THE STRUCTURE OF BELIEFS AMONG SELECTED COLLEGE FRESHMEN Thesis For The Degree of Ed. D. MICHIGAN STATE UNIVERSITY Edwin G. Spacie 1956 U ’ (I 0-169 This is to certify that the thesis entitled The Structure of Beliefs Among Selected College Freshmen presented by Edwin G. Spacie has been accepted towards fulfillment of the requirements for __jkfll____dqpeeinAdminietrative & Educational 9ervices(Guidance and Counseling) THE STRUCTURE OF BELIEFS AMONG SELECTED COLLEGE FRESHMEN by Edwin G. Spacie AN ABSTRACT Submitted to the School of Advanced Graduate Studies of Michigan State University of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF EDUCATION College of Education Guidance and Counselor Training 1956 ~ \ _, f I t/ V \ \‘\ l (,1 77/ ,. ‘ {7 a \\ Approved r11 / : 'Mfi ' 4% fix? Factor analysis was used to study the organization of beliefs among college freshmen who attained "middle scores" on the Inventory of Beliefs. From among 2953 freshmen who entered Michigan State University in the fall of 1954, all those having a score on the Inventory of Beliefs equal to the mean score of the total group, :1; one raw score unit, were ' These comprised 179 men and 80 women, selected as the "1954 population.’ treated separately because of the significant difference in the group means found when divided on sex. As a "second population" 150 men and 100 women were selected by the same criterion from among some 3200 freshmen who had voluntarily attended summer counseling clinics at Michigan State University in 1950, 1951 and 1952. Treating the 120 items of the Inventory of Beliefs as separate tests, the inter-item correlations for the 1954 populations were obtained as tetrachoricr's. The resulting correlation matrices Were factor analyzed, using the Thurstone Centroid method. Factors were rotated in accord with the criteria for simple structure. To demonstrate the non-chance character of the factors calculated, "pseudo-factors" were constructed, each having the same weight distribution as the corresponding factor, but with items and signs determined by a table of random numbers. Keys were made for the factors and pseudo-factors and the populations were re-scored. The t-test was used to estimate the significance of the differences in variance of the resulting score distributions. Lists of items having maximum loading on one factor and minimum loadings on the other were submitted to a panel of judges in an attempt to determine the psychological meanings of the factors. The following results were obtained: 1. Two factors were calculated from each matrix. More factors probably could have been determined. The factors accounted for significantly more variance within each population than did the pseudo-factors. The variance for each factor differed significantly between the two populations, while the variance for each pseudo-factor did not. The judges reached very little agreement on the psychological meanings of the factors. following conclusions were drawn: Most, perhaps all, "middle scorers" have structured belief systems. It is generally these structures and their polar nature which accounts for the middle score. The factors were recognizable in groups separated by two to four years. . . The factors demonstrated some stability in the face of certain social changes . Groups known to differ, but indistinguishable by their total Inventory of Beliefs scores, were differentiated by the factors. THE STRUCTURE OF BELIEFS AMONG SELECTED COLLEGE FRESHMEN by Edwin G. Spacie A DISSERTATION Submitted to the School of Advanced Graduate Studies of Michigan State University of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF EDUCATION College of Education Guidance and Counselor Training 1956 ACKNOWLEDGMENTS The author wishes to express his sincere appreciation to Dr. Walter F. Johnson, under whose supervision this research was con- ceived and executed, for guidance, warm interest and unstinting assist- ance far beyond the duty of a major professor. He is also deeply grateful to Dr. Paul L. Dressel for sharing in the supervision of the research and for invaluable assistance and incisive criticism in its planning and development. For the freely loaned facil- ities of the Board of Examiners, for the data so generously made avail- able and for the faculty appointment which made the study possible, he is also greatly indebted to-Dr. Dressel. The writer is also deeply indebted to Dr. Harold L. Dahnke for his assistance in planning the statistical treatment of the data, for de- signing the International Business Machine procedures used and for assisting in. and supervising their application to the research. Grateful acknowledgment is also due Dr. Willard G. Warrington for reviewing the research at several points in its development and for assisting in the psychological assessment of the factors. Sincere thanks are also given Dr. Walker H. Hill, Dr. John Kidd and Dr. Lewis B. Mayhew for their kind cooperation in judging the psychological content of the factors. Finally, the writer deeply appreciates the valuable assistance of Mrs. Paul F. Young in the computation of the correlations involved and at other points in the statistical deve10pment. E.G.S. I/ Edwin G. Spacie candidate for the degree of Doctor of Education Final Examination, August 1, 1956, 1:30 P. M. , Morrill Hall Dissertation: The Structure of Beliefs Among Selected College Freshmen Outline of Studies Major subject: Guidance (Education) Minor subjects: Administration (Education) Evaluation, Psychology, Mathematics Biographical Items Born, November 28, 1909, Chicago, Illinois Undergraduate Studies, Northern Illinois State Teachers College, 1925-1929 Graduate Studies, The University of Chicago, 1929, 1930, 1939, 1940, 1941; State University of Iowa, 1931, 1932; Michigan State University, 1953-1955 Experience: Electrical Engineer, 1929-1931, 1933-1935; High School Science Teacher, 1935-1938; Instructor, Assistant Professor, Associate Professor, Michigan College of Mining & Technology, 1938-1943.; Student Counselor, Michigan College of Mining and Technology, 1943-1953; Instructor, Michigan State University, 1954-1955; Counselor Trainer and Counseling Consultant, Central Michigan College, since 1955 Member of Phi Delta Kappa, Society of the Sigma Xi TABLE OF CONTENTS CHAPTER I Introduction Purpose of Study Background of the Problem Statement of the Problem Limitations and Scope of the Study Definition of Terms Plan of the Study CH APTER 11 Procedures and Methods Selection of the Population Statistical Design Correlation Coefficient Calculation of Correlations Reliability of Correlations Analysis of Structure Factor Analysis Accuracy Checks The Factors Rotation of the Factors Keys and Pseudo-Factors CHAPTER I II Results and Discussion Results for Men Results for Women Discussion of Results Psychological Meanings of Factors Relation to Main Factors CHAPTER IV Summary and Conclusions The Problem The Sample and Methodology The Findings Conclusions Implications for Further Study OCDCDUINH 12 13 14 16 19 21 25 27 27 28 29 32 34 38 38 40 42 42 43 44 44 APPENDIX 1. Tables 11. Sample Form 111. Inventory of Beliefs Sample Copy of the Inventory of Beliefs Scoring Key for the Inventory of Beliefs LIST OF REFERE NC ES 46 69 70 71 72 II. III. IV. VI. VI 1. VIII. IX. XI. XII. LIST OF TABLES Confidence Limits of Tetrachoric Correlations for Certain Category Divisions Comparisons of Variances Obtained With Each Key: Men Comparisons of Variances Obtained With Each Key: Women Distribution of Inventory of Beliefs Scores for Entire Freshman Class, 1954 Inventory of Beliefs Item Discrimination for Freshman Class, 1954 Number of Responses to Each Item of Inventory of Beliefs by 1954 Population Computation of Correlations Factor Loadings Before Rotation Rotated Factor Loadings Keyed Weights of Factors and Pseudo-Factors Criterion Items Distribution of Scores for 1954— and Second Population With Each Key 20 32 34 46 47 49 52 53 55 60 64 68 CHAPTER I INT RODUC T ION Purpose of Study This investigation is a study of the organization of beliefs expressed by certain college students through a test designed to measure person- ality along a stereotype-non-stereotype axis. The study is primarily concerned with determining the structure of such expressed beliefs among those students whose scores place them near the mean of a population, and is guided by the hypothesis that most individuals attain the mean position by holding organized systems of beliefs. Other hypotheses are possible. For example, most individuals may arrive at their middle position by answering the questions in an unsystematic fashion, reacting to that portion of the subject matter which is unique to each item. Or, mean scores may result from the reactions of persons attempting to maintain a "middle of the road" position in the face of the forced choice technique of the test; such attempts would result in seemingly random answers. The problem seems worthy of investigation because while "stereo- typy" of personality has been demonstrated to be of fundamental im- portance to education, knowledge of the syndrome and its effects is largely limited to the extremes of the axis. An exhaustive review of the literature has been made and has revealed nothing of the structure of the "middle" scorer's belief system beyond guesses, impressions, and some negative results. While no complete review of the literature is attempted here because thorough discussions and bibliographies are readily available elsewhere, (1, 7, 12, 25) a brief review of the re- lated research will describe the background of the problem. Background of the Problem A long series of investigations of prejudice, fascism, and ethnocentrism as aspects of personality began about 1930 and culminated with the publication in 1950 of the Authoritarian Personality by Adorno, et. a1 (1). In these studies and in others, such as those of Rokeach (19, 20, 21), Srole (23), Davis (10), and Christie (5), amount of education appears to be strongly related to the personality factors involved. Rokeach (19) found that ethnocentrism is related to rigidity in problem solving situations. Since problem solving is a concern of education, this finding is of interest to educators. In another series of studies Rokeach (20) found that dogmatism is a personality factor related to ethnocentrism and fascism. An attack on this phase of personality from the educational point of view was made by a committee of the American Council on Education which developed the Inventory of Beliefs (8) as a measure of certain outcomes of General Education. That the test succeeds in measuring -3- the desired educational outcomes was demonstrated by the Cooperative M (12) and by Horowitz (15). These studies also indicate that the variables measured by the Inventory are factors of personality. Other studies show the Inventory of Beliefs correlates highly with the F-scale of the California study (8) and with Rokeach's Dogmatism Scale (22). Stern, Stein and Bloom (25) found extreme scores on the In- ventory of Beliefs to be highly significant predictors of success in certain academic fields and to be significantly related to choice of vocation and to interests as measured by the Strong Vocational Interest Inventory (3). The studies cited above have shown that the Stereotype-Non- Stereotype dimension of the Inventory of Beliefs is related to the Authoritarian-Non-Authoritarian axis of the California Study and to the Dogmatic-Non-Dogmatic scale developed by Rokeach, and that the Inventory of Beliefs shares the status of these tests as measures of certain personality factors. These factors have been shown to be signif- icant in certain forms of problem solving, in predicting academic success in certain areas and in vocational choice, and have also been shown to be affected by education. It seems clear, therefore, that these scales in general and the Inventory of Beliefs in particular are of real importance to education and to educators. In all the studies in this area the investigation of the personality syndromes involved has been limited to the extreme scorers. In the .. 4 .. work of Stern, Stein and Bloom (24, 25, 26) findings were limited to the 16% in each of the "tails" of the distribution because no significant re- lationships were found for the middle 68%. The California Study was based on groups of extreme scorers, and Rokeach's studies of "dogmatism" have followed the same pattern. On this point Adorno et. a1. (1:26) say: Very few 'middle' subjects - the 50% whose scores fall between the high and low quartiles - were interviewed. It was believed that for the understanding of anti-democratic trends the most important first step was to determine the factors which most clearly distinguish one extreme from the other. . . . The in- tensive study of representative middle scorers should form a central part of any future research along the lines of the present study. Since they are more numerous than either extreme, it is essential to know their democratic and anti-democratic poten- tialities. The impression gained from a few interviews with middle scorers, and from the examination of many of their questionnaires is that they are not indifferent or ignorant with respect to the issues of the scales, or lacking in the kinds of motivation or personality traits found in the extremes. In short, they are in no sense categorically different; they are, as it were, made of the same stuff but in different combinations. Allport (2:368) says of these studies of "prejudice:" Any psychologist who admits these basic mechanisms of cat- egorization, displacement, and rationalization to good standing ought to admit the product of their joint operation. Prejudice _i_§_ something, and does something. It is not the invention of liberals. Its importance in society merely adds urgency to what is in any case a basic psychological problem. Allport (2:376) also is concerned about the lack of studies of the middle of the distribution: Most research on the dynamics and structure of personality in relation to prejudice utilizes extreme and contrasting groups. The subjects are chosen from those ‘high' and 'low‘ in prejudice. Median subjects are usually discarded. This heuristic device leads to an overemphasis upon types. We study confirmed bigots and emerge with the concept of an 'authoritarian' personality; -5- or we focus on markedly tolerant people and emerge with the 'productive', ' mature,’ or 'self-actualizing' personality. This procedure, though defensible, leads subtly to the depiction of 'ideal types. ' We sharpen our finding so that the whole complex subject falls a little too readily into a neat 'schema of comprehen- sibility. ' This procedure tempts us to forget the many mixtures that occur in ordinary run-of—the-mill personalities. It thus appears highly probable that the aspects of personality measured by the Inventory of Beliefs are of fundamental importance to education. It appears equally probable that knowledge of organization of beliefs in median subjects is important in the use of the stereotype- non-stereotype syndrome by educators. Statement of the Problem This study is concerned with the organization of beliefs among students whose scores on the Inventory of Beliefs place them in the middle portion of the population. The study is especially concerned with belief structures common to subgroups of the middle portion of the p0pu1ation as opposed to belief structures unique to the individual. It is also con- cerned with the relative permanence of such belief structures with respect to time and social change. This investigation was, therefore, designed to attain the following objectives: 1. To determine whether patterns of belief can be found among the beliefs expressed by college students who obtain middle scores on the Inventory of Beliefs. 2. To determine whether any such belief patterns can be recognized in a second population. 3. To determine whether any changes in such belief patterns can be recognized in populations before and after such events as the rise and fall oF'McCarthyism'l 4. To determine whether any such belief patterns differ between the sexes of the given pepulation. Limitations and Scope of the Study The study is limited to populations drawn from a single educational institution. While some of the usual sampling errors are avoided by the fact that the populations involved in this study are comprised of_a_ll_ members of the given classes, there is reason to question the represent— ativeness of any single university student body. For example, at Michigan State University significant differences are found between the Inventory of Beliefs scores of men and women, while at the University of Chicago this difference does not exist. In discussing this fact Bloom (4) has pointed out that Inventory of Beliefs scores are related to vocational choice; curricular differences between institutions should therefore affect the distributions of Inventory of Beliefs scores. Again, while no scientific comparisions of the populations from which universities draw their students are available, it seems reasonable to suppose that the students from the fundamentalist religious sects of Western Michigan or from the Saginaw Valley, would have differing .. 7 _ beliefs, hence differing Inventory of Beliefs scores from those students from, say, urban Northern Illinois. The study is also limited to certain freshman classes of Michigan State University. While it is usually assumed that freshman classes do not differ significantly from year to year, proof of the validity of the assumption in the area under study is not available and has not been attempted here. The study also has no control over such transient social factors ' interracial tensions, or a resurgence as the rise of "McCarthyism,’ of formal religion. It does, however, include some rough indication of at least one such general social factor since one of the populations in- volved was taken when "McCarthyism" was at its height and the other after its fall. The study is also limited by the difficulty attendant upon determin- ing the psychological meaning of the results of a factor analysis. "Tracer ' which are usually used to ease this difficulty, were impossible tests,' to apply in this case, and psychological interpretation must, therefore, rest on deductions from the test items themselves and on other known relations . Definition of Terms The Inventory of Beliefs is an ideological inventory constructed by the Attitudes Sub- Committee of the Cooperative Study of Evaluation in General Education of the American Council on Education. Items were selected from among several thousand cliches and overgeneralizations gathered by teachers of courses in general education. The basis for sel- ection was the following conceptual framework here adapted from Dressel and Mayhew (12:2170: 1. Ideas and Institutions (ideocentrism) 1.1 Philosophy 1.11 Materialistic, manipulative, power, cynical. 1. 2 Religion 1.21 Mystical, ritualistic, nonpersonal I. 3 Arts 1.31 Romantic, antisensual, anti-intellectual, anticultural 1. 4 Sciences 1.41 Application, limitation, antirational 1. 5 Politics, economics 1. 51 Dependence, adherence to outmoded ideas, distrust, denial of conflicts 2. Social groups (ethnocentrism) 2.1 Out-groups (Negroes, Jews, other minorities, foreigners) 2. 11 Personal characteristics (offensive, immature, threaten- ing, intrusive, seclusive) .. 9 _ 2. 12 Solutions (pseudodemocratic: segregation, limited participation; antidemocratic: elimination, exclusion; insoluble: fatalism, despair, cynicism 2. 2 In-groups (Chauvinism) 2. 21 Uncritical acceptance of values, exclusive pride in memberships, blindness to or dismissal of shortcomings 3. Individuals, interpersonal relations (sociocentrism) 3. 1 Family (parents) ' 3. 11 Irrational acceptance of external authority, unwillingness to assume 3. 16 Personal responsibility, shift of responsibility to others, blaming others for failure (extra-punitive), re- sistance to departure from tradition, depersonalization of ' etc.) relationships, sentimentality ("momism,’ 3. 2 School (teachers) 3. 3 Church (ministers) 3. 4 State (public officials) 3. 5 Business and consumer relations (tradesmen) 3.6 Friends, peers, siblings 4. Self (egocentrism) 4.1 Self-concept, self-evaluation 4. 11 Perception of external world as threatening, as manipulable: submission and agression, rigidity and com- pulsion, superstition and stereotypy, destructiveness and .. 10 - cynicism: free-floating anxiety, preoccupation with health and sex: over-spiritualization, denial, fear, and deperson- alization of sexuality. The stereotype personality would, by definition, be expected to accept these statements while the non-stereotype would be expected to reject them. Counseling clinic students were entering freshman at Michigan State University who had responded to an invitation, extended to all applicants for admission, to attend pre-college counseling clinics during the summer immediately preceding entrance into college. Attendance was purely voluntary. The program has been described in detail by Matteson (17). Goodrich (13) has shown that these students differed scholastically from their fellows who did not accept the clinic invitation. Plan of the Study The present chapter has stated the problem and discussed its background and significance. A complete review of the literature has not been included since thorough reviews are available in the references cited. Also included in this chapter is a discussion of the limitations and scope of the study and a definition of certain terms used. Chapter I I contains a discussion of the statistical design of the study, including the selection, applicability and reliability of techniques used. The chapter also includes a description of the methods by which the _ 11 - techniques were applied. Chapter I I I presents the results of the study and a discussion of their significance. Chapter I V summarizes the study and its results, states certain conclusions drawn from the study and suggests certain implications for further study. The appendix contains certain raw data and the computational steps referred to in the discussion. It also contains samples of certain forms used in the study. A list of references quoted in the text follows the appendix and includes certain works which contain exhaustive bibliographies. CHAPTER I I PROCEDURES AND METHODS Selection of the Population In the fall of 1954 the Department of Social Science, Basic College, began a broad sociological study of the entire class of freshman entering Michigan State University in September 1954. As a part of that study the Board of Examiners administered the Inventory of Beliefs to the entire entering class, 1240 women and 1713 men, and these test results were made available for this study. This population will be called the "1954 population. " The Inventory of Beliefs had also been administered to all freshman attending the Counseling Clinics in 1950, 1951 and 1952 and the original answer sheets for these groups were also made available by the Board of Examiners. These serve as the "second population. " On examining the scores for the 1954 group it was found that the means and standard deviations for men and women were significantly different. 1 It was therefore necessary to deal with the sexes as separate populations . The group about which previous research involving the Inventory of Beliefs has furnished the least information is the 68% of the population falling between the 16th and the 84th percentile; this is the group with 1. Appendix, Table IV \- .. 13 - which the present study is concerned. On examination it appeared highly probable that the personalities receiving scores at, say, the 17th per- centile would be very like the "stereotyped" personalities known to be found in the lower "tail" of the distribution while those at, say, the 83rd percentile would closely resemble the "polar" or "non-stereotyped" groups of the upper "tail. " If any "middle" personality structure exists it would, therefore, appear to increase as scores nearer the mean were examined, be a maximum at the mean and decrease as the "polar" group is approached. Thus if any typical structure (or lack of structure) exists among the middle scorers it would be maximized at the mean. Following this reasoning, it was decided that the subjects for this study should be those students who obtained an Inventory of Beliefs score equal to the mean score of the population. Since it was desired to include those cases not significantly different from the mean, those having a score equal to the mean or the mean I". one raw score unit were included. Since the standard error of the mean for men was found to be approximately 0. 32, the range of scores accepted for the study was approximately the mean 1’3 S. E. Statistical Design If the beliefs expressed by a significant portion of the selected population have a common structure, then that structure will be reflected 2. Appendix, Table IV .. 14 .. in the intercorrelations of the test items, as discussed by Thurstone (27:55). The present statistical design must, therefore, include selection of a suitable correlation coefficient, a method for calculating the cor- relations and a method for determining the structure of the resulting correlation matrix. The design must also include a technique for estimat- ing the efficiency of factors found. Correlation Coefficient The Inventory of Beliefs items are answered on a four-point scale: strongly agree, tend to agree, tend to disagree, and strongly disagree. The scoring, however, is dichotomous, agree - disagree, and so the Pearson product-moment correlation could not be used to describe the correlations between the items. Guilford (14:310) says, "The Pearson coefficient cannot or should not be computed, for example, unless x and y are measured on continuous metric scales . . . " According to Peters and Van Voorhis (18:366) the tetrachoric correlation coefficient must be used". . . where we have'our data in both variables merely in the number of individuals, or the proportion of individuals, in each of two categories. " Guilford (14:332) says, "The tetrachoric requires that both x and y be continuously variable, normally distributed, and linearly related. " He gives an example involving the numbers of students responding "yes" and "no" to two questions in a personality questionnaire, and then says: The assumption of continuity and normality of distribution can be defended as follows: It is unlikely that all who respond 'yes' .. 15 _ do so with equal degree of affirmation. It is similarly unlikely that those who respond 'no' do so with equal degree of negation. It is most likely that either question represents a continuum of behavior extending from strong affirmation at the one extreme to strong negation at the other. Continuity is thus the probable state of affairs, not a real dichotomy. If a continuum is granted, the gen- eral law of unimodal distribution approaching normality in psychological traits may be cited in defence of the other requirement. In the present case, the fact that the students actually used a four- point scale, rather than a dichotomous one, suggested that a ten-point scale or even a continuum would have been utilized by the students if the test makers had seen fit to provide it. The assumption of continuity thus seemed even more strongly supported in this case than in Guilford's example. Further support for the assumption of normality lay in the fact that the total scores for the entire 1954 population approximated a normal distribution; the aspect of personality measured by the total test was, therefore, most probably normally distributed. Each item was known to be strongly related to the test as a whole (8) and so was known to be a measure of the same aspect of personality. Each item score thus represented a measure of a continuous, normally distributed aspect of personality, partially dichotomized by the omission of a mid-point among the answers permitted by the test, and further dichotomized by the scoring procedure. Each pair of such item scores, therefore, was assumed to satisfy the requirements for the tetrachoric correlation as quoted from Guilford. .. 15 - Calculation of Correlations Since 14, 280 correlations were involved, direct calculations were prohibitive, and Thurston's Computing Diagrams (28) were used. Of these tables Thurstone says, (28:1): In applying the multiple factor methods which the writer has recently developed it has been found that certain problems can be handled with tetrachoric correlation coefficients and that the data are not available for the usual product-moment coefficient. This is especially true when the problem involves the correlation of traits about which one had judgments of mere presence or absence of the trait but no scaled measurements of them, . . . It was soon found that when the traits are rather numerous, such as fifty, it becomes almost prohibitive in time and expense to calculate the complete table of tetrachoric coefficients. A table of fifty traits about which one has records of presence or absence in each of a group of subjects would necessitate the computation of 1225 tetrachoric coefficients. While the facilitating tables prepared by Pearson and his students do reduce the labor of compu- tation, it is possible to reduce the labor still more if it is not required that the coefficients be computed with a high degree of accuracy. It is hardly ever advisable to calculate a tetrachoric correlation coefficient to more than two decimals and the second decimal is not to be taken seriously even for a large population. The reason for this is that when either of the two distributions in the four-fold table is cut near the tail, a very slight deviation from normality of the underlying distribution seriously affects the computed value of the coefficient. The probable error of a tetrachoric coefficient is large even under the most favorable circumstances such as when the population is large and when the two distributions are cut near their medians. It is, therefore, of rather slight practical value to be able to compute such a coefficient to more than two decimals. The tetrachoric coefficient should not be computed at all if the division of either or both of the two distributions is near the tail. This is the reason why computing diagrams have not been prepared for values of _a_ which are smaller than . 05. In general the diagrams should not be used for values of a less than . 10. By this criterion five items were dropped for men and seventeen for women. The items thus dropped are also among the items which .. 17 - failed to discriminate for the entire freshmen class. 3 That some items did not discriminate well for this population, after having shown high discrimination when standardized on a national sample of college students, suggests three possibilities: either the population used here is not entirely representative of the standardizing group, or time and its accompanying social changes have affected college students in a fashion capable of reflection by the test, or both. Since Bloom's observation (3) on the relation between vocational choice and the Inventory suggests a reason for expecting this population to be somewhat different from the national population, and since other supporting evidences have already been seen, a difference between Michigan State University freshmen and the national standardizing group seems the most probable cause. To apply Thurstone's Correlation Tables to the problem it was necessary to determine a four-fold table for each correlation; i. e. , to determine first how many agreed with item 1, and of that group, how many agreed with item 2, similarly for those disagreeing with item 1. This goal was accomplished by means of a IBM scoring machine with item counter attachment and a hand-sorting procedure, in the following steps: 1. A count of all items was made by machine using an "agree" key, and the process repeated with a "disagree" key. The results are 3. Appendix, TableV .grhp . .. .43..“ _ 18- shown in the appendix, Table VI. The total for each item was checked against the total number of papers. Whenever an overcount was indicated the papers were examined for over-marks and all such extraneous marks were carefully removed. Cases of two or more answers to each item would also be found by this procedure, but there were no such cases in this population. 2. The smaller total, agree or disagree for each item, was entered in the appropriate box in the right hand column (boxes a and b) of Table VII, Appendix. 3. The population was sorted by hand into two groups, agree and disagree on, say, item k, and the smaller group item-counted on all items again. These results were entered into the appropriate box c on Table VII. This process was repeated until a four-fold table was determined for each item against every other item. Each item count required two trips through the scoring machine for each paper counted and 122 counts were made, so wearing of the answer sheets was a problem. Four c0pies were made of each answer sheet and new sets were substituted whenever the total count failed to equal the total determined on the original count. 4. All items for which the smaller group, agree or disagree, included 10% or less of the population, were discarded because the corresponding correlations could not be determined with sufficient accuracy. .. 19 .- 5. Thurstone's charts were changed so as to enter by item count instead of by percentages. This was done by making suitable scales based on the populations involved and temporarily applying these to the charts. 6. Each correlation was determined by entering the Thurstone charts with the smaller agree-disagree count. 7. Since complete duplication of the 14, 280 correlations for checking purposes seemed prohibitive, 100 items were chosen at random and the r‘ 's re-calculated by using the larger sides of the four- fold tables. In no instance did the second calculation differ from the first by more than 0. 05, which is well within the accuracy to be expected from the method. 8. The correlations were arranged in square matrices and divided into five-column groups which were labeled alphabetically so as to facilitate identification of each row and group when key-punched. Reliability of Correlations Guilford says (14:335): The tetachoric r is less reliable than the Pearson r, being as much as 50 percent more variable. rt is most reliable (I) when N is large, as is true of all statistics, (2) when r is large, as is true of other r's, but also (3) when divisions into two categories are close to the medians. The complete formula for estimating 0h; is entirely too long to be practical so it will not be given here. But when rt: 0, the formula is much simpbr and reads @tJBILSL‘L. (Standard error of a zero tetrachoric r) . . . This is yy' W the most useful of the formulas for’O-rt , at any rate, because it permits testing the null hypothesis. - 20 _ In the present problem the r's found were thus most reliable when the categories for both items were divided at the mean and least reliable when the categories for both items were divided 89%-11%. Application of Guilford's formula to these limiting cases yielded the following results: TABLE 1 Confidence Limits of Tetrachoric Correlations for Certain Category Divisions Category (0.05 level ) T0. Olievel) Division 07“ . 1. 96 Oh . 2. 6 0'13, Men 50-50 0.11 . 0. 22 0. 29 " 89-11 0.21 0.41 0. 54 Women 50-50 0.17 0. 33 0. 44 " 89:11 0. 32 0. 63 0. 78 ; Those correlations found, for categories divided near the mean, to be greater than 0. 29 for men or O. 44 for women were thus significantly different from zero. (14:336) Each row of the correlation matrices was found to contain one or more correlations meeting this criterion of sig- nificance . For men, all rows representing items categorized near the extreme contained one or more correlations meeting the criterion at the 0. 01 level of confidence. For women, all rows representing items categorized near the extreme contained one or more correlations differing from zero at the ’0. 05 level of confidence, and some .. 21 - of these rows contained one or more r's meeting the O. 01 criterion. It thus appeared that a considerable number of the correlations found were significantly different from zero, and that the correlations for men were probably more reliable than those for the women. This is also the result which would be expected from the sample sizes, for Guilford says: For very dependable results, it is recommended that N be at least 200 and preferably 300 when rt is computed. In smaller samples than these, even less than N = 100, a tetrachoric r can be used to test the null hypothesis but it cannot be depended upon to give very accurate estimates of the size of correlation unless r is very large. The number in the sample for men in this case was 179, meeting Guilfords' criterion reasonably well, but N for women was 80, making for rather large standard errors of estimate. However, no practicable increase in sample size would so decrease the standard error of estimate as to make 31L correlations in the matrix lie within the usual limits of confidence; also, there was no reason to expect that a meaningful subset meeting such a criterion could be selected from the matrices at hand. For these reasons the study was carried out with the original correlation matrices, even though they contained some correlations of doubtful Significance. Analysis of Structure Of the two general methods for describing the variance of a test matrix, cluster analysis and factor analysis, the latter was chosen for - 22 _ the present application because it seemed more readily adaptable to the available equipment and because it is much more thoroughly discussed in the literature. From among the various factor analysis techniques available in the literature, the Thurstone Centroid method (27:149) was chosen, again on the grounds of apparent adaptability to the avail- able equipment. It seemed probable that the method of selection of the population would affect the factors to be found; indeed, the method was chosen so as to maximize any structure peculiar to the group near the mean of the population. But the question of what happened to the main factors of the test through this selection remained to be answered, along with the question of the relationship of factors found in the limited population to the main factors of the test. Exclusion of the extremes of the population eliminated those contributing most heavily to the main factors, and so it was to be expected that the main factors would be reduced in proportion to the restrictions placed on the population. The selective criterion used was that the sum of the item scores be a constant. If the correspondence between the items and the main factors is less than perfect, then the criterion requires that the sum of main factors lie in a certain interval, "partial selection on the composite variable" as Thurstone has called it (27:453). To discuss this condition he used a population of boxes, in which the length, width and height of the boxes are the main factors: -23.. We turn next to an even more stringent condition, namely, partial selection on the composite variable S=( 17443;]; ). This is the sum of the three box dimensions . . . [when plotted] the inside triangle shows the given configuration as before [no selection on S}. The composite variable S is in the middle of the configuration at test 10. The new configuration, after partial selection on the composite variable S, is shown by the outside triangular configuration. The new test vectors all lie in the outside triangle, and their locations are determined by the radial lines from the selection variable as before: It can be seen that the simple structure is here again retained after selection and the primary factors are the same, namely, those which are determined by tests 1, 2, and 3. Hence we conclude that partial selection on a linear combination of even all the factors leaves the simple structure, as well as the physical interpretation of the primary factors, invariant. It thus appears that if the criterion of selection used in this problem imposes a partial selection on a linear combination of the main (primary) factors, then the main factors will appear in the analysis, with their relationship to the test and to each other unchanged. On the other hand, the more heavily the items are loaded with the main factors, the more nearly selection on the mean score becomes "total selection." Of this Condition Thurstone says (27:462): However, if the boxes are assembled so as to satisfy the con— ditions of total selection on the composite variable S, then the rank is reduced to 2;[_and on plotting the new configuratiorfl . . . it is seen that the simple structure has been destroyed, so that the primary factors are no longer identifiable. ThurStone concludes (27:470): In partial selection, when the rank of the reduced correlation matrix is invariant, the simple structure is also invariant, but the correlations between the primary parameters are altered. The interpretation remains the same in both groups. In total selection, when the rank of the reduced correlation matrix is reduced by one, the simple structure may be distorted, depending on the relationship of the total selection to the primary .. 24 - factors. The interpretation of the primary factors affected by the total selection may then be obscured . . . For this reason it is well to allow as much variation as possible among the subjects to be investigated, thus practically eliminating the possibility of total selection within the parameters that are to be sought. Previous attempts (9) to factor the Inventory of Beliefs have failed to discover more than a single main factor. It was to be expected, there- fore, that the selection criterion used here would amount to total selection on the main factor or factors, and would prevent such factors from appearing among the present findings. Conversely, those factors which did appear must be those on which total selection had _r_1_4 .J 13 3' 0 .2 l 0 ‘1 Q S l 3 '2 F 0 '73 Lo 0 6 .7 o o .5" v y 7- :‘ F 1 0 0 7 1 5 ’ ‘5" O {f ....... Y ‘78 2 P 0 Li 3 O 0 S 0 l o- o - Y x .X __ P I “f l 3 .1 (I 0 § 0 3 X-.. Y >< _ 75 (O 10 0? 2.0 05:: /O. X X x gm P ac 06 08 .2! 00 x 1% k 1;; P I L L o I ‘4 <3 C 0% X ‘K. Y f “,3 F o o I ‘1‘ l 0 0" ‘7’ 2 ( -70.. III The Inventory of Beliefs Cooperative Study of Evaluation in General Education of the American Council on Education INVENTORY 0F BELIEFS FOR] I This inventory consists of 120 statements which range over a wide variety or topics. As you read each statement you are asked to indicate quickly your agreement or disagreement with it in terms of the key given below. People have different reactions to these statements. This is not a test in which there are "right" and "wrong" answers. What is wanted here is your own quick personal reaction. You should be able to finish taking the inventory in 30 minutes or less. In responding to these statements you will notice that there is no way provided for indicating a neutral position. It is desired that you indicate a tendency toward either agreement or disagreement even though you may prefer to remain undecided. It is important that you respond to every one of the 120 statements. iBelore beginning work please record at the top of your answer sheet (1) your name, (2) date, (3) the :name of your school, (4) your sex, (5) your academic class i.e.. (Freshman, Sophomore, etc.), and (6) the name of this inventory. 'The key you are to use in responding to these statements is reproduced at the top of each page. (Note that you will ngger use the fifth response space on your answer sheet.) Copyright, 1951, by the American Council on Education 10. 11. 12. 13. 14. 15. Key: If you want a thing done right, you have to do it yourself. There are times when a father, as head of the family, must tell the other family members what they can and cannot do. Lowering tariffs to admit more foreign goods into this country lowers our standard of living. Literature should not question the basic moral concepts of society. Reviewers and critics of art, music and liter- ature decide what they like and then force their tastes on the public. Why study the past, when there are so many problems of the present to be solved. . Business men and manufacturers are more impor- tant to society than artists or musicians. . There is little chance for a person to advance in business or industry unless he knows the right people. . Ian has an inherent guide to right and wrong-- his conscience. The main thing about good music is lovely melody. It is only natural and right for each person to think that his family is better than any other. All objective data gathered by unbiased persons indicate that the world and universe are with- out order. Any man can find a Job if he really wants to work. We are finding out today that liberals really are soft-headed, gullible, and potentially dangerous. A man can learn as well by striking out on his own as he can by following the advice of others. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. . I strongly agree or accept the statement. . I tend to agree or accept the statement. I tend to disagree or reject the statement. I strongly disagree or reject the statement. The predictions of economists about the future of business are no better than guesses. Being a successful wife and mother is more a matter of instinct than of training. A person often has to get mad in order to push others into action. There is only one real standard in Judging art works--each to his own taste. Business enterprise, free from government interference, has given us our high stand- ard of living. Nobody can make a million dollars without hurting other people. Anything we do for a good cause is Justi- fied. Public resistance to modern art proves that there is something wrong with it. Sending letters and telegrams to congress- men is mostly a waste of time. Many social problems would be solved if we did not have so many immoral and inferior people. Art which does not tell a human story is empty. You can't do business fits are profits; and not evidence in a law on friendship: pro- good intentions are court. A person has troubles afford to worry about of his own; he can't other people. Books and movies should start dealing with entertaining or uplifting themes instead of the present unpleasant, immoral, or tragic ones. . Children should be made to obey since you have to control them firmly during their formative years. «:L‘ 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. I strongly agree or accept the statement. . I tend to agree or accept the statement. 1 2 3. I tend to disagree or reject the statement. 4 . I strongly disagree or reject the statement. The minds of many youth are being poisoned by bad books. Speak softly, but carry a big stick. Ministers in churches should not preach about economic and political problems. Each man is on his own in life and must deter— mine his own destiny. New machines should be taxed to support the workers they displace. The successful merchant can't allow senti- ment to affect his business decisions. Iinisters who preach socialistic ideas are a disgrace to the church. Labor unions don't appreciate all the advan- tages which business and industries have given them. It's only natural that a person should take advantage of every opportunity to promote his own welfare. We should impose a strong censorship on the morality of books and movies. The poor will always be with us. A person who is incapable of real anger must also be lacking in moral conviction. If we allow more immigrants into this country, we will lower our standard of culture. People who live in the slums have no sense of respectability. We acquire the highest form of freedom when our wishes conform to the will of society. 46. 47 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. Iodern paintings look like something dreamed up in a horrible nightmare. Voting determines whether or not a country is democratic. The government is more interested in winning elections than in the welfare of the people. Peeble-minded people should be sterilized. In our society, a person's first duty is to protect from harm himself and those dear to him. Those who can, do; those who can't, teach. The best government is one which governs least. History shows that every great nation was destroyed when its people became soft and its morals lax. Philosophers on the whole act as if they were superior to ordinary people. A woman who is a wife and mother should not try to work outside the home. We would be better off if people wOuld talk less and work more. In some elections there is not much point in voting because the outcome is fairly certain. The old masters were the only artists who really knew how to draw and paint. lost intellectuals would be lost if they had to make a living in the realistic world of business. You cannot lead a truly happy life without strong moral and religious convictions. 61. 62. 63. 64. 65. Key: If we didn't have strict immigration laws, our country would be flooded with foreigners. When things seem black, a person should not complain, for it may be God's will. liracles have always taken place whenever the need for them has been great enough. Science is infringing upon religion when it attempts to delve into the origin of life itself. A person has to stand up for his rights or people will take advantage of him. 66. A lot of teachers, these days, have radical 67. 68. 69. 70. 71. 72. 73. 74. 75. ideas which need to be carefully watched. Now that America is the leading country in the world, it's only natural that other countries should try to be like us. Most Negroes would become overbearing and dis- agreeable if not kept in their place. Foreign films emphasize sex more than American films do. Our rising divorce rate is a sign that we should return to the values which our grand- parents held. ' Army training will be good for most modern youth because of the strict discipline they will get. When operas are sung in this country they ought to be translated into English. People who say they're religious but don't go to church are just hypocrites. What the country needs, more than laws or politics, is a few fearless and devoted leaders in whom the people can have faith. Pride in craftsmanship and in doing an honest day's work is a rare thing these days. 76 77 78. 79. 80. 81. 82. 83. 84. 85s 86. 87. 88. 89. 90. I strongly agree or accept the statement. . I tend to agree or accept the statement. I tend to disagree or reject the statement. . I strongly disagree or reject the statement. The United States may not have had much exper- ience in international dealings but it is the only nation to which the world can turn for leadership. In practical situations, theory is of very little help. No task is too great or too difficult when we know that God is on our side. A sexual pervert is an insult to humanity and should be punished severely. A lot of science is just using big words to describe things which many people already know through common sense. lanual labor and unskilled jobs seem to fit the Negro mentality and ability better than more skilled or responsible work. A person gets what's coming to him in this life if he doesn't believe in God. Public officials may try to be honest but they are caught in a web of influence which tends to corrupt them. Science makes progress only when it attempts to solve urgent practical problems. lost things in life are governed by forces over which we have no control. Young people today are in general more immoral and irresponsible than young people of previous generations. Americans may tend to be materialistic, but at least they aren't cynical and decadent like most Europeans. The many different kinds of children in school these days force teachers to make a lot of rules and regulations so that things will run smoothly. Jews will marry out of their own religious group whenever they have the chance. The worst danger to real Americanism during the last 50 years has come from foreign ideas and agitators. Key: 1. I strongly agree or accept the statement. 2. I tend to agree or accept the statement. 3. I tend to disagree or reject the statement. 4. I strongly disagree or reject the statement. 91. Europeans criticize the United States for its 106. Members Of religious sects who refuse to materialism but such criticism is only to salute the 11:8 should be punished for their cover up their realization that American cul- lack of patriotism. ture is far superior to their own. 107. Political parties are run by insiders who are 92. The scientist that really counts is the one not concerned vith the public velfare. who turns theories into actic l . pr a use 108. As young people grow up they ought to get over 93. No one can really feel safe when scientists their radical 19°‘S' continue to explore whatever they wish without any social or moral restraint. 109. Negroes have their rights, but it is best to keep them in their own districts and schools 94. Nudist colonies are a threat to the moral life and to prevent too much contact '1th whites. of a nation. 110. The twentieth century has not had leaders with the vision and capacity of the founders of 95. One trouble with Jewish businessmen is that this country. they stick together and prevent other people from having a fair chance in competition. 96. No world organization should have the right to 111. There are a lot of things in this worldthat tell Americans what they can or cannot do. will never be explained by science. 97. There is a source of knowledge that is not 112. Sexual relations between brother and sister dependent upon observation. are contrary to natural law. 98. Despite the material advantages of today, 113. There may be a few exceptions, but in general family life now is not as wholesome as it used Jews are pretty much alike. to be. 114. The world will get so bad that some of these 99. The United States doesn't have to depend on times God will destroy it. the rest of the world in order to be strong and self-sufficient. 115. Children should learn to respect and obey their teachers. 100. Foreigners usually have peculiar and annoying habits. 101. Parents know as much about how to teach 116. Other countries don't appreciate as much as children as public school teachers. they should all the help that America has given them. 102. The best assurance of peace is for the United States to have the strongest army, navy, air 117. We would be better off if there were fewer force, and the most atom bombs. psychoanalysts probing and delving into the human mind. 103. Some day machinery will do nearly all of man's work, and we can live in leisure. 118. American free enterprise is the greatest bub wark of democracy. 104. There are too many people in this world who do nothing but think about the opposite sex. 119. If a person is honest, works hard, and trusfli in God, he will reap material as well as 105- Iodern people are superficial and tend to lack spiritual rewards. the finer qualities of manhood and womanhood. One will learn more in the school of hard knocks than he ever can from a textbook. 120 -71- Scoring Key for the Inventory of Beliefs The test is scored by counting each "disagree' or "strongly disagree" response. The total score is the total number of such responses. 10. ll. 12. - 72 .— LIST OF REFERENCES Adorno, T. W., Frenkel—Brunswick, Else, Levinson, Daniel J., and Sanford, R. Nevitt, The Authoritarian Personality, New York: Harper and Brothers, 1950. Allport, Gordon W. , "Prejudice: A Problem in Psychological Causation", in Parsons, Talcott, and Shils, Edward A. (Eds.) Toward a General Theory of Action , Cambridge: Harvard University Press, 1952. Bloom, Benjamin S. , "The Stereotype Personality Model and Vocational Choice ", unpublished paper read at meeting of Michigan College Counselors Association, East Lansing, 1955. Personal communication. 9 Christie, R. , "The effect of frustration upon problem solving", American Psychologist , 1950, 5, 296-297. Christie, R. , and Garcia, J. , "Subcultural variation in Authoritarian Personality ", Journal of Abnormal and Social Psychology, 1951, 46, 457-469. Christie, R. , and Jahoda, Marie, (Eds) Studies in the Scope and Method of the Authoritarian Personality, Glencoe: Free Press, 1954. Cooperative Study of Evaluation in General Education, Instructors Manual for the Inventory of Beliefs, Washington: American Council on Education, 1950. Dahnke, Harold L. , Personal communication concerning work of the Cooperative Study of Evaluation subcommittee on attitude measure- ment. Davis, Allison, Social-Class Influences upon Learning, Cambridge: Harvard University Press, 1951. Dixon, Wilfrid J. , and Massey, Frank J. , Introduction to Statistical Analysis, New York: McGraw-Hill, 1951. Dressel, Paul L. , and Mayhew, Lewis B. , General Education: Explorations in Evaluation , Washington: American Council on Education, 1954. 13. 14. 15. 16. 17. 18. 19. 20. 2.1. 22. 23. 24. 25. 26. -73- Goodrich, Thomas A. , A Study of the Michigan State College Summer Counseling Clinics , unpublished doctoral thesis, 1952. Guilford, J. P. , Fundamental Statistics in Psychology and Education , New York: McGraw-Hill, 1950. Horowitz, Victor, A Study of Selected Interest Factors as Related to Outcomes of the Program of General Education at Michigan State College , 1954, unpublished doctoral dissertation. Kendall, M. G. , and Smith, B. B. , Tables of Random Sampling Numbers , Tracts for Computers XXIV, London: Cambridge University Press, 1939 . Matteson, Ross W. , "Counseling clinics for high school grads", Occupations , 29: 502-505, April, 1951. Peters, Charles C. , and Van Voorhis, Walter R. , Statistical Procedures and Their Mathematical Bases , New York: McGraw- Hill, 1940. Rokeach, M. ,"Generalized mental rigidity as a factor in ethno- centrism ", J. of Abnormal & Soc. Psych. , 1948, 43: 259-278. , "'Narrow-mindedness' and ethnocentrism", American Psychologist , 1950, 5: 308 , "Prejudice, concreteness of thinking, and reification of thinking", J. of Abnormal 8. Soc. Psych. . 1951, 46: 83-91. Spacie, Edwin G. , "Dogmatism among non-stereotype subjects ", unpublished manuscript. Srole, Leo, " Social dysfunction, personality and social distance attitudes ", 1951 Stern, George G. ,Studies in Personality Typolo ies , Chicago: (mimeographed) 1952 Stern, George G. , Stein, Morris J. , and Bloom, Benjamin S. , Methods of Personality Assessment , Chicago : Free Press, 1956. Assessment of Personality 11 , Chicago : (mim'eographed) , 1952. 27. 28. 29. -74- Thurstone, L. L. , Multiple-Factor Analysis , Chicago: The University of Chicago Press, 1947. Thurstone, L. L. ,Chesire, Leone, and Saffir, Milton, Computing Diagrams for the Tetrachoric Correlation Coefficient ,Chicago : The University of Chicago Bookstore, 1933. Warrington, Willard G. , A Longitudinal Study of Michigan State University Students , (Study still in progress, 1956) t.» .3 4"“ F“ '9. 1 \EQ’;