.6. Procedures followed for preparation and validation of the
"positive" and "negative" stimuli resulted in communications which exercised significant influences on the attitudes of the experimental subjects.
7. The control group attitude toward censorship changed from pre- to post-test. On the basis of a t test, the forced-choice instrument did not reflect a statistically significant change in the
attitude of the control group, but the Thurstone scale changes were significant at the .05 level. During conduct of the study, national discussion of censorship dealing with speeches by military officers
88
and consideration of local movie censorship may have caused the attitude shift by the control group and thus render some findings questionable.
8. Both the "positive" and "negative" stimuli caused changes in attitude toward censorship which were statistically significant on the Thurstone and forced-choice instruments. Data from both scales suggested that the "negative" stimulus caused greater changes in attitude than the "positive" communication. Forced-choice data indicated significance at the .05 level. These findings suggest that subjects exposed to a communication will make greater attitude changes in the direction of the message if the stimulus attacks or points out the problems in a proposed course of action. A message which stresses the beneficial aspects of an alternative choice will be less effective. This "negative" approach may result in a considerable number of subjects "boomeranging" or moving in a direction
opposite to the message because they are rebelling at an attack upon their point of view. The "positive" approach may not induce such a magnitude of attitude change, but fewer "boomerangs" may
result.9. The data on comparison of changes of homogeneous groups
and .heterogeneous audiences is inconclusive. Of the ten comparisons of scores made, eight showed greater changes on the part of subjects in heterogeneous groups. Only two of these changes were statistically significant. One of the significant changes indicated
89
greater changes for the heterogeneous group and the other change
indicated a significant change for the homogeneous group. Generally the findings suggest that groups having greater distances to move made the greater changes regardless of homogeneity or heterogeneity. This would indicate that persuasive stimuli exercise greater effects on persons opposed to them because changes in the direction of the message are more easily made than the strengthening of already favorable views.
10. Persons who were initially in disagreement with the commu
nication stimulus made greater changes in the direction of the message than persons initially in agreement with the stimulus.
11. The time required for administration of the forced-choice scale is approximately twice that required to administer the Thurstone scale.
12. The forced-choice scale can be scored much more rapidly and accurately than the Thurstone instrument.
13. The "positive" stimulus was considerably more difficult to read, on the basis of the Flesch formula, than the "negative"
communication.14. The Thurstone and forced-choice scales differ in sensi
tivity in the measurement of attitude. The significance of changes in the control group indicate that the Thurstone scale is more sensitive to the social desirability of reflecting a particular
90
attitude. The forced-choice approach seems to offer a better method for the examination of attitude and attitude changes on social issues where the effect of social desirability on recorded attitude scale measurement can be controlled. In these situations, the tester wishes to avoid receiving responses which are a reflection of a conscious willingness of the testee to reflect a given point of view. The forced-choice appears to be more effective when attempts are made to tap a true picture of the latent variable. It is better able to do this because it attempts evaluation through description while the open-faced-structured instruments are capable of manipulation by the testees because the value of statements is readily determined and persons can rate rather than describe themselves.
92
ATTITUDE TOWARD CENSORSHIP
Write your name here ______________________________________________
This is a study of attitudes toward censorship. On the following pages you will find forty statements expressing different attitudes toward censorship.
Put a check mark ( ) if you agree with thestatement.
Put a cross ( X ) if you disagree with the statement.
If you cannot decide about a statement, you may mark it with a question mark.
This is not an examination. People differ in their opinions about censorship.
Please indicate your own attitude by a check mark when you agree and by a cross when you disagree.
93
PUT A CHECK MARK ( / ) IF YOU AGREE WITH THE STATEMENT.
PUT A CROSS ( X ) IF YOU DISAGREE WITH THE STATEMENT.
1. Censorship is a good thing if there isn't too much of it.
2. The judgment of intelligent people is the only effective censorship.
3. Nobody has the right to distate to me what I shall read.
4. Censorship is needed because most people are unable to judge for themselves.
5. There is much to be said on both sides of the censorship question.
6. Morality varies so much with different places and times that censorship is arbitrary.
7. Censorship is absurd because no two people agree about morality.
8. Our national morality is safeguarded by censorship.
9. The censors are needed, but they go too far.
10. It is a shame that so many fine books and plays have been suppressed by the censors.
11. Censorship is a disgrace to our country.
12. Censorship when reasonably exercised is desirable for morality.
13. Whether censorship is good or not depends entirely upon the censor.
14. People should be allowed to make their own distinctions between good and bad.
15. What we need is more and better censorship.
94*
PUT A CHECK7MARK ( f ) IF YOU AGREE WITH THE STATEMENT.(PUT A GROSS ( X ) IF YOU DISAGREE WITH THE STATEMENT.
) 16. Our system of censorship Isn't perfect but it is better than none.
) 17. The education of public opinion would be a great improvement over censorship.
) 18. Censorship can never be justified in a free country.
) 19. Some authorized power is certainly needed to keep obscene literature in check.
) 20. Censorship can never make people moral.
; 21. I doubt if censorship is wise.
) 22. A truly free people must be allowed to choose their own reading and entertainment.
) 23. We must have censorship to protect the morals of young people.
) 24. The theory of censorship is sound, but censors make a mess of it.
) 25. Only narrow-minded Puritans want censorship.
) 26. The whole theory of censorship is utterly unreasonable.
) 27. Until public taste has been educated, we must continue to have censorship.
) 28. Many of our great literary classics would be suppressed if the censors thought they could get away with it.
) 29. Everything that is printed for publication should,first be examined by government censors.
) 30. Plays and movies should be censored, but the press should be free.
) 31. Censorship has practically no effect on people's morale.
95
PUT A CHECK MARK ( ^ ) IF YOU AGREE WITH THE STATEMENT.
PUT A CROSS ( X ) IF YOU DISAGREE WITH THE STATEMENT.
) 32. Censorship is a gross violation of our constitutional rights.
) 33. Censorship protects those who lack judgment or experience to choose for themselves.
) 34. Censorship is a very difficult problem, and I am not sure how far I think it should go.
) 35. Censorship is a good thing on the whole although it is often absurd.
) 36. Education of the public taste is preferable to censorship.
) 37. Human progress demands free speech and a free press.
) 38. Censorship is effective in raising moral and aesthetic standards.
) 39. Censorship might be warranted if we could get reasonable censors.
) 40. Morality is produced by self-control, not by censorship.
97DESCRIPTION OF CENSORSHIP
NAME ___________________________
Directions: Check two items in each of the twenty four groups belowas follows:
(1) In the column headed MOST, check the one item in each group which is most descriptive of censorship.
(2) In the column headed LEAST, check the one item in each group which is least descriptive of censorship.
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
MOST LEAST1. Prudish............................f ~ Z Z _______________
Sober...................................... ...............Confused .................. .. . ________ _______________Sound Decisions . . . . . ........
2. Inconsistent . . . Negative Approach . Assists World Peace Educated ........
3. Classifies Materials.......... .Lacks Objectivity . . . . . . . . .Prevents Comparisons ............Good Sense and Moderation ........
4. Reduces Aziti-Social Development . .Authoritarian ........ ..........Pure Literature . . ..............Biased ...........................
5. Vague Standards ..................Decisive .........................Stereotyped Material ..........Distinguishes Right from Wrong , .
6. Practical . . . . . . . . . . . . .Well Grounded . * ............* . .Hypocritical . . .......... . .Restricts Voicing of Opinion . . .
and studies by such researchers as Knower and Hartmann have clarified the use of appeals and their effects. The literature in the field makes no mention of the effect of positive and negative types of argumentative communications upon audience predispositions. The positive type of communication involves a message which stresses only the favorable aspects of the issue under consideration while the negative communication presents the disadvantages and weaknesses of the problem area. Although communications organized in these manners are often heard, experimental work has, thus far, not examined the comparative effectiveness of the two approaches.
Cognitive consonance and dissonance. Since 1957 when Festinger^ proposed the theory of cognitive dissonance, researchers have questioned whether discrepant material induces greater change than consonant material. The significance of this approach seems clear when action states are explored. Is attitude change or reinforcement more
5Franklin H. Knower, "Experimental Studies of Changes in Attitudes: I. A Study of the Effect of Oral Argument on Changes of Attitude," Journal of Social Psychology. 1935, 6, pp. 315-347.
^G. W. Hartmann, "A Field Experiment on the Comparative Effectiveness of Emotional and Rational Political Leaflets in Determining Election Results," Journal of Abnormal and Social Psychology. 1953,48, pp. 78-92.
7Leon Festinger, A. Theory of Cognitive Dissonance. Evanston: Row-Peterson, 1957, 291 pp.
MOST98
LEAST
7. Prohibits Information . . . . .Rational.................... ..Unsound ......................Justified .................. .
8. De-emphasizes Crime ........Deceptive . .. . . . . . . . . .Public Education . . . . ; . .Unrealistic ..................
9. Sinful . . . . . . . . . . . .Professional Competence . . . .Lowers Morale....... ... . . .Constructive Criticism . . . .
10. Reorganizes our Moral StatusNaive .......... ..Establishes Minimum Standards . Outmoded . ..................
11. Rigid . . . . . . . . . . . . .Qualified Censors . . ........Social Responsibility . . . . . Puritanical ..................
12. Sound in P r i n c i p l e ..........Dangerous.......... ..Intolerant . . . . ..........Prevents Security Leaks . . . .
13. Concentration of Power . . . .Shelters People . . . . . . . .Eliminates False Impressions Helpful to the Uninformed . . .
14. Creates False Standards . . . .Discourages Individualism . . .Protects Against False Doctrine Prevents Low Morale ..........
13 * Senseless . . . . . . . . . . .Blacklisting . . . . . . . . .Common Security ..............iHelpful.................... .
99
MOST
16. Alda Moral Standing . . ........Narrow Minded . . . . . . . . . . ______Promotes National Security . ,. . .Creates Gullibility ............ _____
17. Legally Permissible .......... ........Unrealistic . ______Creates Stereotypes . . . . . . . _____Removes False Statements . . . . (_____
18. Irritating . . . . ............ ........Self Protection................ ........Antiquated ............ ........Helps Eliminate Impurity . . . . [_____
19. Trustworthy . . . . . . . . . . . ________Monotonous......................... _____Forsakes Individual Reason . . .________Maintains M o r a l i t y .............. L _ _ _ _
20. Hides Obscenity................ ........Unjust.................... _____Democratic........ ....................Prejudiced................ (______
21. Purges Communist Literature . . . ______Public Conscience . ............ ......Dogmatic ............ . . . . . ________Minority Rule ................ [_____
22. Fairly L i b e r a l ................ ........Legal R i g h t .............................Haphazard ..................... .Ambiguous . ................ .
23. Not Essential ................ ........Fights C o m m u n i s m .............. ........Ineffective ................ .. ......Maintains Mature Standards . . .
24. Assists Parents ................ ......Rejects Undesirable Material . .________Prejudgment . . . . . . _____Historically Unsound .......... £_____
LEAST
101
SCORE CARD FOR FORCED-CHOICE SCALE
DIRECTIONS:(1) Place scoring sheet on top of scale so that the two boxes
coincide with the words MOST and LEAST.(2) Count the checked squares which give the plus values and
record total number in the first box, lower right.(3) Count the checked squares which give minus values and
record total number in second box, - lower right corner.(4) Determine difference between boxes and record final score
in third box, lower right corner.
r/- c
/ + i t
K t
'-r/+ r
/+ c
/-f r_
a
(C
r:r
x i+
y i~
1 1 +
S ' -
I Kyi-
S'+l h
I- E r~ — 1 f +
/- L
£Z 3l i +
f + m
/-c/- t+
c :i i -
( _ Z 3 / +
f =ii-■ / -N ,
EOT/
JLVJfcf
HE3 E3 -/y-/
- i r — >H C _ ' ]
y/
- 1 C
t l C
- ' Fy-/L
1
I
r~=n>/t m - /n=3+'
1 t
:_j
3 - /
U +t
+ I C Z
- / m
-+;e-/ ,i—
♦ ' t =
y / p
' / t Z
ifJ:-
J - /
J . + I
E3;/n — j ♦<
i— :~- i -/
I= - ! - /C
n □ □
106
NEGATIVE STIMULUS
Since the. beginning of the United States, various persons and organizations have attempted to censor ideas and forms of expression of citizens to which they object. These advocates of censorship made one of their early appearances in America when Theodore Dreiser's novel. An American Tragedy was banned by the New England Hatch and Ward Society. Earlier censors considered such books as The Scarlet Letter unfit reading for women. They felt that these written materials were obscene. But, in recent times, legal experts and scholars have expressed widely different impressions of the term obscene. For example, an international conference at Geneva on Suppression of the Circulation and Traffic in Obscene Publications accomplished little because the delegates could not agree upon the meaning of obscenity.
The American judiciary has frequently changed its. interpretation of the term obscene. During the early portion of this century, the courts held that any material which created obscene thoughts in the minds of those who read it was obscene. If we were to apply this criterion to literature, such books as The Sun Also Rises and Candide would be banned today. To carry this interpretation to an extreme which was characteristic of many of the advocates of censorship, the plays of William Shakespeare and the Bible should be suppressed because of their language. In fact, such a move did take place in the Eighteenth Century in Great Britian when Twelfth Night was ruled unfit for public viewing. The American courts of this century have not upheld this extreme view of the interpretation of obscene.
Even more disturbing is the ease with which censors sometimes can prevent the sale of a book at a public store. If a child brings home a copy of, for instance, The Grapes of Wrath, you, as a parent, could underline or note the immoral passages and take the book to the nearest policeman. This policeman can then force the dealer to cease sale of the book and remove all the remaining copies from display. Finally, the bookseller can be sued by the parent for offering for sale material containing impure or indecent language. The prosecution must not prove that the entire book is impure or indecent but only that selected passages taken from context are objectionable.
The theory of censorship is that it can prevent the lowering of moral standards. No educated and sensible person would quarrel with this objective if it were possible to prove that it accomplishes this effect without imposing other ill effects. Men such
107
as Aristotle and Oliver Wendell Holmes have maintained that man is free only so long as he may make choices - choices such as whether he desires to read Lady Chatterly’s Lover or Toynbee’s History; to see "Gone With The Wind,11 or "La Dolce Vita." But censorship would deny these choices by not permitting man to be exposed to some of these materials because they could be considered objectionable.
What the censors would do, then, is to deprive us of the right to freely choose that which we wish to examine. What is even more frightening is that this moral censorship might spread to financial and political matters. The effect in these areas is obvious.
In the moral area, various religious and civic groups such as the Legion of Decency, The National Organization for Decent Literature, Citizens Group For Clean Literature and thousands of others are dedicated to the removal of all publications which violate what they consider the established bounds of moral acceptability. If all they did was to suggest to their members the films they should view and the books they should read, no American would question this as one of their rights. But, when they come into bookstores and tell the bookseller that if he retains materials they consider objectionable on his bookshelves they will urge their friends to discontinue buying his products, they have violated individual rights. And, if they impose a boycott on movie houses showing films which present what they deem questionable themes, then they have overstepped the boundaries of legitimate rights of any social or religious group.
In some instances the censorship campaigns have been extended to almost ridiculous extents as in Cleveland in 1953. There the police forced booksellers to withdraw Sigmund Freud's General Introduction to Psychoanalysis because it had a chapter on sex. Then, to show their moral concern with classics, they banned The Golden Abb because they said the title might be offensive. The current controversy over the showing of the film, "The Lovers," offers further evidence that the forces for elimination of free expression are still at work.
Down through the ages, man's progress has been directly related to his opportunity for freedom of expression. Nations such as ancient Greece, France, England and the United States led or are leading the world and, in large measure, this leadership results from their freedom to speak, write and experiment. Authoritarian states like Nazi Germany and Soviet Russia have made
108
military advances but only at the expense of their citizenry whose right to self expression is denied. When nations like Russia find it necessary to so restrict the voice of public opinion that only the state may publish newspapers and operate radio and television stations, then it is apparent that thought control ranks, in their heirarchy, at the top of the list.
It is ironic indeed, that American citizens who have experienced freedom of expression are willing to deprive others of this right. These persons, of the pro-censorship faction, would act as the conscience of the community. Fortunately these people are a small minority of our population but their lack of numbers should not be regarded as an indication that they will exert little influence. Small minorities have, in the past, wielded influence far beyond their size. As examples we can point to the Nazi Party in Germany and the Communist Parties in many of the captive nations.
If we look at the problem in broad and long range terms, it should be obvious to clear thinking people that censorship springs from fear. And those who have not learned by now that fear is our worst enemy had better learn it before it is too late.