",~ "2* .... AN APPLICATION OF GUTTMAN FACET ANALYSIS, T0 ATTITUDE SCALE CONSTRUCTION: _' A METHODOLOGICALSTUDY I * Thesis for the Degree ofsPh‘ Du _ _ MICHIGAN STATEUNIVERSITY ‘ I JOHN PAUL MAiERLE .. 1969 .mhgxs This is to certify that the thesis entitled AN APPLICATION OF GUTTD‘IAN FACET ANALYSIS TO ATTITUDE SCALE CONSTRUCTION: A METHODOLOGICAL STUDY presented by John Paul Maierle has been accepted towards fulfillment of the requirements for Ph. D. degree in Counseling, Personnel Services, and Educational Psychology Major ofessor Date August 15, 1969 0-169 ~ ——-——.-—om 'J‘.‘ ‘ 1 a Midfigan S 2:: :3 University ABSTRACT AN APPLICATION OF GUTTMAN FACET ANALYSIS TO ATTITUDE SCALE CONSTRUCTION: A METHODOLOGICAL STUDY By John Paul Maierle Statement of Problem Guttman, analyzing the structure of attitude items, identified four levels, or types of attitudes, generated from permutations of three facets within each item. He hypothesized that, according to his principle of contiguity, the matrix of level-by-level correlations would approximate a simplex. Jordan extended Guttman's analysis to a five-facet six—level system. Data from a scale on attitudes toward the mentally retarded did generate simplex approximations. Jordan did not examine fully (a) his choice of specific facet permutations for each level; (b) the effect of order of administration on relative size of correlations among levels; and (c) the effect of invariant directionality in answer foils. Several related problems therefore remained: (a) identification of all possible facet permutations; (b) John Paul Maierle examination of effects from order of administration and from item directionality; (c) construction of an experi- mental instrument dealing with the emotionally disturbed rather than the mentally retarded; (d) application of appropriate statistical tests. Research Design Of 32 permutations of five dichotomized facets, only 12 appeared semantically possible-—Jordan's six and an additional six. These 12 level members were hypothe- sized to be ordered within seven semantic paths; each path comprised six or four of the 12 level members. Directions for items of each level member were simplified and phrased in parallel form; all items were phrased so that all sets of answer foils were identical. A set of experimental instruments was deveIOped, two for each of the semantic paths (one instrument with the level members in a random order and one instrument with the level members in the hypothesized order). Items were written so that while content of answer foils re- mained identical, directionality of foils and grammatical emphasis within items varied randomly. The 1A varying instruments were administered in random order, one to a subject, to 825 students in under- graduate psychology and education. Data from the lu sub-groups were analyzed by Kaiser's g2 procedure, which evaluates individual correlation matrices and orders the John Paul Maierle variables within the matrices. A Q2 value was determined for the hypothesized and best orderings of level members within each semantic path; for the random—order admini- strations of semantic paths, the Q2 value was also deter- mined for the particular random order of administration. Results For six of the seven paths analyzed, the Q2 value for the randomly administered, randomly ordered matrix was less than the Q2 value for the randomly administered, hypothetically ordered matrix; in no case did the hypothe— sized ordering of level members generate the best simplex approximation. On the other hand, the lack of an ordering principle obviously better than the hypothesized one and the generally close correspondence between hypothesized and best orders suggested that the hypothesized ordering principle, the level members identified, and the orders hypothesized among those level members, are useful exten- sions of the Guttman--Jordan formulations. Among implications discussed were (a) the relation- ship of situation and content specificity to level—member specificity; and (b) possible relationships among semantic paths. Suggestions were made for related analyses of existing data and for refinements of the instrument. AN APPLICATION OF GUTTMAN FACET ANALYSIS TO ATTITUDE SCALE CONSTRUCTION: A METHODOLOGICAL STUDY By John Paul Maierle A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Counseling, Personnel Services and Educational Psychology College of Education 1969 (Eda/Igfly’ 4/1777” T1Copyright by JOHN PAUL MAIERLE 1970 PREFACE This study is one in a series, Jointly designed by several investigators, as an example of the "project" approach to graduate research. A common use of instru- mentation, theoretical material, as well as technical and analysis procedures were both necessary and desirable. The authors, therefore, collaborated in many aspects although the data were different in each study (Erb, 1969; Gottlieb, 1969; Hamersma, 1969; Harrelson, 1969; Morin, 1969) as well as certain design, procedural, and analyses methods. The interpretations of the data in each study are those of the author. ii ACKNOWLEDGMENTS I am deeply grateful to many people for their help with this thesis. I owe the greatest gratitude to Dr. John E. Jordan, my major advisor: his constant help and encouragement both in the completion of this thesis and in all my doctoral studies have been far more extensive and personal than his duties as major professor. I am grateful to Dr. George Ferree, Dr. Carl Frost, and Dr. Gregory Miller for their assistance as members of my doctoral committee: their pointed questions and sug- gestions have been most helpful. Dr. Hiram Fitzgerald was chiefly responsible for the ease with which I obtained my experimental subjects. Two fellow students, Dave Erb and Dick Hamersma, provided help, criticism and encouragement from original proposal to final copy: their assistance is almost as important to me as their friendship. Several technical difficulties could not have been overcome without the help of many persons. I particularly want to thank John Draper, for his contributions to analysis procedures; Marcus Vale and Jim Hanratty, for their computer iii programming and operation; Ann Brown, for her meticulous attention to typing and printing; and Pat Sullivan, for long hours of proofreading. This study was supported in part by training grant number 20-T-67, 68 from the Rehabilitation Services Admin- istration, Department of Health, Education and Welfare, Washington, D. C. iv TABLE OF CONTENTS |--’\OCD\] O\ Page PREFACE . . . . . . . . . . . . . . ii ACKNOWLEDGMENTS . . . . . . . . . . . iii LIST OF FIGURES . . . . . . . . . . . viii LIST OF APPENDICES . . . . . . . . . . xii Chapter I. INTRODUCTION Need . . . ‘. . 1 Purpose . 2 Hypotheses 3 Overview of Thesis 3 II. REVIEW OF RESEARCH AND THEORY Substantive Research Parental Attitudes . . Professional Attitudes. Socio-cultural Factors. . Summary. . . . . . . . . . 1 Theory and Methodology. . . . . . 11 Attitudes and Personality. . . . l2 Attitudes and Behavior. . . . . 13 Special Problems in Item Formulation . . . . . . . 14 Related Projects. . . . . . 15 Cu ttman-type Formulations. . . . . 16 Guttman Four-level Theory. . . . 16 Jordan Six-level Adaptation . . . 21 Summary. . . . . . . . 28 Chapter Page III. NEW THEORY FORMULATION . . . . . . . 29 Introduction. . . . . . . . . . 29 Conjoint Struction Facet Analysis. . 30 Element Designations . . . .‘ . . 3O Schematic Comparisons. . . . . . 31 Item Formulation . . . . . . . . 35 Comparison and Interaction . . . . 35 Directions for Various Levels . . . 36 The Tense of Level VI. . . . . . 37 Summary . . . . . . . . . . 38 Translation of Facet-element Combinations. . . . . . . . . 38 Nature of Response. . . . . . . 38 Attitude and Behavior. . . . . . 39 Implications of Elements. . . . . 39 Item and Direction Formulations . . Al Semantic Possibility Analysis . . . . A2 Semantic Map of Conjoint-struction Verbalizations . . . . . . . . A7 Arrangement . . . . . . . . . A9 Semantic Paths . . . . . . . . 6A Summary . . . . . . . . . . 68 IV. RESEARCH DESIGN. . . . . . . . . . 70 Problems . . . . . . . . . . . 7O Simplex Approximation Test . . . . 71 Order of Administration . . . . . 72 Item Phrasing . . . . . . . . 73 Approaches . . . . . . . . . . - 73 Simplex Approximation Test . . . . 73 Order of Administration . . . . . 75 Item Phrasing . . . . . . . . 76 Summary . . . . . . . . . . 78 Attitude Object. . . . . . . . . 79 SAF content . . . . . . . . . . 80 Test Population. . . . . . . . . 82 Scoring . . . . . . . . . . . 83 vi Chapter Page V. ANALYSIS OF RESULTS. . . . . . . . . 8A Semantic Path A . 85 Semantic Path B . . 87 Semantic Path C . 89 Semantic Path D . . 91 Semantic Path E . 91 Semantic Path F . 93 Semantic Path G . . . . . . . . . 95 Summary. . . . . . . . . . . . 97 VI. SUMMARY AND CONCLUSIONS . . . . . . . 100 Summary of Research. . . . . . . . 100 Statement of Problem . . . . . . 100 Research Design . . . . . . . . 101 Results. . . . . . . . . . . 102 Discussion of Results . . . . . . . 103 Semantic Path B . . . . . 103 Ordering of 12 Level Members. . . . 104 Conjoint- Disjoint Struction Inter— action . . . . . . . . . . 106 Problems in Theory . . . . . . 107 Suggestions for Further Study . . . . 108 Present Data . . . . . . . . . 108 New Experiments . . . . . . . . 109 Conclusion. . . . . . . . . . . 110 APPENDICES 111 GLOSSARY. 218 REFERENCES 22A vii Figure LIST OF FIGURES Basic Facets Used to Analyze Semantic Structure of an Attitude Item . . Level, Profile Composition, and Labels for Four Types of Attitude Item Hypothetical Matrix of Level- -by- -Level Correlations Illustrating Simplex Characteristic . . . . Basic Facets Used to Determine Conjoint Struction of an Attitude Universe. Conjoint Level, Profile Composition, and Labels for Six Types of Attitude Struction. . . A Definitional System for the Facet Analysis of Conjoint and Disjoint Struction of an Attitude Universe. Correlation Matrix for 88 Graduate SER Students Illustrating Expected Simplex Ordering of Items Con- structed on Basis of Figures 2.05 and 2.06 . . . . . . . . Correlation Matrix for 633 Education Students Illustrating Expected Simplex Ordering of Items Con- structed on Basis of Figures 2.05 and 2.06 . . . . . . . . Correlation Matrix for 523 British Honduras Elementary School Teachers Illustrating Expected Simplex Ordering of Items Constructed on Basis of Figures 2.05 and 2.06. . viii Page l8 19 2O 22 23 2A 25 26 27 Figure 3.01. 3.02. 3.03. 3.0M. 3.05. 3.06. 3.07. 3.08. 3.09. 3.10. 3.11. 3.12. 3.13. 3.1“. Comparison of Guttman and Jordan FaCet Designations . . . . .~ Schematic of Guttman Four-level Semantic Analysis Schematic of Jordan Six-level Semantic Analysis . . . Schematic of Guttman Four—level and Jordan Six—level Semantic Analyses: Comparison. Permutations of Five Two-element Facets Permutations of Five Two-element Facets and Basis of Elimination Five-Facet Six-level System of Attitude Verbalizations: Twelve Hypothesized Level Members. . . . . . . . Five-Facet Six-level System of Attitude Verbalizations: Levels, Facet Pro- files, and Definitional Statements for Twelve Permutations . . . Five-Facet Six—level System of Attitude Verbalizations: Jordan's Set of Six Permutations Five—Facet Six-level System of Attitude Verbalizations: Correspondence with Jordan Names . . . . Five-Facet Six-level System of Attitude Verbalizations: Twelve Hypothesized Level Members: Definitional State- ments and Descriptive Names . Members of Level II. . . . Members of Level III Members of Level IV. ix Page 29 32 33 3“ AA “5 A9 51 53 54 55 62 63 6A Figure 3.15. 3.16. 3.17. 3.18. 3.19. 3.20. 3.21. 4.01. “.02. H.03. 5.01. 5.02. 5.03. 5.04. 5.05. 5.06. Semantic Path A Semantic Path B . Semantic Path C Semantic Path D Semantic Path E Semantic Path F Semantic Path G Hypothetical Matrix of Level- -by-Leve1 Correlations Indicating a Special Case of Perfect Simplex Hypothetical Matrix of Level-by-Level Correlations Indicating Various Violations of Simplex . . General Content of 12 Hypothesized Level Members. . . . Semantic Path A, Randomly Adminis- tered: Correlations and Q2 Values. Semantic Path A, Administered in Hypothetical Order: Correlations and QQ'Values. Semantic Path B, Randomly Adminis- tered: Correlations and Q2 Values. Semantic Path B, Administered in Hypothetical Order: Correlations and Q2 Values. Semantic Path C, Randomly Adminis- tered: Correlations and Q2'Va1ues. Semantic Path C, Administered in Hypothetical Order: Correlations and g2 Values. . . . . . . Page 65 65 66 66 67 67 68 71 72 81 86 86 88 88 90 90 Figure Page 5.07. Semantic Path D, Randomly Adminis- tered: Correlations and Q2 Values. 92 5.08. Semantic Path D, Administered in Hypothetical Order: Correlations and Q2'Va1ues. . . . . . . . 92 5.09. Semantic Path E, Randomly Adminis- tered: Correlations and Q2 Values. 9“ 5.10. Semantic Path E, Administered in Hypothetical Order: Correlations and Q2 Values. . . . . . . . 9“ 5.11. Semantic Path F, Randomly Adminis- tered: Correlations and Q2 Values. 96 5.12. Semantic Path F, Administered in Hypothetical Order: Correlations and Q2 Values. . . . . . . . 96 5.13. Semantic Path G, Randomly Adminis- tered: Correlations and Q2 Values. 98 .1“. Semantic Path G, Administered in Hypothetical Order: Correlations and Q2 Values. . . . . . . . 98 U1 6.01. Matrix of Averaged Correlations Among 12 Level Members, Tentatively Ordered. . . . . . . 105 xi LIST OF APPENDICES Appendix A. Attitude-Behavior Scale--M. R. B. Item Variants for 12 Level Members C. Attitude—Behavior Scales, SAF Scale: Form ED-l . . . . . . . . D. Data Coding and Re-Scoring. E. Code Book F. Identification of Variables G. Sequence of Level Members in Semantic Paths. xii Page 112 118 12A 156 167 208 215 CHAPTER I INTRODUCTION Attitude research has held a prominent place in social psychology for many years. From Allport's classic paper (Allport, 1935) to the varieties of theory and method found today, many differing definitions of atti- tude and many differing emphases in measurement technique have been proposed. Among other theorists, Guttman is noted for his contributions of scalogram analysis (Guttman, 19A“) and facet design and analysis (Guttman, 1955). Jordan (1968a), reviewing current attitude research, found few studies which employed many attitude items other than stereotypic ones, although Guttman (1950) had pro— posed that attitudes exist on various levels, from stereo- typic to concrete behavioral. Jordan constructed a new attitude scale (Jordan, 1967) according to an adaptation of Guttman's proposals and found that preliminary adminis- trations of the instrument yielded results consistent with Guttman's theory. Need In the construction of his scale, however, Jordan left unanswered some questions about the relationship of 1 his formulation to that of Guttman; some imperfections in preliminary results may have been related to those unanswered questions. Jordan constructed his scale with the "mentally retarded" as attitude object; parallel research with the "emotionally disturbed" as attitude object is non-existent. Two needs existed, therefore: (a) the need for a reformulation of the Guttman and Jordan constructs and (b) the need for an instrument paralleling the original one of Jordan but dealing with the emotionally disturbed. Purpose The present study, therefore, had the following purposes: (a) to propose a reformulation of the Guttman and Jordan theories and to make a preliminary test of that reformulation; (b) to construct, according to the formu- lations of Guttman and Jordan and the new formulations here proposed, an attitude scale with the emotionally disturbed as attitude object. Some additional purposes have also been served: (a) to suggest a comprehensive system for writing attitude scales with any specified person as attitude object; (b) to suggest relationships among varieties of attitudinal behavior which have not yet been investigated; and (c) to suggest a variety of implications for such areas as attitude definition and change, personality dynamics, and the relationships among such variables as beliefs, values, laws, feelings, and overt gestures. Discussion of these additional purposes has been left to the con- cluding chapter. Hypotheses As noted in Chapters III and IV, the present study comprised a statement of a revised attitude-item con- struction theory and an empirical test of that theory. In particular, matrices of correlations among responses to various types of items were expected to approximate a "simplex" form (cf. Chapter IV). Of two tests for correlation—matrix goodness-of-fit, the Q2 index (Kaiser, 1962) appeared more appropriate; however, because the distribution of Q2 is presently un— defined, significance statements--and, therefore, standard statistical hypotheses-~were not applicable. On the other hand, level of significance has been indicated for all correlation ratios. Further discussion of Kaiser's Q2 may be found in Chapters IV and VI. Overview of Thesis Chapter I, "Introduction," comprises the above back- ground information and the following overview of the remaining chapters. Chapter II, "Review of Research and Theory," is in four parts: (a) substantive research; (b) theory and methodology; (c) directly related research projects; and (d) Guttman—type formulations. Chapter III, "New Theory Formulation," states a new formulation of attitude—item construction theory. Two emphases are present throughout the chapter: (a) an examination oijuttman and Jordan formulations, and (b) a proposal of certain changes and extensions in those formulations. Chapter IV, "Research Design," comprises a dis- cussion of apprOpriate data—gathering and statistical methods for evaluating the theoretical system proposed in Chapter III. Additional discussion at the conclusion of Chapter IV covers specification of the particular attitude object to be considered and an indication of general scoring principles. Chapter V, "Analysis of Results," presents the data gathered in the study and the various statistics generated from that data. Chapter VI, "Summary and Conclusions," reviews the theoretical formulations, research design, and actual data of the study, with additional discussion of implications for further study. Because a number of terms are introduced or are given a specific technical meaning in the present study, particu- larly in Chapter III, a glossary has been written to aid in interpretation. In general each word in the glossary has been defined at its first appearance in the study; glossary discussions expand on those original definitions. On the other hand, a preliminary list of definitions has not been included in the present introduction: such isolated definitions, presented out of context, did not appear useful, in View of the complexity of the proposed theoretical formulations. Various appendices have been added to include material such as sample items and directions from Jordan's original scale, various phrasings of items considered for the present study, a complete set of items used in the present research instrument, and a summary of scoring and data-processing procedures. The final section, "References," lists the sources which are of importance for an understanding of the present study. CHAPTER II REVIEW OF RESEARCH AND THEORY The literature on attitudes in general and attitudes toward emotional disturbance in particular is currently large and constantly growing. The present review, there- fore, is limited to those works most directly related to the present study. The review is in four parts: (a) substantive research; (b) theory and methodology; (c) directly related projects; and (d) Guttman-type formu- lations. Substantive Research Reviewing literature on mental illness and emotional disturbance, Sinha (1966) noted that most attitude studies dealt with the attitudes of parents of schizophrenic or neurotic children and that differences found between such parents and parents of "normal" children were more of degree than of kind. Among the most widely used instru- ments was the "Parental Attitude Research Instrument" (PARI) (Shaefer and Bell, 1958). Thurstone (1959) also developed an instrument for measuring such parental attitudes. Parental Attitudes A number of studies did not include control groups. Tietze (19A9) concluded that mothers of schizophrenic children were generally insecure, superficial, domineer- ing, and rigid. In a factor-analytic study, Shepherd and Guthrie (1959) identified five factors in the attitudes of such mothers: (a) detached authoritarianism, (b) inadequacy and inconsistency, (c) pervasive control, (d) sophisticated denial of adequate mothering, and (e) annoyance and rejection. Kasanin, Knight and Sage (1934) had identified both parents of schizophrenic children as overprotecting and rejecting; Reichard and Tillman (1950) distinguished three types of schizophreno- genic parents: (a) overtly—rejecting mother; (b) covertly— rejecting mother; and (c) domineering, sadistic father. Comparing the perceptions of parents and children on child—rearing attitudes, Farina and Holzberg (1967) found the perceptions of children and parents similar but noted that the relationship between expressed attitude and overt behavior was complex and very much a function of individual differences. Two studies utilizing similar control groups (Prout and White, 1951; Mark, 1953), although differing somewhat in specific conclusions, found few major differences among mothers of male schiZOphrenics. In a control—group study utilizing the PARI, Zuckerman, Oltean and Monashkin (1958) found no support for the widely accepted hypothesis that mothers of schizophrenics are generally controlling and rejecting. Differences between parents of schizophrenic chil- dren and parents of other children have been found. McKeown (1950) concluded that parents of schizophrenics tended to be antagonistic while parents of "normals" tended to be encouraging. On the other hand, Klebanoff (1958) found fewer pathological tendencies in mothers of schizophrenics than in mothers of brain-damaged and retarded children. Field (1940) and Bolles, Metzger and Pitts (1941) studied parents of neurotic children and found: (a) inadequacy in parental role, (b) rejecting attitude, (0) poor marital adjustment, and (d) infantile neurotic traits. Shoben (1949) found parents of neurotic and problem children dominant, possessive, or ignoring. Professional Attitudes Several studies have dealt with attitudes of hospital or other mental-health personnel. Cohen and Struenning (1962) tested large numbers of hospital personnel; they then identified five factors which vary with occupation and degree of education: (a) the tendency to consider patients different and inferior; (b) approval of strong restrictions on patients during and after socialization; (c) a sense of moral obligation to help the unfortunate; (d) professional attitudes regarding the treatability of mental illness; and (e) attitudes regarding the etiology of mental illness (Struenning and Cohen, 1963). They further developed certain occupational profiles for mental hospital personnel (Cohen and Struenning, 1963). Stotsky and Rhetts (1967) found the same charac— teristics among non—trained hospital personnel. Gilbert (1954) distinguished between custodial and humanistic ideologies among hospital personnel. In a factor—analytic study of doctors' attitudes, Taylor (1965) found no personality variables among physi- cians to correlate with five identified factors: (a) self- confidence in treating mental illness, (b) dogmatism and authoritarianism, (c) perceived status of psychology and psychiatry, (d) acceptance of the counseling role, and (e) a group of questions about psychiatric institutions. Wright and Klein (1966) did find that degree of education and training was directly related to positiveness of atti— tude toward mental illness. Socio-cultural Factors Recent studies have noted the relationship between socio-cultural factors and attitudes toward mental illness (Askenasy, 1963; Zavalloni and Askenasy, 1963) across cul— tures. In a series of studies (1965a, 1965b, and 1965c), Adis-Castro and Waisenen found that modernity and socio- economic level were directly related to attitudes about mental illness. 10 Such socio—cultural studies have used increasingly SOphisticated methodology. Greenbaum and Wang (1965) identified semantic-differential differences among the concepts of "retarded," "mentally ill," "emotionally disturbed," and "neurotic." Swarte (1968), using a semantic-differential approach to distinguish among "criminal," "mentally ill person," "physically ill person," "ex—mental patient," and "myself," noted gener- ally unfavorable attitudes and support for the hypothesis that authoritarian personality is related to such un— favorableness; lack of contact, however, was not found related to unfavorable attitudes. Swarte stated that "evaluative meaning may be taken as the attitude dimension" (Swarte, 1968, p. 15). Earlier, Gatherer and Reid (1963), concluded that "lack of knowledge encourages the persistence of stereotyped ideas amounting . . . at worst, to rejection of the mentally disordered and to the continuation of stigma" (p. 42). Both studies dealt exclusively with varieties of verbal behavior; in an experimental setting, after noting that attitudes toward the mentally ill were generally unfavorable, Farina and Ring (1965) found that the perception of working with a mentally ill person improved task performance but that inadequacies in jointly performed tasks were inaccurately ascribed to the partner identified as "mentally ill." 11 Summary Substantive research on attitudes toward mentally ill persons, much of it on parental or professional atti— tudes, has become more sophisticated. Earlier studies used no control groups and included few socio—cultural variables. Recent trends include use of control groups, cultural and cross-cultural data, and behavioral indices. Sinha's conclusion (1966, p. 36) that "methodologically superior research under a comprehensive framework is desperately needed" is perhaps somewhat less true today. On the other hand, the use of the term "attitude" remains ambiguous both in such research and in attitude theory in general. The following section deals with such general theory. Theory and Methodology Fishbein (1967) has collected most of the classic papers in attitude theory; Zinnes (1969) reviewed the most recent work in the special area of scaling. Proctor (1967) discussed the more obvious problems in attitude measurement. Among other general works are the data-theory emphasis of Torgerson (1958) and the scale-construction emphasis of Edwards (1957). Among the better-known names in attitude theory are those of Allport, Thurstone, and Likert. Allport's dis— cussion of the nature of attitudes (1935) is perhaps the classic in the field. The Likert technique (1932) is 12 widely used in scales designed to measure "attitudes," "opinions," "beliefs." Thurstone (1928) was among the first to suggest a specific scale construction technique. Both theory and measurement continue to receive extensive treatment. Katz and Stotland (1959) present a comprehensive theory of attitude structure and change; a new approach to scale construction, using Stephenson's Q technique (1953) and analysis of variance procedures, has been suggested by Kerlinger (1956). Among general works appearing recently may be mentioned Shaw and Wright (1967), Insko (1967), Sherif and Sherif (1967) and Halloran (1967). Although the term "attitude" remains ambiguous, some trend toward precision may be noted. In 1928 Symonds noted that the term could mean drive, muscular adjustment, generalized conduct, readiness, emotional response, feel- ings, or verbal responses (Symonds, 1928). By 1966, much of attitude theory (excluding the work of Guttman, to be discussed in a following section) agreed with the Ker— linger notion of a predisposition to perceive, think, feel, and behave (Kerlinger, 1966). Attitudes and Personality A number of investigators have suggested links be— tween various patterns of attitudes and particular personality types. The so—called "authoritarian person- ality" (Adorno et al., 1964) was identified in research l3 originally designed to measure anti-semitic attitudes. Rokeach (1961), noting that the authoritarianism studied by Adorno appeared to be associated with only one end of a political ideology continuum, proposed a more general "dogmatism" scale; in his more recent work (1968), Rokeach identifies a comprehensive system of beliefs, attitudes and values within basic personality structure. Following the theoretical framework of Fishbein (1967), and using both the F scale (Adorno g£_al., 1964) and the Rokeach Dogmatism scale (Rokeach, 1961), Anderson (1968) concluded that degree of attitude change is a function of personality types. Osgood and Tannenbaum proposed (1955, 1957) that congruity among various attitudes was a basic personality rleed. Crano and Schroder (1965) concluded, however, that only integratively simple subjects are bound by the prin— ciple of internal consistency. Among studies on the related concept of "balance" is the work of Price, Harburg and Newcomb (1966), who pointed out that attitudes must be studied in the light of reciprocity from the personal attitude object, rather than being viewed as the simple product of the isolated subject. Attitudes and Behavior The relation between a verbal response and an ob- served behavior has long been a problem in attitude research. Vinacke (1952) noted three aspects of the l4 problem: (a) the indirect character of attitude instru— ments, such that attitudes are inferred rather than measured directly; (b) the lack of an established unit for measuring attitudes; and (c) persistently low validity and reliability in attitude instruments. Boyd assumed that unless attitudes become "estab- lished to the point of awareness, they cannot be measured by the endorsement of attitude statements" (Boyd, 1943, p. 10). Insko (1965) discovered that verbal reinforcement has an effect on responses to attitude questionnaires; Loew (1967) found that hostile responses could mediate overt aggressive behavior. Some tentative connections between verbal responses and overt behavior have there- fore been identified. Tittle and Hill (1967), commenting on the generally poor correlations between verbal attitude responses and overt behavior, noted that such correlations were a function of (a) the measurement techniques used, (b) the degree to which the behaviors measured were common or unique, and (c) the repetitive or single character of the behaviors measured. Special Problems in Item Formulation DeBaty (1967) concluded that degree of perceived ambiguity was a function of intellectual level of the judges but was independent of the particular attitudinal position held by the judges. A particular item, there- fore, might be perceived as relatively simple by one l5 subject—~hence, easier to give assent to or to disagree with--and considerably more complex by another subject. Anderson (1965), reversing the phrasing of question- naire items, discovered that items phrased negatively tended to produce significantly lower subscale average ratings, although the relationships among various sub— scales remained constant. Total score, therefore, could be artificially lowered or raised by manipulating the direction in which items are phrased. Related Projects As noted in the Preface, the present study is one of several related studies. Jordan (1968a) reported on an ll-nation study of attitudes toward education and physical disability. Jordan's Attitude-Behavior Scale--MR (ABS—MR) (1967), indicated in Appendix A, is the principal instru— ment in the studies of Harrelson (1969), Gottlieb (1969), and Morin (1969). Hamersma and Jordan's scale (1969), a refinement of the Jordan and Hamersma scale (Jordan and Hamersma, 1969), dealing with attitudes toward Negroes and whites, has been developed from the work of Jordan (1968b) and Hamersma (1969). 'Phe original version of the scale also appears in the work of Erb (1969), who investigated racial prejudice and empathy. The work of Jordan and of Hamersma is of two types, attitude-scale construction and substantive research. 16 The work of Erb, Harrelson, Gottlieb, and Morin is pri— marily substantive research. The present study is one in attitude—measurement theory. Guttman-type Formulations As indicated in Chapter III, the present study in- volves a reformulation of certain attitude—measurement constructs first proposed by Guttman and later refined by Jordan. The following section of the present chapter, therefore, comprises a summary of Guttman's original formulations and the adaptation proposed by Jordan. Guttman Four-level Theory Guttman's early work on attitude theory was concerned with the measurement of attitudes during World War II (Stouffer, 1950). In contrast to Lazarsfeld (1959), whose contribution to attitude measurement emphasized the latent continuum, Guttman stressed the simple response to a given instrument item. For Guttman, whether or not some latent structure existed was not of importance; for him the crucial point was the possibility of ordering a series of specific responses to instrument items and from that ordering to make predictions. Guttman defined attitude as "a delimited totality of behavior with respect to something" (1950, p. 51). Within the limits of such a definition, both verbal responses and overt behaviors can be construed as atti- tudes-—in one sense, the attitude is what is measured. 17 But if the particular response to an attitude item constituted an expression of attitude, then individual items could be analyzed to see what makes any kind of ordering possible. One form of such ordering is empirical, the Guttman contribution of "scalogram analysis" (Guttman, 1947). Another way to order responses is to examine the various semantic factors, or "facets," involved in a particular response to a particular item. Commenting on the work of Bastide and van den Berghe (Guttman, 1959), Guttman distinguished three "facets" in- volved in a particular attitude response: the subject's behavior (belief or overt action), the referent (the subject's group or the subject himself), and the referent's intergroup behavior (comparative or interactive). He labeled the first of the two options, or "elements," of each facet as the "weaker." A particular attitude item, then, was as strong as the number of strong elements which appeared. If all attitude items could be distinguished semanti- cally in terms of three facets, then an individual item could have none, one, two, or three strong facets--a total of four combinations. Guttman further indicated a logi- cal reason for considering only four permutations of Strong-weak facets. If elements are correctly ordered within facets and facets are correctly ordered with respect tO each other, a semantic analysis of attitude items 18 according to n dichotomous facets will reveal n + 1 types of attitude items. Guttman called these types "levels." Each level had one more strong facet than the level immedi- ately preceding, and one less strong facet than the level immediately following. Figures 2.01 and 2.02 are adapted from Guttman (1959). Figure 2.01 contains the three facets originally identified by Guttman and the elements identified within those facets. Figure 2.02 contains the four permutations of strong-weak facets which Guttman identified and the descriptive names which he attached to each of those permutations. FIGURE 2.01 Basic Facets Used to Analyze Semantic Structure of an Attitude Item A = Subject's B = Referent C = Referent's Facets Behavior Intergroup Behavior a = belief b = subject's c = comparative 1 l rou 1 Elements g p a2 = overt b2 = subject c2 = interactive action himself 19 FIGURE 2.02 Level, Profile Composition, and Labels for Four Types of Attitude Item Level Profile Descriptive Label I Stereotype II Norm III HypothetiCal Interaction IV Personal Interaction An example of an attitude item identified as belong- ing to a particular level may be helpful. The statement "All members of the --—- race are mentally defective" would be a level-I response, or stereotype, whether accepted or rejected by the subject who responds. The behavior of the subject is a belief about how the attitude object compares with other persons such as the subject's 53232. The statement "I have invited members of the —--- race to a dinner or party at my house" would be a level-IV response, or personal interaction. The response indicates an interaction of the subject himself, a specific overt action. Guttman hypothesized that, according to what he called the "principle of contiguity" (cf. Foa, 1958), responses at any given level would be most closely related to the most similar levels--the levels having the largest 20 number of common facets--and less related to less-similar levels. Thus level—I responses would be more similar to level-II responses than to responses of any other level. If such similarities were expressed in correlation ratios, the matrix of level-by-level correlations would have a distinctive appearance. Figure 2.03 indicates what such a hypothetical matrix might approximate. Such a matrix Guttman labeled a "simplex." FIGURE 2.03 Hypothetical Matrix of Level-by-level Correlations Illustrating Simplex Characteristicl Level I II III IV I 1.00 II .90 1.00 III .80 .90 1.00 IV .70 .80 .90 1.00 lExact magnitude of correlations not determined; note that simplex requirement does not necessitate either identical correlations in diagonals or identical differences between diagonals: the case given is sometimes called a "perfect simplex." The fundamental requirement in any simplex is that correlations decrease as they are farther from the main diagonal. According to Guttman, if attitude items are correctly written-~i.e., to correspond to each of the hypothesized levels-—then the matrix of level-by-level correlations 21 should approximate the simplex. If, on the other hand, a simplex did not appear, the items were incorrectly or ambiguously assigned to levels. Jordan Six-level Adaptation Jordan (1968a) proposed an expansion of the facet- theory analysis. According to Jordan, five facets--and, therefore, six levels--are identifiable. Figures 2.04 and 2.05, corresponding to Figures 2.01 and 2.02, indicate the facets, elements of facets, and levels identified by Jordan. Jordan further defined the characteristics of items written for each level of response. Figure 2.06 (Jordan, 1968a), indicates "definitional statements," or specifi- cations of the response characteristics, proper to each hypothesized level. Jordan has prepared an instrument designed to con- tain the newly hypothesized six levels (Jordan, 1967). General directions and sample items from each level are given in Appendix A. Figures 2.07 and 2.08 indicate simplex approximations obtained in preliminary testing of the instrument. Similar results have been obtained in a cross-cultural study: Figure 2.09 indicates the simplex approximation obtained in Belize, Central America. 22 FIGURE 2.04 Basic Facetsl Used to Determine Conjoint Struction2 of an Attitude Universe3 (A) (B) (C) (D) (E) Referent Referent Actor Actor's Domain of Behavior Intergroup Actor's Behavior Behavior al others bl belief 01 others dl comparison el symbolic a2 self b2 overt c2 self d inter- e2 opera- action action tional lAs B qualifies A's behavior, so E qualifies C's behavior. Frequently, but not necessarily, A and C are identical. In such cases, B and E must be "consistent," i.e., some combinations seem illogical; BlEg. It should be noted that sometimes the subject filling out the questionnaire is identical with either referent or actor or both, but not necessarily so: i.e. in Level 1 and 2 referent and actor are identical, the subject is asked to report about them; in Level 3 the subject is identical with the referent, but not with the actor; in Level 4, 5, 6, subject, referent, and actor are identical. (See Figure 2.05). 2Conjoint Struction: Operationally defined as the ordered sets of these five facets from low to high across all five facets simultaneously. The more subscript elements a set contains, the greater the "strength" of the attitude. It should also be noted that not all combi- nations are logical. The selection of a "best" group of sets is still partly a matter of judgment. Two continua run through the facets: other-self and verbal-action. 3Figure and notes adapted from Jordan (1968a, p. 76). John E. Jordan Michigan State University Louis Guttman Israel Institute for Applied Social Research February 9, 1966 23 FIGURE 2.05 Conjoint1 Level, Profile Composition,2 and Labels for Six Types of Attitude Struction3 2 Type—Level Struction Profile Descriptive Conjoint Term l Societal Stereotype 2 Societal Interactive Norm 3 Personal Moral Evaluation 4 Personal Hypothetical Behavior 5 Personal Feelings 6 Actual Personal Behavior lConjoint order: Level 1 < level 6 and a < 02; d1 < d2; el < e2. < a ' 1 2’ b1 < b2’ Cl 2Based on facet order of February 9, 1966 (see Figure 2.04). 3Adapted from Jordan (1968a, p. 77). John E. Jordan Michigan State University Louis Guttman Israel Institute for Applied Social Research March 7, 1968 24 FIGURE 2.06 A Definitional System for the Facet Analysis of Conjointl and Disjoint2 Struction of an Attitude Universe_ (0 FJ A. J. (‘D Level Descriptive Term Respondent B s/Reports a b c l l 1 etrl (P) Lelieve3 the pop. a3 a whole, believes, that the pop. e ‘1 mLolioally (would), combires, the disabled . G in life situation invOlving level of—impertance under V . . specified conditions via evaluat n process on trait T j tvne and tr'i 1* level. __r_.-__ —_——____— .r“ b l C 1 G I \ ~ 1 v \ y l 2 booietal (a) Lelieves the 0p. as A whole, believes, that tne T'iter"i‘t P Live el 39 " 'licilly (would), inter acts, with the C4. 9 a r (U 12 L1 31 3 Per.ona1 (S) Leli eve: he himself, belitxv , thit thegpopulat it on .':’.1I'fi ‘3'1 1;) Exallzition as a whole, symLolically (ought), interacts, with the 1, d9 C1 LI 631 ‘ L disable} in (etc.). 13 L], 32 4 Feracnal (C) Lelieves he nimselI, Lelieven, that he hintelf, Hvoothetical e1 1, u . (_ menavipr symbolically (chli), interaci , with the disabled in (etc.). Cl- 21 I. b 8. 2 2 a 1 1 19 be 5 Personal (S) reports he himself, h}; astgal feelings, such as thrt Feelih; C: e: 1* '——"-' ‘7 5 ,1 ‘ ' v ‘j .2 ., . -. ‘7. ' ' ‘ ' n a, a? 3, b3 el he n1fi Cl}, Hyntcllaally, interic.3, With the disabled L. 4_ L. z. {i "1 ‘L '; (_ .ctual (S) reports he himself, has actual experiences, such that eonal c; C. \ 3 .4 ‘.vior He hinself, operationally (does), inter cts, with the r\ |’\J disabled in (etc.). l n o 0 IV 0 Facets "A" through "E" denote conjOint Htruction or level. 2 . . . . . Facets "F" through "J" deno e attitude ;tem content or DiSQOint Struction; tne ordering system has not been devel ped. 3Adapted from Jordan (1968a, p. 82). John E. Jordan Michigan State University Loui3 Guttma Israel Institute for Applied Social Research March 7, 1968 25 FIGURE 2.07 Correlation Matrix for 88 Graduate SER Students Illustrating Expected Simplex Orderingl of Items Constructed on Basis of Figures 2.05 and 2.062 Descriptive Term Level 1 2 3 4 5 6 Societal Stereotype l —- Societal Interactive Norm 2 .56 -- Personal Moral Evaluation 3 .17 .34 -— Personal Hypo~ thetical Behavior 4 .10 .lg .48 -- Personal Feelings 5 .04 .13 .98 .24 -— Actual Personal Behavior 6 .00 .05 .Q4 .13 .21 -- l Underlined correlations indicate instances in which simplex ordering was not maintained. 2Adapted from Jordan (1968a, p. 84). 26 FIGURE 2.08 Correlation Matrix for 633 Education Students Illustrating Expected Simplex Orderingl of Items Constructed on Basis of Figures 2.05 and 2.062 Descriptive Term Level 1 2 3 4 5 6 Societal Stereotype 1 -- Societal Interactive Norm 2 .44 -— Personal Moral Evaluation 3 .Q5 .Q1 -- Personal Hypo- thetical Behavior 4 .15 .21 .15 -- Personal Feelings 5 .17 .12 .19 .38 -- Actual Personal Behavior 6 .01 .04 .05 .19 .22 -- lUnderlined correlations indicate instances in which simplex ordering was not maintained. 2Adapted from Jordan (1968a, p. 84). 27 FIGURE 2.09 Correlation Matrix for 523 British Honduras Elementary School Teachers Illustrating Expected Simplex Orderingl of Items Constructed on Basis of Figures 2.05 and 2.062 Descriptive Term Level 1 2 3 4 5 6 Societal Stereotype l —— Societal Interactive 'Personal Moral Evaluation 3 .11 .32 -- Personal Hypo- thetical Behavior 4 .21 .28 .39 -- Personal Feelings 5 .17 . 6 .19 .31 —- Actual Personal Behavior 6 .13 .10 .15 .32 .16 -- lUnderlined correlations indicate instances in which simplex ordering was not maintained. 2Unpublished data; used with permission of J. E. Jordan, Michigan State University. 28 Summary Guttman proposed a four-level system of attitude items; within the system, levels were hypothesized to be related to each other according to the principle of contiguity, so that a matrix of level-by-level corre- lations would approximate a simplex. Jordan proposed a five-facet, six-level adaptation of the system and has preliminary data within and across cultures on a research instrument; the data obtained support Guttman's hypotheses. CHAPTER III NEW THEORY FORMULATION Introduction In Jordan's adaptation of Guttman's original facet analysis, certain facet names have been assigned new meanings, although the adaptation corresponds in other details to Guttman's original designations. What Guttman called "referent" is now called~"actor"; what Guttman called "subject" is now called "referent." Figure 3.01 indicates the original Guttman facet names and the cor— responding facets and facet names in the Jordan adaptation. FIGURE 3.01 Comparison of Guttman and Jordan Facet Designations Facets in Jordan Adaptation Designa- tion A B c D E Jordan Referent Referent Actor Actor's Domain behavior inter- of group actor's behavior behavior Guttman --- Subject's Referent Referent's --- behavior intergroup behavior fie 29 30 Conjointl Struction Facet Analysis Such facet designation, whether of three facets (im— plying four levels) or of five facets (implying six levels) is a designation of "conjoint struction"-—i.e., of that part of the semantic structure of attitude items which can be determined independently of specific response situations. Corresponding to "conjoint struction" is "disjoint struction,"v«i.e., that part of the semantic structure of attitude items directly dependent upon specification of a situation. Thus, both the statement "I invite ———- to parties at my house" and the statement "I work for someone who is-uuud'are designated as "personal interaction" (Guttman) or "actual personal behavior" (Jordan); on the other hand, the situations specified in the two statements are different; that is, the conjoint struction is identical, but the dis- joint struction is not. Element Designations One inconsistency between Jordan and Guttman designa— tions does appear, however. Guttman designates the elements of his facet B (referent) "subject's group" and "subject himself"; Jordan designates the corresponding facet (actor) elements "self" and "others." Similarly, Jordan designates the elements of his facet A (referent) as "self" and "others." Some of the redundant expressions which appear in Figure 2.06 1Not to be confused with "conjoint measurement" (Zinnes, 1968). 31 may be eliminated by reedesignating the elements in the Jordan five-facet adaptation. If facets A and C are given the element designations of "we" and "I" and facet B elements are designated "believe" and "act," while facet D elements are designated "compare" and "interact" and facet E elements are designated "symbolically" and "operationally," the lengthy defini— tions of Figure 2.06 can be reduced to such definitional statements as "we believe we compare symbolically . . . ." The statement "we believe we compare symbolically " may be considered the operational introduction to stereotypic behavior. Thus stereotypic behavior may be considered the statement (expressed by agreement with an item) that members of agparticular group ("we," not "I") believe that they themselves ("we") compare favorably with some designated outgroup on some symbolic (e.g., value- system) basis. Schematic Comparisons Figure 3.02 is a schematic presentation of the de- finitional statements corresponding to Guttman's original three-facet system. Figure 3.03 is a corresponding pre— sentation of Jordan's five-facet system. Figure 3.04 is a presentation of Figure 3.03 superimposed on Figure 3.02 to illustrate the semantic correspondence between Guttman's four-level and Jordan's six-level systems. 32 FIGURE 3.02 Schematic of Guttman Fourelevel Semantic Analysisl Facet Subject's Referent Referent's Element behavior intergroup behavior Weak Element believe we compare Lux— —T-——-——x— — ---—-x II--x--—-b ————— x \ \. x 77 \ Strong \\ \ Element act \ I interact \ \ l -< - ‘ \ \ \ X \ x- —- ————— x IV x— - _ —---——--x—-—mn H> a > Ham>oa page mpoz .AMAmcfi = .mfimmawc< muaafinammom oapcmsom= .moV mo:oEmpMum.HmcoHu .mo.m mhzwfim co AHV mp0: mom a H>Illllnxlll41lnxlnllxnll..|xllllllx xln.ITx-l.I1.lxII.uL.ulx \ \xullllx x x . I x \ \ x I \ x \\\’ \ c I x _ \ . x I \ \\ \. I \ \_V\\ II / :\ xx \ \ /I I x x xx x (A. x I \ newsman om zaamcofiumnwWox \ pompouwM H I ax \ H wcohum u \\ “A II x x u \ .-- ----«... a. , \. \ x I x \ \ I I\ >H..1 I.I....x\ I x \ x \ II \ HHHI..Iule\ leunntucx \ I HWI::III-x xull..ulxluunlu-x Hll lltxlllluuuxullalxnlllflxuuulu-ux .aHHmQHHonEam mLMQEoo o3 o>oaaon o3 ucoaoaw xmo: poH>mnmn hoa>mnon Ho>oq m.pouo< asohwpoucfi poa>mcmn mom no chan m.nouo< Louo< pcmpouom pcmsouom u m H mammamc< caucwsom Hm>oHIwa eunuch no ospmsonom mo.m mmDGHm 34 Cannon was CMEppso opozz coxopncs mam mocHH wcHUCOQmOLLoo on» .opflocfioo mesoEoumpm Hmcoapficficoo ”aflpmsmcom on» Ca muoomc mappwe omen» one mflco Eonm mpcmEon omHLQEoo mucoEODmum Hmcofipflcfiump Ho>ochmEupso m .mo.m onswfim co AHV One: new mo.m msswfim :o AHV mpoc mom.H H> I I f .I I I.xIIII.I I.I I.x x xI.I I.I I I x x I I.I I .Ix I.I I I I.Ix I IIII I Ix \ \_x x. x x . x x x x \ x / \\ x. / x ‘\ \ N / \ \ \ \ V. \\ ‘LK / k\ zfifimcowpmsoaw..\ DomLODCHI H Dom \ H pcoEmHm x \ I // \ \ wcopum N x // k \.~ 4 x .\ x x 11.: I I I I.I.Ix\.\\ [/1 \\ >H.I.II I I I I x..\ / R \ . I \ HHHI... I I I.Ix xI I I.I.I- -.x \ HHII III..I I Ix x VI.II I I Ix HI I I. I I I I.xI.I I I I I Ix x H? I I I.I I.Ix xflfisoHHoDEmm mLmQEoo o3 escaaah m? pcmEmHo xmoz L0H>m262 I I I I dsosmgoDCH Lca>mzch I I I I :mEppso w.s:mLOdc: Hcmpmcmn m.LUOchsw mow>mEOQ Lofi>mch Ho>oH m.soDom dsosmLoDCH Loflpmch cansom mcmpsow do :HsEOQ m.LoHo< pnwua Dutsmdom pcopocom peomm Empmmm HCOHfiLeroo Hwomzfimc< capcmEmw Hm>mHIwa cmesom new Ho>mHIszom :mEppsm mo oaumsozom :o.m mmDmHm 35 Item Formulation Comparison and Interaction According to the Jordan six-level system, only levels I and II involve the weaker element, comparison, of facet D, Actor's intergroup behavior. But the implication of comparison seems to appear also in his directions and answer foils for level V (see Appendix A). On the other hand, the implication of interaction, the stronger element of facet D, is found only in the content of individual items for level VI and not at all in the foils for that level. On the other hand, the comparison implied in level I items (directions and foils, but not in item content) is pre- sented in such a way as to lead to interpretative problems: which is the "best" response? to say that members of -—-- group are equal to members of other groups? or to say that the members of the specified group are superior? A possible solution to such interpretative problems would be to write items on all levels so that all foils are the same for all items; any statement of comparison or interaction would, therefore, have to be made in the content of the item and not in the foils. Items involving comparative behavior (levels I and II) would be of the‘ type "We believe that we have more of characteristic x than do members of group ---—." Items involving interactive be- havior would be of the type "We act--or think we act-- in such or such a way." The set of foils for every item 36 could then be written "agree, uncertain, disagree"; directionality—-i;e., favorableness or unfavorableness implied in the "agree" response would be entirely de- pendent on the content of the item; no item would contain the ambiguous "equal to" choice. One objection to such an approach to item formula- tion would be that of the danger of response set, or the tendency to answer all questions with identical foils in a similar fashion, independently of the content of parti- cular items (Guilford, 1954). Two techniques of item formulation would appear to resolve the problem raised by the response-bias objection: writing of items such that half imply favorableness toward the attitude object when the foil "agree" is marked and half imply unfavorableness; and random distribution of such "favorable" and "unfavorable" items within the set of items designed for a given level. Directions for Various Levels Present directions for the Jordan adaptation use the words "other people" when the weaker element of facets A or C is involved. As indicated above ("Element Desig- nations"), if items are phraséd in the first person, "we" and "I" may be preferable terms both for definitional statements of levels and for item directions. The direc— tions for level I, therefore, would read "This section contains statements about ideas which you may share with other people . . . . " And items written for level I 37 would be of the form "We think that we are less ——-— than members of group ---—." But how specific should the "we" be made? Two ap- proaches seem defensible: one which leaves the specifi- cation of "we," or the "in—group" of the person responding, entirely undefined. A question might then be added to each level of a set of attitude items to determine what the in- dividual considered as the "we" of his responses. One problem with such an approach is that, if the in-group were indicated as different for varying levels of the attitude items, interpretation from level to level would be difficult. Or, as an alternate approach, one which at least specifies the "we" as all those who are not members of the attitude-object group. Such an approach would be consistent with the assumption that those who respond to an attitude-object person can do so as members of their own in—group. The Tense of Level VI Items presently written for the Jordan adaptation in level VI are of the form "Have you ever done such or such?" An improvement might be made by phrasing all such items in the form "Do you do such or such?" The objection that persons who do not now have an opportunity for some specific interaction but who did interact in that way at some time would be penalized by such an emphasis on present behavior may be answered as follows: (a) the emphasis in 38. level V1 is not on why or why not some specific behavior takes place, but on whether or not it takes place; (b) phrasing in all other levels is in the present tense, so that behavioral items of level VI might reasonably be ex- pected to correlate more closely to items of other levels if all items emphasized present comparisons or interactions; and (c) short of reports by observers, a response to the present—tense statement "I do suCh and such" is the closest to actual behavior of all verbalizations-—i.e., "what I do now" is closer to actual personal behavior than is "what I did then." Summary Uniformity of foils for all levels, specification of "we" as the weaker element of facets A and C, and phrasing of level VI items in the present tense seem de- fensible changes in the present item formulations of Jordan. Examples of all such changes are indicated below in the final section of this chapter, "Semantic Map of Conjoint Struction Verbalizations." Translation of Facet-element "Combinations Nature of Response Attitude items such as those discussed here are-verb- alizations of behavior, or verbal expressions of a set of behaviors. The behaviors-~and, therefore, the verbalizations-- 39 are hypothesized to be of varying types or levels. Such verbal attitude items are open to the criticisms leveled at all questionnaire-type instruments: effects of response bias and of analogous pressures such as social desirability. Attitude and Behavior A traditional distinction made between attitude and behavior is that between inclination to act and the action itself. The present formulation is consistent with that distinction: attitude items are considered verbalizations of behaviors, not the behaviors themselves. The behaviors, in turn, are hypothesized to exist at varying levels, so that the degree of favorableness toward an attitude ob- ject which is evidenced on one level is correlated in varying degrees with degrees of favorableness evidenced on other levels of behavior. Attitude, then, is not seen as a single psychic position, but as a "delimited totality of behavior with respect to something" (Guttman, 1950). The present study, therefore, is an analysis of systems of verbalizations which may be made by a member of an identifiable group about the behavior of his group and of himself relative to some specified non-group persons. I I Implications of Elements The present Jordan adaptation ofIGuttman's facet- analysis system comprises five facets of two elements each. A brief consideration of each facet and the distinction 40 between the weak and strong elements of each facet may be helpful. Facet C, Actor, represents that person or persons who are stated, by response of agree or disagree, to com- pare or interact with the specified attitude object in some specified way. The actor, therefore, may be the person who states his own response, or the actor may be the group to which that actor belongs. Hence, the weak element of facet C, Actor, is "we," and the strong element is "I." The actor may be seen as in comparison with or as interacting with the attitude object. Facet D, actor's intergroup behavior, comprises the weak element "compare" and the strong element "interact." But the form of comparison or interaction may be only symbolic--e.g., hypothetical, or seen as conforming to group norms, or seen as expected of self by others--or operational--i.e., an instance of overt behavior. Facet E, therefore, domain of actor's behavior, comprises the weak element "symbolically" and the strong element "operationally." Any behavior of the actor, as well as any status of superiority or inferiority in comparison with the attitude object, may be perceived by the actor or by the actor's group. Thus "we" may have beliefs about "us" or about "me" (i.e., about how we or I compare or interact with the at- titude object)-Similarly, "I" may have beliefs about how “1 "we" or "I" compare or interact. Facet A, the referent, therefore, comprises the weak facet "we" and the strong facet "I." Referent behavior, Facet B, comprises the weak ele— ment "believe" and the strong element "act." Thus I or we (Facet A) may believe that I or we (Facet C) compare or interact (Facet D) either symbolically or operationally (Facet E) with the attitude object. But the Corresponding statement comprising the strong element of Facet B, act, produces sentences which are either redundant (I act I act operationally, for example) or contradictory (We act I act operationally, for example). A closer analysis of semantic structure reveals that, when Facet B is expressed through the strong element, act, the elements of Facets A and C must be identical--either both "we" or both "I." A more complete analysis of semantic possibilities con- stitutes the following section of this chapter. Item and Direction Formulations The level names and choices of permutations of ele- ments, as well as the choice of specific direction and item formulations to match particular levels, are not as- sumed to be the only possible ways in which such levels of attitude may be expressed. The names, permutations of elements, direction and item formulations, as well as the definitional statements proposed by Jordan and simplified 42 here, are assumed to be consistent with the semantic im- plications indicated above. The system of verbalizations proposed here, therefore, is not assumed to be all inclusive, but is rather a con- venient way to classify and relate varieties of behavior within a "limited totality of behavior." As Guttman in- dicates, the degree to which the matrix of level—by-level correlations approximates a simplex indicates the degree to which items have been correctly written to represent interrelated levels of behavior. The conclusion of the final section in this chapter comprises sample directions and items proposed as representative of those hypothesized levels. Semantic Possibility Analysis Jordan (l968a)comments that "not all combinations of the ordered sets of these five facets . . . are logical" and that "the selection of.a 'best' group of sets is still partly a matter of Judgment." The following con— siderations indicate how some combinations of facet ele- ments are illogical. The complete set of all permutations of five facets, where each facet comprises two elements, would include 32 permutations. Figure 3.05 gives the complete listing of all such permutations. Following the original notation of Guttman, weak elements have been assigned the subscript "l" and strong elements have been assigned the subscript "2," "3 so that, for example, the permutation l - l - l - l - 2 is that permutation having all weak elements except the final one. The particular order in which the permutations have been written in Figure 3.05 was chosen to correspond to the ordering of facets proposed by Jordan, although the explanations which follow are not dependent upon that ordering. If in place of the subscripts "l" and "2" are writ- ten the first letters of the corresponding element names (e.g., W=We, B=believe, I=interact), then Figure 3.05 would appear as Figure 3.06. For various logical or semantic reasons, however, only 12 of the 32 permutations are possible. ”As indicated above, when the strong element of facet B is present, items written would be redundant if expressed completely--e.g., "I act I act . . . ." But while a weak element in Facet B--viz., "believe"--can be preceded or followed by either "we" or "I," no implication of thinking about behavior is present in the strong element, "act." Basis l for the elimination of some permutations, therefore, is expressed "an 'a' in Facet B must be preceded and followed by iden- tical elements, both 'w' or both '1‘. Given the redundancy implied by the strong element "act" of Facet B, the actor's intergroup behavior (Facet D) and the referent's behavior (Facet B) must be consistent: they both refer to the same person or persons. If the NH FIGURE 3.05 Permutations of Five Two—element Facetsl Facets 1111111111111111 2222222222222222 1212121212121212 1212121212121212 1111222211112222 1111222211112222 1111111122222222 1111111122222222 1122112211221122 1122112211221122 Element Subscripts lSubscript "1" indicates weak element; "2" indicates strong element. 145 FIGURE 3.06 Permutations of Five Two-element Facetsl and Basis of Elimination Permutations2 Facets Basis of Elimination3 A B C D E l w b w c s 2 w b w i s 3 i b w c s A i b w i s 5 w b i c s 6 w b i i s 7 i b i c s 8 i b i i s 9 w a w c s 2 10 w a w i 8 ll 1 a w c s l 2 12 i a w i s 1 13 w a i c s l 2 l“ w a i i s l 15 i a i c s 2 16 i a i i s 17 w b w c o 2 3 U 18 w b w i o H 19 i b w c o 3 A 20 i b w i o H 21 w b i c o 3 A 22 w b i i o u 23 i b i c o 3 U 24 i b i i o 4 25 w a w c o 2 3 26 w a w i o 27 i a w c o l 2 3 28 i a w i o l 29 w a i c o l 2 3 30 w a i i o l 31 i a i c o 2 3 32 i a i i o lowed a ”0” Basis Facet a "b" See te 1See Figure 3.05 2Numbering arbitrary, for identification only. 3Logical semantic analysis as follows: Basis 1: an "a" in Facet B must be preceded and fol- by identical elements, both "w" or both "1"; Basis 2: in Facet D cannot be preceded by an "a" in Facet B; 3: a "c" in Facet D cannot be followed by an "o" in E; Basis A: an "o" in Facet E cannot be preceded by in Facet B xt for explanation. N6 referent and actor are acting (strong element of Facet B), he or they cannot simultaneously be seen simply in compari- son (weak element of Facet D). Basis 2 for the elimination of some permutations is expressed "a 'c' in Facet D cannot be preceded by an 'a' in Facet B." The domain of the actor's behavior (Facet E) can be symbolic whether the actor's intergroup behavior is com- parative or interactive, but the operational domain applies only to overt acts, so that, as basis 3 for the elimination of some permutations, "a 'c' in Facet D cannot be followed by an 'o' in Facet E." Finally, if the domain of the actor's behavior is operational (strong element of Facet D), then the ex- . pression of belief (weak element of Facet B) would seem inconsistent--i.e., the presumption would be that although the individual or group actually performs some intergroup action overtly (operationally), they only believe they so act and do not actually act (weak element, "belief," of Facet B). Provision is made for the situation in which an individual is not certain whether a particular item-- of whatever level--applies: such provision is made in the foil "uncertain." Note, however, that such uncertainty does not imply the concurrent action and unawareness of action which would be implied by the combination of a weak element, belief, in Facet B, and a strong element, opera- tionally, in Facet E. The basis A for the elimination of “7 some permutations, therefore, may be expressed "an 'o' in Facet E cannot be preceded by a 'b' in Facet B." The final column in Figure 3.06 indicates in summary fashion which of the bases for elimination of permutations apply to which permutations. Twelve permutations remain semantically possible. The concluding section of this chapter indicates the facet profile for each permutation, some relationships amoung the 12 permutations, and sample directions and items which can be written consistently with the requirements of each facet profile. Semantic Map of Conjoint-struc- 'tibn‘Verbalizations In the second part of the preceding chapter and in the preceding parts of this chapter, Guttman's original four-level system of attitude-item analysis, Jordan's six- level adaptation of the system, and some implications of that adaptation, have been described. A summary of those' implications may be helpful. Jordan suggested, as did Guttman, that the choice of a "best" system remains arbitrary. He further suggested that some permutations of five two-element facets may be semantically inconsistent. 0f the 32 permutations which might be formed, only 12 seem semantically consistent. A further examination of the 12 permutations reveals that varying numbers of those permutations belong to different levels--i.e., if a level is defined by the number of U8 strong or weak elements found in the attitude items of that level, then one permutation exists on level I, three permutations exist on level II, four permutations exist on level III, two permutations exist on level IV, and one permutation exists on each of levels V and VI. This con- cluding section of Chapter 3 comprises a description of the 12 facet-element permutations which appear semantic- ally possible and an examination of some possible re- lationships among those permutations. Arrangement Assumptions Following the formulations of Guttman and Jordan, levels of attitude—item behavior are defined by the number of strong and weak elements in the facets of items be- longing to those levels. "Level members," or permutations which belong to various levels, are as follows: level members of level I have no strong elements, members of level II have one strong facet, members of level III have two strong facets, members of level IV have three strong facets, members of level V have four strong facets, and members of level VI have strong elements in all five facets. Figure 3.07 is a schematic presentation of the six levels and corresponding level members which appear possible. Following the letter designations of Figure 3.06, the 12 permutations have been identified by the first letters of ’49 .mumomm xmm3 ccm wCOAum mo popes: beam on» m>m£ .mCOHpmHSEhou chpOh new cmEpuso on» mcH3oHHom .Hm>wH wco mam no mpmnEme HHoH m no :onmmpqxo oHHonezm m mchhcoo xon comm mmmxon chuH3 mamupoH mo COHuonuHucoOH mom mo.m ocm mo.m monstm mmmH m _ o H H m H H> : fimHHwHL > m _ ofiosm_3_ _m HH pH _ >H N — m H 3 n.3fi —1\m H H n 3— flm o H n H _ — m H 3 n H_ HHH H _ m o H n 3 — F m H 3 n 3 A — m o 3 n H _ HH 0 _ m o 3 n 34 H mumomm wCOme mo amnssz mHm>oq HmamnEmz Hm>mq cmNHmepoamm m>Hm39 "mQOHpmNHHmnpm> mUSpHpu< mo Emumzm Hm>m4|me pmommlm>Hm wo.m mmeHm 50 the appropriate elements for each facet. To summarize: in Facet A, "w" stands for "we" and "1" stands for "I"; in Facet B, "b" stands for "believe" and "a" stands for "act"; in Facet C, "w" stands for "we" and "1" stands for "I"; in Facet D, "c" stands for "compare" and "1" stands for "interact"; and in Facet E, "8" stands for "symbolic- ally" and "0" stands for "operationally." The definitional statement for the one permutation on level I, therefore, identified as "wbwcs," is read "we believe we compare sybolically." Figure 3.08 indicates the six levels, 12 permutations, facet profiles, and definitional statements for the hypothesized five-facet, six—level system. An examination of Figure 3.07 reveals that only certain sequences of level-to—level facet changes have the character of progressive, irreversible element changes found in the Guttman and Jordan formulations: i.e., if a "semantic path" is defined as any sequence of level mem- bers such that each level member has one more strong facet than the immediately preceding level member and one less strong facet than the immediately following level member, and such that, within the sequence of level members, no facet becomes weak once it has been changed from weak to strong, then only a limited number of semantic paths exist within the hypothesized system. The restrictions of such a definition of semantic path are based on the assumption that addition of strong facets has a progressive, cumulative l erbalizations Definitional Five-Fa- A d Permu C m ___ -4--_._. _ tation e 'welv . FY‘ f‘ ’n1 1 *ef Y‘ 1.1 -le 4 a ”V .1. awe” Hr n (perceive l A y-assiqned group uati; interactive norm . 'pe (group-assigned v s. 1.4.A- eva‘ 'CGPEOC‘ . \ f" status group compare ombare compare ft v we I ~ '5: e eve ‘J 6 ically e . l 1 J. olioallv , J. bel e D A. symbo 'e I v v S S w b w c s i b w c s w b w i s b i i b w W Facet Prof Level II III «a ,—-w ,LJ 5.4 C) b .atus) of n). Felf-concent (nersonallv- assigned personal (group ex— U) ancthetical personal C terac 1 Y‘ A u .1 .L “‘27 symbolically believ. p Jehavior l ual group at . x, . 11'“? C.) L13 (1) A.) w—l Actual personal behavior Actual personal J ‘7 interact onall 1' (I act) I operat- ODO i a i i s a i i o 1 VI Figures (.4 cf. 9 evel members his. I A P‘ (V D statem 6- sten ms of variou o 1 C‘ x) o 1 ‘ ate relationsh 0 11C of redundant but con inc f'V Q 939 es and 3.11. A 3.10, 3 .- Alternate names in parenth .00 Words in parenthes 2 3 igures 3 D A 52 effect upon the strength of an item: i.e., a particular level member can be hypothesized to be stronger than a member of a preceding level if and only if the preceding level member has no strong facets which do not also appear as strong in the following level member. The system has not yet been sufficiently defined or tested to permit hypotheses about the relative strength of two level mem- bers which have the same number but not the same combination of strong and weak facets. A further discussion of seman- tic paths constitutes the second part of this final section of Chapter III. Level-member Names‘ Guttman assigned descriptive names to the four level members, or permutations of the three two-element facets, in his original system. Jordan similarly assigned addi- tional descriptive names to the level members in his five— facet, six-level system. Such assignment of descriptive names was made so as to be consistent with the definitional statements for various level members (see Figures 2.02 and 2.05). In the present study, additional descriptive names have been assigned to the six level members implied in but not stated in the Jordan adaptation. Figure 3.09, an adaptation of Figure 3.07, indicates that Jordan's original level-member names have been preserved. Figures 3.10 and 3.11, with the accompanying commentary, indicate the 53 .mHonemm 30Hmn mmxon cH espouse moses pom mo.m oustm mom “COHpmqumEocom no cmemv Lou no.m mpsmHm .mo HmCOmAmm Hmsuo< h0H>mnom 7:3: Hmcomnom Hm3u0< mwcHHoom Fm H H m H— H AoH>mcmm Hmsomumm HMoHpmcuodzm — o H 3 m 3— 17m HH 9H _ Flo H 3 m 3 _ _ m H H a 3 _ AmmsHm> um>Hmohmav COHumsHm>m Hmhoz Fla 0H BH— rm.HzIMH _ m>HpompoucH HmumHoom Ehoz Fin 0 H n 3 _ Hm H 3 n 3 _ os.zH_ Anopmpm .hu .mm< .Luv mazuomampm HmumHoom F m o 3 n 3— HmQOHpmpSEmmm me mo pom m.cm@p0h "mQOHpmNHHmnpm> mwsppr< ho Smpmzm Hm>mqlem pmomm|m>Hm mo.m mmeHm 54 .moHocmpmHmcoo wcHomoHUcH mzoppm anz .mnmmSmH mmmvnuo30H new HmuHQmo CH moss: capaow ”moss: cannon HocHano cpH3 zncopmimcoo Lou :wmoco whopumh HmpHowo CH mmEmzH LoH>m£mm Hmcomnmm Hmspo< FoH Hm H— mwcHHmmm Hmcomomm Hmspo< AmmsHm> cm>Hmopmmv COHumsHm>m Hmpoz _ m Hz.nH_ _ an as H— h0H>mnmm fOH>HpompmucH HapmHoom — m o H a 3 _ _m H 3 o 3 _ _ “mauMum .ho .mm: .30 mazuomuopm Hmpchn _ m o 3 m 3H HmmEmz chLOh an3 mocmvcoammhaoo "mQOHpmNHHmnhm> cooppr< mo Empmmm Hm>mq|me OH.m mmeHm m o 3 n H _ pmomm|o>Hm .mo.m mcstm .Ho H noH>m£om, Hmcomuom Hmsuo< —0Hfi mwcHHmom Hmcomnom Hmspo< H. m H H m H— . noH>mnom 90H>mnom asohu Hmsuo< Hmcomuom HmoHponuoamx H o H z m :A_ , A m H H n H_ AumcHHoou asopm AmGOHpMpooaxo QsOhov Amoumum Hmcomuom .mm< AmosHm> oo>Hoonmmv HmsuoHuomumch HmpmHoom cocmemHpQHpome one mucmsmpmpm HmQOHpHcHHmQ Hummmnamz Ho>mq UmNHmmnuommm m>Hm3B "mQOHumNHHmnAm> mUSuHup< mo Empmzm Hm>qume umommum>Hm HH.M mmDmHm 56 consistency of the new names with the original six assigned by Jordan and the inter-relationships of all level-member names. The following paragraphs indicate general instrument directions and examples of specific items corresponding to each of the 12 hypothesized level members. Societal Stereotype (group-assigned group status): for the definitional statement "we believe we compare symbolically" general directions would read This section contains statements about ideas which you may share with other people about persons who are ----- If a statement describes the ideas of most people like yourself, mark the answer "agree" 0 O 0 etc. A sample item might be "We think that we are less intelligent than most —--- persons." Personally-assigned group status: for the definitional statement "I believe we compare symbolically" general dir- ections would read This section contains statements about ideas which you may have about yourself and others who are not ----- If a statement describes how you think, mark the answer "agree" . . . etc. A sample item might be "I think that we are less intelligent than most ---- persons." Societal interactive norm (group-assumed group be- havior): for the definitional statement "We believe we interact symbolically" general directions would read This section contains statements about ways in which you and others may think you behave toward persons 57 who are ----;- If a statement describes how you think you behave toward ---- persons, mark the answer "agree". . . . etc. A sample item might be "We think that we have some friends who are ----." Group-assigned personal status: for the definitional statement "We believe I compare symbolically" general dir- ections would read This section contains statements about ideas which other persons who are not ---— may have about you. If a statement describes what people like yourself think about you, mark the answer "agree" . . . etc. A sample item might be "Other people think that I am less intelligent than most ---- persons." Personal Moral Evaluation: for the definitional statement "I believe we interact symbolically" general directions would read This section contains statements about ideas which you may have about how you and others should act toward ---- persons. If a statement describes how you think you and others should act, mark the answer "agree". . . . etc. A sample item might be "We should be friends with persons who are ----." Self—concept (personally—assigned personal status): for the definitional statement "I believe I compare symbol- ically" general directions would read This section contains statements about ideas you may have about yourself. If a statement describes your opinion of yourself, mark the answer "agree" etc. A sample item might be "I think that I am less intelligent than most ---- persons." 58 Proclaimed laws (stated group expectations): for the definitional statement "We believe I interact symbo- lically" general directions would read This section contains statements about ideas which people like you may have about how you act toward ---— persons. If a statement describes how others expect you to act toward ---- persons, mark the answer "agree" . . . . etc. A sample item might be "Other people expect me to have friends who are -——-." Group identity (actual group feelings): for the de- finitional statement "We act we interact symbolically" general directions would read This section contains statements about feelings that you and people like yourself may have about —--- persons. If a statement describes your feelings about ——-- persons, mark the answer "agree". etc. A sample item might be "We are not frightened by ---— persons." Personal Hypothetical Behavior: for the definitional statement "I believe I interact symbolically" general dir- ections would read This section contains statements about how you think you would act toward ---— persons. If a statement describes how you think you would act, mark the an— swer "agree" . . . . etc. A sample item might be "I would have friends who are -——-." Actual group behavior: for the definitional statement "We act we interact operationally" general directions would read 59 This section contains statements about how persons like yourself actually behave toward persons who are ----. If a statement describes how you and persons like yourself actually behave, mark the answer "agree" . . . . etc. A sample item might be "We have friends who are ----." Personal Feelings: for the definitional statement "I act I interact symbolically" general directions would read This section contains statements about feelings which you yourself may have about ---- persons. If a statement describes a feeling which you have, mark the answer "agree" . . . . etc. A sample item might be "I am not frightened by ---- persons." Actual Personal Behavior: for the definitional statement "I act I interact operationally" general direc- tions would read This section contains statements about your own be- havior toward persons who are ----. If a statement describes an action of yours, mark the answer "agree" etc. A sample item might be "I have friends who are ----." Some remarks about the above interpretations of dir— ections and items for various level members may be helpful. Some of the level members for which items might be written may also be open to more valid methods of measurement. For example, which is more accurate—-a compilation of in- dividuals' scores on the items of "Actual Personal Behavior" or the score of one individual on items of "Actual Group Behavior"? The answer depends upon the particular research emphasis: which is more important in a particular study-- 60 a summary of what a group of individuals thinks about their own individual actions (many scores on "Actual Personal Behavior"), or a statement of what one group member thinks about the actions of his entire group (one score on "Actual Group Behavior")? An examination of the above item samples reveals that item content is in part related to item definitional statement. Note, for example, that the content of "Group Identity" and "Personal Feelings" is similar, but that statements about feelings would be inappropriate in such level members as "Proclaimed Laws" or "Societal Interactive Norm." A closer examination of the item examples above, with reference to specific levels of the members involved, reveals that, as items change in definitional statement so as to become stronger--i.e., so as to include more strong and fewer weak elements--the content of the items tends to change from ideas or intellectual behavior to feelings or emotive behavior to instances of concrete overt actions. Such a trend may in part clarify what Guttman refers to as stronger and weaker items and levels. Finally, the directions suggested above have been written so as to be independent of each other. A general set of introductory directions which would be consistent with any set of level members as indicated above would read: You yourself and other persons like you often behave in the same way toward persons who are ---—. You also have some general ideas about yourself, about others like you, and about persons who are ----. Sometimes you behave the same way toward everyone, 61 and sometimes you may behave differently to- ward persons who are -—--. This questionnaire has statements about ideas and about behavior. Each section of this questionnaire is different from every other section, although some of the statements in each section are similar. Your answers in one section, therefore, may be the same as answers in another section, or your answers may differ from section to section. Try to mark each set of statements as accurately as possible. If you agree with a particular statement, mark the answer "agree". . . etc. Comparisons within Levels Because more than one level member appears within some levels of the presently hypothesized system, an ex- amination of the various members found within levels II, III, and IV may be helpful. Three members exist on level II, four on level III, and two on level IV. The difference between "Personally-Assigned Group Status" and "Group-Assigned Personal Status" is implied in the two descriptive names: the former is verbalized in statements about the relative status of a group by a group member; the latter is verbalized in statements about the group member's perception of status assigned to him by his group. Finally, both "Group-Assigned Personal Status" and "Personally-Assigned Group Status" differ from "Societal Interactive Norm" in the same way: the two for- mer level members are verbalized in statements about sym- bolic comparisons; the latter is verbalized in statements about symbolic interactions. See Figure 3.12. 62 FIGURE 3.12 Members of Level Ill Profile Definitional Statement Descriptive Name 1 b w c s I believe we compare Personally-Assigned symbolically Group Status w b w i s We believe we inter- Societal Inter- act symbolically active Norm w b i c s We believe I compare Group—Assigned symbolically Personal Status 1 cf. Figures 3.08 and 3.11. Within level III, "Moral Evaluation" differs from "Proclaimed Laws" in that items of the former express the individual's perceptions of behavior to be expected from his group and items of the latter express group expectations as perceived by the individual. "Self-Concept" differs from "Group Identity" (Actual group feelings) in that items of the former express personal beliefs about personal act- ions and items of the latter express group feelings; the behavior expressed in each is, however, symbolic. See Figure 3.13. The violations of the simplex ordering which appear in Jordan's testing of his system (see Figures 2.07 and 2.08) may be due in part to the fact that four permu- tations, or level members, appear possible on level III: most of the violations of simplex ordering found by Jordan involve level III. 63 FIGURE 3.13 Members of Level III1 Profile Definitional Statement Descriptive Name 1 b w i s I believe we inter- Moral Evaluation (Per- act symbolically ceived Values) i b i c s I believe I compare Self-Concept (Person- symbolically ally-Assigned Per— sonal Status) w b i i s We believe I inter- Proclaimed Laws (Group act symbolically Expectations) w a w i s We interact Group Identity (Actual symbolically Group Feelings) 1 cf. Figures 3.08 and 3.11. Within level IV, "Hypothetical Personal Behavior" dif— fers from "Actual Group Behavior" in that items of the former express personal beliefs about symbolic interaction while items of the latter express actual group interaction. As indicated in the following, concluding part of this section of Chapter III, the present condition of theory formulation does not permit hypotheses about which of the two level members, "Hypothetical Personal Behavior" or "Actual Group Behavior," is closer to "Actual Personal Feelings" or "Actual Personal Behavior." See Figure 3.1“. 6H FIGURE 3.1“ Members of Level IV1 Profile Definitional Statement Descriptive Name 1 b i i s I believe I inter- Hypothetical Person- act symbolically al Behavior w a w i 0 We interact opera- Actual Group tionally Behavior 1 cf. Figures 3.08 and 3.11. Semantic Paths A "semantic path" has been defined above as any se- quence of level members such that each level member has one more strong facet than the immediately preceding level member and one less strong facet than the immediately fol- lowing level member, and such that, within the sequence of level members, no facet becomes weak once it has been changed from weak to strong. The simplex ordering hypo- thesized by Guttman and tested by Jordan applies only to the set of level members found within one semantic path. By extension of Guttman's reasoning, if items are written for each of the level members of a given semantic path, then any failure to approximate a simplex (matrix of level- by-level correlations) is due to inappropriate assignment of items to levels. Discussion of the relationship of level members from different semantic paths is presented in the Concluding Chapter, "Summary and Conclusions." 65 An examination of Figure 3.07 reveals six six-level semantic paths and one four-level semantic path (comprising levels I to IV, and excluding levels V and VI). Figures 3.15 through 3.21 indicate the combinations of permutations constituting those seven semantic paths.. FIGURE 3.15 Semantic Path A Profile Descriptive Name Level w b w c s Societal Stereotype I i b w c s Personally-Assigned Group Status II 1 b w i s Moral Evaluation III 1 b i i s Hypothetical Personal Behavior IV 1 a i i 5 Actual Personal Feelings V i a i i 0 Actual Personal Behavior VI FIGURE 3.16 Semantic Path B Profile Descriptive Name Level w b w c s Societal Stereotype I i b w c s Personally-Assigned Group Status II 1 b i c s Self-Concept III 1 b i i s Hypothetical Personal Behavior IV 1 a i i s Actual Personal Feelings V i a i i 0 Actual Personal Behavior VI 66 FIGURE 3.17 Semantic Path Cl Level Profile Descriptive Name w b w c s Societal Stereotype I w b w i s Societal Interactive Norm II 1 b w i s Moral Evaluation III 1 b i i s Hypothetical Personal Behavior IV 1 a i i s Actual Personal Feelings V i a i i 0 Actual Personal Behavior VI 1 Semantic path C comprises the set of "best" per- mutations of elements proposed by Jordan. FIGURE 3.18 Semantic Path D Profile Descriptive Name Level w b w c s Societal Stereotype I w b w i s Societal Interactive Norm II w b i i s Proclaimed laws III 1 b i i s Hypothetical Personal Behavior IV 1 a i i 5 Actual Personal Feeling V i a i i 0 Actual Personal Behavior VI 67 FIGURE 3.19 Semantic Path E Profile Descriptive Name Level w b w c s Societal Stereotype I w b i c 8 Group Assigned Personal Status II 1 b i c s Self—Concept III 1 b i i s Hypothetical Personal Behavior IV 1 a i i s Actual Personal Feelings V i a i i 0 Actual Personal Behavior VI FIGURE 3.20 Semantic Path F Profile Descriptive Name Level w b w c s Societal Stereotype I w b i c s Group—Assigned Personal Status II w b i i s Proclaimed Laws III 1 b i i s Hypothetical Personal Behavior IV 1 a i i 8 Actual Personal Feelings V i a i i 0 Actual Personal Behavior VI 68 FIGURE 3.21 Semantic Path G1 Profile Descriptive Name Level w b w c s Societal Stereotype I w b w i s Societal Interactive Norm II w a w i s Group Identity (Group Feelings) III w a w i 0 Actual Group Behavior IV 1 Note that (of. Figure 3.11) present theory formula- tion does not specify relationship of "Actual Group Be- havior" to level members outside semantic path G: in particular, theory does not provide for inclusion of levels V and VI in this particular semantic path. Summary Guttman identified four levels of attitude items and suggested a specific choice of level members within those levels comprising more than one member. Jordan, expanding the Guttman system to six levels, suggested that not all permutations of strong-weak facets are logical and that, among the logical permutations, the choice of a "best set" of permutations remains arbitrary. The discussion above comprises an analysis of which permutations are possible and which sets of those permu- tations--i.e., which semantic paths--are subject to the criterion of simplex approximation. The design of the present study, therefore, as stated in the following Chapter, comprises a series of tests of simplex approximation of the 69 semantic paths identified in the present chapter. The implications are as follows: if data from the adminis- tration of items as discussed above result in a series of simplex approximations, then the implications of Guttman's original system and of Jordan's adaptation have been correctly identified; if, on the other hand, analysis of the various matrices from the identified semantic paths does not reveal a series of simplex approximations, then at least some of the implications of Guttman's system have been incorrectly identified. A more complete discussion of research-design implications is found in the following Chapter. CHAPTER IV RESEARCH DESIGN Problems As indicated in the conclusion of Chapter III, the design of the present study comprised a series of tests of simplex approximation; for each of the seven hypothesized semantic paths, the matrix of level—by—level correlations was expected to appear similar to the ideal simplex proposed by Guttman. Two problems were implied in such a design: (a) how test the "goodness" of a particular simplex approxi— mation? and (b) since the analysis was of attitude verbal- izations, how insure that simplex approximations were more than a reflection of verbal behavior which might or might not have had any relation to non—verbal behavior? In particular, did the order in which the various level mem— bers of a semantic path were administered have any effect upon the correlations among those level members? A third problem, noted also in Chapter III, was that of uniformity of response foils. Uniformity of all foils ("l--agree; 2-—uncertain; 3-—disagree"), although defensible on the grounds noted in Chapter III, was Open to the criticism of response bias. 7O 71 Simplex Approximation Test Figure A.Ol is a hypothetical matrix of level—by- level correlations for a five-facet six-level system; the Figure presents not an approximation of a simplex but rather one special case of a perfect simplex. Figure 4.02 is also a hypothetical matrix, an approximation of a simplex. Note that the violations of simplex requirements in Figure 4.02 are in both rows and éolumns; the crucial question is "how much of a violation of simplex?" FIGURE “.01 Hypothetical Matrix of Leveléby-level Correlations Indicating a Special Case1 of Perfect Simplex Level I II IIILevel IV v VI I 1.00 II .60 1.00 III , .50 .60 1.00 IV .uo .50 .60 1.00 v .30 .uo .50 .60 1.00 VI .20 .30 .AO .50 .60 1.00 *7— v v ~— ‘fi— ' V 1The simplex requirement specifies only that, within such a matrix, the magnitude of individual correlations de- ’creases down each column, and, from right to left from the main diagonal, across each row. The special case here in- volves identical correlations along each diagonal and equal sized differences among adjacent correlations. 72 FIGURE 4.02 Hypothetical Matrix of Level—by—level Correlations Indicating Various Violations of Simplex Level Level I II III - IV V VI I 1.00 II .58 1.00 III .55 .59 1.00 IV .53 .50 .55 1.00 V .A8 .A5 .5A .61 1.00 VI .30 .A0 .66 , .50 .62 1.00 1 Correlations involved in a simplex violation are en- closed in boxes. Order of Administration Preliminary data from administration of Jordan's adap- tation indicated the possibility of empirical simplex approx— imation. But all of Jordan's data were obtained from adminis— tration of various level—member scales in the same order: the Jordan instrument presented all items of level I first, then all items of level II, etc. Was the correlation between two levels a reflection of the relationship between the attitude levels being verbalized or a reflection of the fact that the two verbalizations were measured near to each other in time? In each of the simplex approximations (cf.Figures 2.07, 2.08, and 2.09), the correlation between levels I and II was greater than that between levels I and VI. But such a difference 73 might not have been due to a difference in items of the three levels. The second problem, therefore, was "does order of administration of level—member items affect the simplex approximation?" Item Phrasing As noted in Chapter III, a possible solution to some interpretative problems was to write items on all levels so that all foils were the same for all items. One ob— jection to such an approach to item formulation was that of the danger of response set, or response bias, or the tendency to answer all questions with identical foils in a similar fashion, independently of the content of a specific item (Guilford, 1954). Approaches Two characteristics seemed necessary for any research- design approach to the problems indicated above: (a) con— sistency with the formulations indicated in Chapter III and (b) assurance of appropriate data for answering the questions posed. The approaches indicated below appeared to have both characteristics. Simplex Approximation Test Kaiser (1962) suggested a procedure for scaling the variables of a Guttman simplex . . . . The procedure . . . orders the variables. A measure of the goodness of fit of the scale to the data is suggested (p.155). 7A Kaiser's approach may be seen as performing two functions: (a) a "sorting" of virtually all possible arrangements of data so as to generate the best empirically possible simplex approximation; and (b) an assignment of a descriptive statistic, "g2," to specified matrices. The index g2 is a descriptive one, with a range of 0.00 to 1.00. A computer program was developed which (a) re—ordered the level members of each semantic path, by Kaiser‘s pro— cedures, so as to generate the best empirically possible simplex approximation; and (b) calculated Q2 for the hypo- thesized ordering and for the empirically best ordering of members in each semantic path. At the time of the research completion, appropriate likelihood ratios for measuring goodness of fit were not available. Mukherjee (1966) suggested a method which appeared appropriate for matrices of equally spaced corre« lations, but neither the theory set forth in Chapter III nor the actual data suggested that the matrices in the present study had equally spaced entries. No statistical comparisons were made across matrices—- i.e., from simplex approximation to simplex approximation. Guttman (1950) indicated that such comparisons could gen- erally be made within the framework of his "radex" theory, but that attitude scales should not be so analyzed. Finally, because directionality and grammatical em- phasis were controlled for by randomization, the analysis 75 of the effect of directionality and grammatical emphasis was not within the sc0pe of the study. A follow—up study, utilizing procedures indicated at the conclusion of Appendix F, is discussed in Chapter VI. Order of Administration If the order in which a set of level—member verbal? izations is presented has an effect upon the correlation of that level member with other level members, then sets of level—member verbalizations should generate better simplex approximations when listed by order of administration than when listed by relative position within a semantic path. One way to check such a possibility was to administer all the sub—tests representing the corresponding level members of a particular semantic path in some random order. If the matrix generated by listing correlations according to the hypothesized semantic path order remained a better simplex approximation than that generated from the random order of administration, then (a) order of administration was not of major importance in the simplex approximation, and (b) some indirect support would be given to the assumption that more than mere verbalizations were involved in the responses to a given set of items. In other words, if the best simplex approximations are independent of order of administration, than (a) at least, individuals are able to distinguish the levels of behavior verbalized in the various level members; and (b) 76 although the verbalizations would remain open to the criti— cisms leveled at all "paper-and-pencil" tests, some assurance would exist that the specific item verbalizations being used were related to each other in an ordered fashion. Appendix G includes a listing of each semantic path and the following information: (a) specific level members included in the path; (b) the particular random order in which the sets of items, corresponding to the various level members, were administered; and (c) the levelvby—level se- quence in which the level members were also administered. No hypotheses were made regarding a particular random order, but each semantic path was administered both in some random order and in the level—by—level order. Although the dis- tribution of 9? was not known-—hence, probability statements about "better" or "worse" fits could not be made-—the general hypothesis to be examined was that the level members of a particular semantic path generate the best simplex approxi- mation--hence, are best ordered——when listed level—byvlevel, regardless of the order of administration. Item Phrasing The sample items and general directions indicated in Chapter III do not specify a particular attitude object. Further, each item has been phrased in a particular way. In the following section, the choice of "emotionally disturbed" as attitude object is discussed. The present section deals with item phrasing. 77 In Chapter III the sample item for level I (definit— ional statement: we believe we compare symbolically) was given as "We think that we are less intelligent than most -—-- persons." The same item could have been phrased, without changing the directionality of the foils (agree, uncertain, disagree), "We think that most -——— persons are more intelligent than we are." If the directionality of the foils were reversed——i.e., so that “agree" indicated a stereotypic behavior un— favorable toward the attitude object--two more variations of the same item could be written: (a) "We think that we are more intelligent than most --—- persons" and (b) "We think that most -——- persons are less intelligent than we are." Two aspects of phrasing appear in these variations: directionality of the item and grammatical position of the "actor." An individual item could be phrased so that agreement is either favorable or unfavorable toward the attitude object; an item could also be phrased so that either the actor or the attitude object was grammatically emphasized. Each item could therefore be phrased four different ways without a change in specification of sit— uation-—i.e., without a change in "disjoint struction." Appendix B contains examples of the four item variants used in the research instrument of the present study. Appendix C contains the actual items as administered. Each 78 set of disjoint—struction items (one set for each of the 12 hypothesized level members) was randomly assigned to one of the four phrasing combinations. Such random assignment was done to control for the effect of various types of phrasing between levels: if all items on all levels were randomly distributed among the four variants of phrasing, then phrasing might be assumed to have no direct effect on the over—all pattern of level— by—level correlations. On the other hand, since the effect of specific disjoint—struction items upon total scale score or upon score for a given level was not specified, effects of dis- joint and conjoint struction interation within levels could not be specified without the construction of another in— strument analogous to that indicated in Appendix C. Such construction was beyond the s00pe of the present study. Further discussion of conjoint-disjoint struction inter— action is found in Chapter VI. Summary Simplex.approximations could be tested by the procedures suggested by Kaiser. To test the effect of order of adminis- tration on simplex approximation, sets of level members from each semantic path were administered in a random order, so that a comparison could be made of the simplex arranged by order of administration and the simplex of the same data arranged by order of level. Further, an independent test of 79 simplex approximation achieved from an ordering of level members from I to VI was run to provide additional support to the validity of the present formulation as an accurate extension of Guttman's system. Finally, all items in all level members were presented in one of four variant item phrasings. Selection of the specific phrasing for each item was by random assignment, to control for the effect of various types of phrasing--in particular, the possible effect of response bias in an instrument where answer foils are uniform throughout. Attitude Object The sample items and general directions indicated in Chapter III do not specify a particular type of person as attitude object. One characteristic of the present formula- tion is that the analyses involved are independent of a par- ticular attitude object, provided that object is some speci— fiable group of persons. The attitude object specified in the present study was the "emotionally disturbed" person, described in the instrument as including "those children or adults whose behaviors, feelings, or emotions cause them to have difficulties with every-day problems which they are un— able to solve." Such a specification rested in part on the assumption that the term "emotionally disturbed" possessed a large enough commonality of meaning as to be susceptible to measurement by verbal report. An additional reason for the choice of this particular attitude object was the lack of research of the type proposed by Guttman and Jordan on attitudes toward the emotionally disturbed. 8O SAF Content As noted in Chapter III, item content appeared in part related to item definitional statements. Note, for example, that the content of "Group Identity" and "Personal Feelings" is similar, but that statements about feelings would have been inappropriate in such level members as "Proclaimed Laws" or "Societal Interactive Norm." General content was found, in the analyses of Chapter III, to be of three types: (a) status—descriptive statements of the form "We are more intelligent than ----"; (b) action— descriptive statements of the form "We invite ---- to our home"; and (c) feeling-descriptive statements of the form "We feel loathing toward -___," "SAF" is a symbol for the three general content types: status, aption, and feeling. Figure H.03 indicates the definitional statement (of. Chapter III) and corresponding general content for each of the 12 hypothesized level members. In Jordan's later work (Jordan and Hamersma, 1969), content was kept constant across all levels. Such an approach, while permitting direct comparison of items_ across levels, required the use of such items as "We feel uncomfortable about eating with «v-v." In analysis of such items, a problem of interpretation arises: is the attitude object the specified person, or eating, or eating-with—the- specified—person? Since analysis of items across levels was not of primary importance in the present study, general 81 item content was allowed to vary in accordance with the definitional statements for each level member. FIGURE ”.03 General Content of 12 Hypothesized Level Members Level Member Definitional Statement Content I l We believe we compare status symbolically II 1 I believe we compare status symbolically 2 We believe we interact action symbolically 3 We believe I compare status symbolically III 1 I believe we interaCt action symbolically 2 I believe I compare status symbolically 3 We believe I interact action symbolically A We interact symbolically feeling IV 1 I believe I interact action symbolically 2 We interact operationally action V l I interact symbolically feeling VI 1 I interact operationally action 82 Test Population Seven semantic paths were identified in Chapter III (of. Figures 3.15-3.21). Since all level members of a semantic path were administered to the same subject at the same time, 14 groups of subjects were required—-seven groups for administration of the semantic—path level members in the I-VI order, and seven groups for adminis— tration of the same semantic—path level members in some random orders. An N of approximately 50 was set for each of the 1“ groups. Access to sufficiently large and suf- ficiently homogeneous groups of undergraduate students was arranged through the Department of Psychology and the College of Education at Michigan State University. The subjects were all enrolled in an introductory psychology course or an introductory education course. Administration of any one semantic path to a subject (no subject responded to more than one semantic path) took approximately U5 minutes. Sets of instruments were pre— pared for each of the semantic paths in the random arrange— ments indicated in Appendix G; additional sets of instruments were prepared with the level members assembled in order from I to VI. The first group of random sets was then randomly distributed to the first available approximately 350 sub— jects; the second group of sets, arranged in levelvby-level order, was randomly distributed to the next available ap— proximately 350 subjects. 83 Scoring Each set of answer foils was identical in phrasing-- "agree," "uncertain," "disagree." Directionality of foils was dependent upon the phrasing of each individual item. Directionality of foils for each possible variant is in— dicated in Appendix B. Appendix C contains the specific item variants used in the research instrument. For computational purposes, "uncertain" was always scored "2," while the favorable response (either "agree" or "disagree," depending on item phrasing) was scored "3" and the unfavorable response was scored "1." The higher an individual scored within a given level or across levels, therefore, the more favorable, positive, or "over—favorable" responses he made. Appendices D and E indicate how items were re-scored from the original—response answer sheets. CHAPTER V ANALYSIS OF RESULTS As indicated in Chapter III, the present degree of theory precision permitted analysis of only seven semantic paths, or ordered sets of level-member items. Data from analysis of these seven paths-—six level members in paths A - F and four level members in Path G-—are presented below. The level members of each path were administered in two orders: (a) to one group of subjects in a ran— dom order; and (b) to a second group of subjects in the order corresponding to the level identification, level I first, level II second, etc. Data for each path are given below in the following order: (a) the matrix of level-by—level correlations for the randomly administered level members, arranged in the random order; (b) a matrix of the same correlations, arranged in the hypothesized order (I, II, III, etc.); (c) a third matrix of the same correlations, arranged in the "best" order, as deter- mined by Kaiser's procedure; (d) the matrix of level-by- level correlations for the second administration of the 84 85 same path (administered in the hypothesized order); and (e) a matrix of this second set of correlations, arranged in the best order. The Q? value is indicated for each matrix, as well as the critical value, at the .05 level, for the significance of correlations in the matrix. In the final section of the chapter, results across all seven paths are summarized and discussed. Particular attention has been paid to comparisons between 3? values for random and hypothesized orders of the same data. Semantic Path A Semantic path A comprises level members I l (Societal stereotype), II 1 (Personally-assigned group status), III 1 (Moral evaluation), IV 1 (Hypothetical personal behavior), V 1 (Actual personal feelings), and VI 1 (Actual personal behavior). The Q2 values for path A randomly administered (Figure 5.01) were: (a) order of administration: 0.617; (b) hypothesized order: 0.823; and (0) best order: 0.942. 0f 15 distinct correlations, ten were significant at the .05 level. The hypothesized order, therefore, gener- ated a better simplex approximation than did the random order; the best order of levels was I, II, V, IV, III, VI (a reversal of levels III and V from the hypothesized order). The 32 values for path A administered in the hypo- thesized order (Figure 5.02) were: (a) order of adminis- tration: 0.877; and (b) best order: 0.899. The best order 86 .00.0 n He>eH 00. e0 c He esHee HeeHeHee He: u 2 "see: - LmQEmE - LopEmE H > H H> H >H H HHH H H H HH HeemH H +3 H > H :H H HHH H HH H H HeeeH 00.H 0m.0 em.0 HH.0 HH.0 HH.0 H > 00.H 0m.0 HH.0 HH.0 00.0 HH.0 H He 00.H 03.0 03.0 HH.0 00.0 H H» 00.H Hm.0 00.0 HH.0 :H.0 H > 00.H 00.0 0m.0 00.0 H 3H 00.H 00.0 mm.0 00.0 H >H 00.H HH.0 40.0 H HHH 00.H HH.0 «0.0 H HHH 00.H nu.o H H 00.H 03.0 H HH 00.H H HH :0.H H H omm.o H mm "Lona? pmmm mmm.m n ufl “coHmepchHEnm Ho LmULe mmsHm> em mam ncoHpmHmLHom mimosa HweHHQCpoam: CH oanHMHnHfism .q gum; 0szmEmm m0.m meeeHm .mm.0 u Hess m0. 00 m 00 esHme HeeHeHee H00 n z "0002 - emnfioe - . - . Leone: 1 LwnEwE H H> H HHH H 3H H > H HH H - HeeeH H .p H . H 3t H HHH H eH H H Hosea H >H H HH H HHH H H H > H H> HeeeH 00.H 02.0 03.0 0H.0 HH.0 HH.0 H H> 00.H HH.0 HH.0 03.0 HH.0 HH.0 H H: 00.H HH.0 HH.0 HH.0 HH.0 00.0 H >H 00.H HH.0 0.0 0H.0 e0.0 H HHH 00.H HH.0 mm.0 HH.0 HH.0 H e 00.H mH.0 0e.0 Hm.0 HH.0 H HH 00.H mm.0 HH.0 Hm.0 H >H 00.H HH.0 HH.0 HH.0 H 0H 00.H No.0 HH.0 0e.0 H HHH 00.H HH.0 HH.0 H 3 00.H mH.0 e0.0 H HHH 00.H Hm.0 HH.0 H H 00.H 0e.0 H HH 00.H 0H.0 H HH 00.H 0H.0 H > 00.H H H 00.H H H 00.H H H> m:0.0 u mm "emeeo seem mmw.0 u mm "geese HeeHeeeeoeH: HH0.0 u mm "coHeecemHeHse< Ho hmeco mmsHm> m was mcoHpmHmpHoo HommechHEU< zHEoocmm H< comm 0HpcmEmm m Ho.m mmaon 87 of levels was II, I, III, IV, VI, V (a reversal of levels I and II and of levels V and VI from the hypothesized order). Of 15 distinct correlations, 11 were significant at the .05 level. Semantic Path B Semantic path B comprises level members I l (Societal stereotype), II 1 (Personally-assigned group status), III 2 (Self concept), IV 1 (Hypothetical personal behavior), V 1 (Actual personal feelings), and VI 1 (Actual personal behavior). The Q2 values for path B randomly administered (Figure 5.03) were: (a) order of administration: 0.652; (b) hypothesized order: 0.886; and (0) best order: 0.918. Of 15 distinct correlations, 12 were significant at the .05 level. The hypothesized order, therefore, generated a better simplex approximation than did the random order; the best order of levels was I, II, III, V, IV, VI (a reversal of levels IV and V from the hypothesized order). The Q2 values for path B administered in the hypothe— sized order (Figure 5.04) were: (a) order of administration: 0.879; and (b) best order: 0.905. The best order of levels was I, II, III, V, IV, VI (a reversal, paralleling that occurring in the random-order administration, of levels IV and V from the hypothesized order). Of 15 distinct corre- lations, all 15 were significant at the .05 level. '88 .Hm.0 n HmeeH m0. 00 m 00 meHse HmeHeHee H0m u z ”0002 LmnEme Losses H H> H >H H > m HHH H HH H H Hm>wq H H> H > H >0 m HHH H HH H H Hm>mq 00.H H0.0 02.0 02.0 02.0 H2.0 H H> 00.H 02.0 H0.0 02.0 02.0 H2.0 H H> 00.H 00.0 02.0 02.0 m0.0 H >0 00.H 00.0 20.0 00.0 m0.0 H 0 00.H 20.0 00.0 m0.0 H > 00.H 02.0 02.0 m0.0 H >H 00.H Hm.0 m0.0 m HHH 00.H H0.0 00.0 m HHH 00.H 00.0 H HH 00.H 00.0 H HH 00.H H H 00.H H H m00.0 n 00 ”00000 0000 000.0 H mm ”cerme.eHeHEHa 00 eeeee mmsHm> mw 0cm mcoHumHmLLou ”L00L0 HmoHmeuomhr CH memHchHE0< .0 cpmm oHpcmsmm 20.0.0020Hu .0H.0 u He>mH 00. 00 m 00 00He> HmeHeHee H00 u z ”0002 LmnEmE . - . omnfimfi LmnEmE H H> H >H H > m HHH H 0H H H Hm>wH H H> H 0 H >0 0 HHH H H0 H H Hm>ma H >H m HHH H H> H H H HH H > Hm>mq 00.H 00.0 Hm.0 0m.0 0H.0 No.0 H H> 00.H 0.0 00.0 0m.0 0H.0 No.0 H H> 00.H 0m.0 00.0 00.0 0m.0 02.0 H >H 00.H 02.0 0m.0 0m.0 00.0 H >H 00.H 02.0 mm.0 mm.0 mm.0 H > 00.H 0m.0 22.0 20.0 mm.0 m HHH 00.H mm.0 mm.0 0m.0 . H > 00.H 0m.0 0m.0 00.0 H HH 00.H No.0 0H.0 Hm.0 H H> 00.H 20.0 22.0 m HHH 00.H 20.0 22.0 m HHH 00.H 0m.0 0m.0 H H 00.H 0m.0 H HH 00.H mm.0 H HH 00.H mm.0 H HH 00.H H H 00.H H H 00.H H > 0H0.0 u mm “emeto 0000 000.0 u 00 ”emeeo HeeHemeeeeHz 000.0 u 00 neoHpeeemHeH50< Ho smeeo mmsHm> m 0cm mcoHumHmHLoo ”wmepchHEU< HHEovcmm .0 spam oHucmEmm m m0.0 mmaon 89 Semantic Path C Semantic path C comprises level members I l (Socie— tal stereotype), II 2 (Societal interactive norm), III 1 (Moral evaluation), IV I (Hypothetical personal behavior), V 1 (Actual personal feelings), and VI 1 (Actual personal behavior). This semantic path corresponds to the set of level members first identified by Jordan and used in his original scale. The 02 values for path C randomly ad- ministered (Figure 5.05) were (a) order of administration: 0.588; (b) hypothesized order: 0.771; and (c) best order: 0.972. 0f 15 distinct correlations, ll were significant at the .05 level. The hypothesized order, therefore, generated a better simplex approximation than did the random order; the best order of levels was I, V, IV, III, VI, II (a reversal of levels III and IV, and transposi— tion of levels II and V from the hypothesized order). The 02 values for path C administered in the hypoth— esized order (Figure 5.06) were: (a) order of adminis- tration: 0.812; and (b) best order: 0.884. The best order of levels was I, II, VI, IV, III, V (a reversal of levels III and IV and a transposition of level VI from the hypothesized order). 0f 15 distinct correlations, all 15 were significant at the .05 level. 9() .H0.0 n Hm>mH 00. 00 m 00 00Hm> HmoHpHpo M00 u z “000: H > H HHH H >H H H> 0 HH H H LMMWWm H H> H > H 0H H HHH 0 HH H H gmmmmm 00.H 00.0 00.0 00.0 00.0 00.0 H > 00.H 00.0 00.0 00.0 00.0 00.0 H H> 00.H 00.0 00.0 00.0 00.0 H HHH 00.H 00.0 00.0 00.0 00.0 H > 00.H 00.0 00.0 00.0 H >H 00.H 00.0 00.0 00.0 H 0H 00.H 00.0 00.0 H H> 00.H 00.0 00.0 H HHH 00.H 00.0 0 HH . 00.H 00.0 0 HH 00.H H H 00.H H H 000.0 n 00 ”00000 0000 000.0 n 00 "00H00000H0H000 mo 00000 mmsz> 0m 0:0 mcoHumHmLLoo ”00000 HmoHuoLHoqmz CH UmpmpchHEU< .0 £000 OHpcmemm 00.0 0000H0 . .Hm.0 u Hm>mH 00. 00 M 00 msz> HmoHquo 020 u z "@002 0 HH H H> H HHH H >H H > H H pmwwwm H H> H 0 H >H H HHH 0 HH H H memwm 0 HH H > H H> H H H HHH H >H pmwflwm 00.H 00.0 00.0 00.0 00.0 00.0 0 HH 00.H 00.0 00.0 00.0 00.0 0H.0 H H> 00.H 00.0 00.0 00.0 00.0 00.0 , 0 HH 00.H 00.0 00.0 00.0 0H.0 H H> 00.H 00.0 00.0 00.0 00.0 H > 00.H 00.0 00.0 00.0 00.0 H > 00.H 00.0 00.0 0H.0 H HHH 00.H 00.0 00.0 0H.0 H 0H 00.H 0H.0 00.0 00.0 H H> 00.H 00.0 0H.0 H >H 00.H 00.0 0H.0 H HHH 00.H 0H.0 0H.0 H H 00.H 00.0 H > 00.H 00.0 0 HH 00.H 00.0 H HHH 00.H H H . 00.H H H 00.H H >H 000.0 H NM ”00000 0000 H00.0 u mm "nacho H00Humzpoazz 000.0 n mm ”coHpmppchHEu< go Lmono mmst> 00 0:0 wCOHumHmppoo "swampchHE©< szovcmm .0 numm OHucmEmm n0.n unstn 91 Semantic Path D Semantic path D comprises level members I l (Socie— tal stereotype), II 2 (Societal interactive norm), III 3 (Proclaimed laws), IV 1 (Hypothetical personal behavior, V 1 (Actual personal feelings), and VI 1 (Actual personal behavior). The Q2 values for path D randomly administered (Figure 5.07) were: (a) order of administration: 0.650; (b) hypothesized order: 0.791; and (c) best order: 0.920. 0f 15 distinct correlations, all 15 were significant at the .05 level. The hypothesized order, therefore, gener- ated a better simplex approximation than did the random order; the best order of levels was I, V, IV, III, VI, II (a reversal, paralleling that in the random administration of path C, of levels III and IV, and transposition of levels II and V from the hypothesized order). The 02 values for path D administered in the hypo- thesized order (Figure 5.08) were: (a) order of adminis- tration: 0.812; and (b) best order: 0.902. The best order of levels was I, II, VI, III, IV, V (a transposition of level VI from the hypothesized order). 0f 15 dis- tinct correlations, 12 were significant at the .05 level. Semantic Path E Semantic path E comprises level members I l (Socie— tal stereotype), II 3 (Group-assigned personal status), III 2 (Self concept), IV 1 (Hypothetical personal behavior), 92 .H0.0 u Hm>mH 00. 00 M 00 msHm> HmoHuHHo ”00 u “@002 pmnEwE LoQEmE H > H >H m HHH H H> m HH H H Hm>®q H H> H > H >H m HHH m HH H H Hw>wH 00.H mm.0 H2.o 0m.0 0H.0 mH.0 H > 00.H 0m.0 m0.0 H0.0 00.0 0m.0 H H> 00.H 00.0 00.0 m2.0 mm.0 H >H 00.H m0.o H2.0 0H.0 0H.0 H > 00.H H0.o 02.0 Hm.0 m HHH 00.H 00.0 m2.0 mm.0 H >H 00.H 00.0 0m.0 H H> 00.H 02.0 Hm.0 m HHH 00.H 0H.0 m HH 00.H 0H.0 m HH 00.H H H 00.H H H 000.0 u mm ”00000 Hmmm 0H0.0 u um ”coHpmpHchH50< 00 pmcgo C mmzH0> .m 0cm mcoHHmHmLLcu ”00000 HmoHHmcpoamz :H UmLmHchHEU< .0 spam OHHcmEmm n. m0.0 mmimHm .Hm n Ho>mH 00. 00 M 00 wsz> HmoHpHLo H0 u z umpoz meEmE . . . J LmnEmE u m pmpEmE m HH H H> m HHH H >H H > H H Hm>®H 0 H: H > H 3H m HHH a HH H H Hw>mH c HH H >H H H> H H H > HHH Hw>mH 00.H 00.0 00.0 m0.0 22.0 0m.0 m HH 00.H 02.0 20.0 H0.0 m0.0 0m.0 H H> 00.H m0.o m0.0 0m.o 22.0 00.0 m HH 00.H H0.0 20.0 02.0 0m.0 H H> 00.H 00.0 02.0 22.0 02.0 H > 00.H 20.0 0m.0 00.0 m0.o H >H 00.H m0.0 02.0 0m.o m HHH 00.H m0.0 m0.0 0m.0 H >H 00.H 0m.0 02.0 H0.0 H H> 00.H 00.0 0m.0 H >H 00.H 00.0 0m.0 m HHH 00.H 02.0 0m.0 H H 00.H 02.0 H > 00.H 0m.0 m HH 00.H 02.0 H > 00.H H H 00.H H H 00.H m HHH 000.0 a mm ”pmupo ummm H00.0 u mm ”00000 HmoHHmcpoazm 000.0 u mm ”coHummeHcHEU< no nacho mwsHm> m 020 mcoHpmHmLpoo ”vmpwpchHE0< HHEoccmm .0 Spam 0Hucmem m 00.0 mmDOHm 93 V 1 (Actual personal feelings), and VI 1 (Actual personal behavior). The Q2 values for path E randomly administered (Figure 5.09) were: (a) order of administration: 0.361; (b) hypothesized order: 0.861; and (c) best order: 0.955. 0f 15 distinct correlations, eight were significant at the .05 level. The hypothesized order, therefore, generated a better simplex approximation than did the random order; the best order of levels was I, II, III, VI, V, IV (a reversal of levels IV and VI from the hypothesized order). The Q2 values for path E administered in the hypoth— esized order (Figure 5.10) were: (a) order of adminis— tration: 0.865; and (b) best order: 0.872. The best order of levels was I, III, II, IV, V, VI (a reversal of levels II and III from the hypothesized order). 0f 15 distinct correlations, 12 were significant at the .05 level. Semantic Path F Semantic path F comprises level members I 1 (Socie- tal stereotype), II 3 (Group-assigned personal status), III 3 (Proclaimed laws), IV 1 (Hypothetical personal be— havior), V 1 (Actual personal feelings), and VI 1 (Actual personal behavior). The 02 values for path F randomly administered (Figure 5.11) were: (a) order of adminis- tration: 0.885; (b) hypothesized order: 0.850; and (c) best order: 0.9A8. 0f 15 distinct correlations, 91! .Hm.0 u HmpoH 90. pm L La mSHm> HmmHHHLo ”mu u : ”moo: I. SamwnEWE : (Hm. E>E H H> H > H >H m HH m HHH H H s H HH H u H HH m HHH m HH H H stma fl .ng._.fligw 00.H mm.o H©.o Hm.c m.o Ho.o H H» co.H mm.c Hm.o om.o 3m.c Ho.o H H: 00.H mo.o mm.o HH.0 mm.c H > oe.H mm.o HH.0 ma.o wm.o H s ..o mm.o H >H O C) r—Q m Au 0 AX (\J O m (V O c—4 H C‘ C) H (\ J r\. C) (*1 F 00.H mw.o :m.o m HH OH.H mo.o mm.o m HHH mwm.o u :: ”corps Hmmi 04y.o n :0 “:QHHmLHchHrwa at Lanna . n .< o nonHwH la wcm mconmHoLLco ”gonna HmoHuezucuar :H ponomecHqu .m coma oHpcmEmm 0 .Hm.a u HmsmH mo. pm m 00 msHm> HmoHHHLo Hmo u z mopoz H J m . LwQEoE 1 . . a . l l m 4 LmQFQE . s m . . . LQQEmE H :H H > H H> o HHH H. H H HmsmH _H H: H : . :H a HH» H. H H Hm>mH H H: H H H : HH H >H m HHH Hm>mH 2 (‘1 C\ 00.H am.o mm.o mo.o mo.o Hc.o H HH 00.H mm.o mm.o om.o o..o H H> 00.H om.o mm.o Hm.o om.o om.o H H> (\J (x O r—4 O C‘ 00.H 0m.0 0H.0 2H.0 2H.0 H > 00.H 20.0 00.0 H b 00.H H0.0 00.0 2H.0 mm.0 H H :1 m C C) (\l C) r4 H C) C) r-i 00.H 0m.0 0H.0 HH.0 2H.0 H 5H 00.H No.0 2m.0 00.0 H > 00.H mH.o Hm.o a HHH 00.H ms.o As.o m HHH 00.H HH.0 mz.o m HH 00.H we.o m HH 00.H 01.0 m HH . 00.H mH.o H >H 00.H 1 H CC.H H H 00.H (“J H H H F mmm.o n ma “Lovcc pmom H00.0 u 10 ”Lowpo HmmHngpodm: Hmm.o u :m ”coHmeHchHEv< 0o Loopo H1. 0 mosz> 4m ppm mccHHmHoLLoo vapmHmHuHme mecwcmn .m £de QHHCMEmm 95 13 were significant at the .05 level. The hypothesized order, therefore, generated a better simplex approxima— tion than did the random order; the best order of levels was II, I, V, III, VI, IV (a reversal of levels I and II and transposition of levels IV, V, and VI from the hypothesized order). The Q2 values for path F administered in the hypoth— esized order (Figure 5.12) were: (a) order of adminis— tration: 0.887; and (b) best order: 0.977. The best order of levels was II, I, V, IV, III, VI (a reversal of levels I and II and transposition of levels IV and V, somewhat paralleling the random administration of path F). 0f 15 distinct correlations, 1A were significant at the .05 level. Semantic Path G Semantic path G comprises level members I l (Socie- tal stereotype), II 2 (Societal interactive norm), III A (Group identity), and IV 2 (Actual group behavior). As indicated in Chapter III, the present degree of theory precision does not indicate the relationship of levels V and VI to path G. The 02 values for path G randomly ad- ministered (Figure 5.l3) were: (a) order of administration: 0.865; (b) hypothesized order: 0.51M; and (0) best order: 0.965. 0f six distinct correlations, five were signifi— cant at the .05 level. The hypothesized order, therefore, did not generate a better simplex approximation than did 96 .2m.0 n Hm>mH mo. 00 M 00 osHm> HmoHoHpo M02 u : ”mooz LooEoE .. prEmE H H> m HHH H >H H > H H m HH Hm>oq H H: H > H >H m HHH m HH H H Ho>m2 00.H 02.0 05.0 0m.0 Hm.0 mm.0 H H> 00.H 0m.0 05.0 02.0 mm.0 Hm.0 H H> 00.H 05.0 No.0 02.0 m:.0 m HHH 00.H m5.0 No.0 mm.0 m5.0 H > 00.H m5.0 20.0 H2.0 H >H 00.H 05.0 HH.0 20.0 H >H 00.H 30 3.0 H H. 00.H 90 9.10 mHHH 00.H 05.0 H H 00.H 05.0 m HH 00.H m HH 00.H H H 550.0 u mm ”poppo Hmmm 500.0 n mm "coHmepchHE00 Lo pmvuo mmsz> mm 0cm mcoHHmHmLpoo ”bmHLo HmoHumLooomz CH omLmHchH50< .0 nomm oHucmEom mH.m mm00Hm .mm.0 "Hm>mH m0. um M 00 msHm> HmOHoHpo “Hm u "@002 LooEmE . LopEoE meEmE H >H H H> m HHH H > H H m HH Ho>m2 H H> H > H >H m HHH m HH H H Hm>mq H >H H H> H H H > m HHH m HH Hm>mq 00.H 00.0 2m.0 mm.0 0m.0 2H.0 H >H 00.H mm.0 00.0 00.0 Hm.0 Hm.0 H H> 00.H 00.0 0m.0 mm.0 2m.0 2H.0 0 >H 00.H 00.0 mm.0 Hm.0 Hm.0 H H> 00.H mm.0 mm.0 mm.0 mm.0 H > 00.H Hm.0 mm.0 00.0 Hm.0 H H> 00.H mm.0 m2.0 .02.0 m HHH 00.H 2m.0 2H.0 0m.0 H >H 00.H mm.0 m2.0 H0.0 H H 00.H mm.0 mm.0 H > 00.H 02.0 m2.0 m HHH 00.H mm.0 mm.0 H > 00.H H0.0 H H 00.H H0.0 m HH 00.H 02.0 m HHH 00.H m HH 00.H H H 00.H m HH 020.0 mm ”bongo pmmm 0m0.0 u mm ”pmupo HooHpmcpoazz m2m.o n mm “coHHMLuchHEU< no pmnho mmsHm> m 0cm mCOHpmHoHLoo uvmpmpchHEu< mHeoncmm .m cpmm OHucmEmm m HH.m mmDOHm 97 the random order; the best order of levels was II, IV, I, III (transpositions of all four levels from the hypothe- sized order but a reversal of only levels I and IV from the random order of administration). The Q2 values for path G administered in the hypo— thesized order (Figure 5.14) were: (a) order of adminis- tration: 0.768; and (b) best order: 0.957. The best order of levels was I, III, II, IV (a reversal of levels II and III from the hypothesized order). 0f six dis- tinct correlations, four were significant at the .05 level. Summary For six of the seven paths analyzed, the Q2 value for the randomly administered, randomly ordered matrix was less than the 02 value for the randomly administered, hypothetically ordered matrix; the one exception involved semantic path G, comprising only four level members and six distinct correlations. 0n the other hand, in no case-- either of random administration or of hypothetically or- dered administration——did the hypothesized ordering of correlations generate the best simplex approximation. The hypothesized ordering principle, therefore (from no strong facets to all strong facets in succeeding level members), generally produced a better-than-random order but never the best order. On the other hand, no general ordering principle which would improve on the hypothesized ordering principle was immediately obvious; many of the S98 .mm.0 u Ho>mH m0. 00 M mo osHm> HMOHpHpo mmm u z ”moo: LmnEoE LonEmE m >H m HH 2 HHH H H Hm>mH m >H 2 HHH m HH H H Hm>mH 00.H m0.0 5m.0 m0.0 m >H 00.H 5m.0 m0.0 m0.0 m >H 00.H mm.0 m0.0 m HH 00.H mm.0 H0.0 2 HHH 00.H H0.0 2 HHH 00.H 00.0 m HH 00.H H H 00.H H H 5mm.o u mm ”tango pmmm 005.0 n mm ”coHpmpHchHEea 00 .0220 mmsHm> mm 0cm mcoHpmHoLLoo ”gouge HmoHpmcpoqam CH ompmHchHEU< .0 comm oHpcwEom 2H.m mm00Hm - .Hm.0 I Ho>mH mo. pm M mo msHm> HmoHuHLo How u z "ouoz . Lenses LmoEmE LooEmE 2 HHH H H m >H m HH Ho>oH m >H 2 HHH m HH H H Hm>oq 2 HHH m >H H H m HH Ho>m2 00.H H2.0 mm.0 50.0 2 HHH 00.H mm.0 05.0 0m.0 m >H 00.H mm.0 H2.0 50.0 2 HHH 00.H 0m.0 0m.0 H H 00.H 50.0 H2.0 2 HHH 00.H. 0m.0 05.0 m >H 00.H 05.0 m >H 00.H mm.0 m HH 00.H 0m.0 H H 00.H m HH 00.H H H 00.H m HH m0m.o n mm ”emcee ummm 2Hm.o u mm ”canto HmOHumcsoaHz m00.o u mm "coHumpuchH522 Ho emepo wmsHm> m 02m mcoHpmHmppoo “ompmpchHEo< >HEoccmm .0 spam oHucmEmm N MH.m mmDOHm 99 best orders appeared to involve few reversals from the hypothesized order. Since in all but one case the matrix generated by listing correlations according to the hypothesized se- mantic path order was a better simplex approximation than that generated from the random order of administration, (a) order of administration was apparently not of pri- mary importance in simplex approximation, and (b) some indirect support was given to the assumption that more than mere verbalizations were involved in responses to a given set of items. Finally, the lack of an ordering principle obviously better than the hypothesized one, and the generally close correspondence between hypothe- sized and best orders, suggested that the hypothesized ordering principle, the level members identified, and the orders hypothesized among those level members, are useful extensions of the Guttman — Jordan formulations. Further discussion of results obtained, implications for the theoretical framework tested, and suggestions for further study are given in the following chapter. CHAPTER VI SUMMARY AND CONCLUSIONS The primary purpose of the present research was to examine a reformulation of the Guttman and Jordan-attitude scale construction theory. A summary of that research is followed below by an extended discussion of some results, a listing of some additional problems in theory, and some suggestions for further study. Summary of Research The present research was primarily methodological. The problem, research design, and results are summarized below. Statement of Problem Guttman, analyzing the structure of attitude items, identified four levels, or types of attitudes, generated from permutations of three facets within each item. He hypothesized that, according to his principle of contiguity, the matrix of level-by-level correlations would approximate a simplex. 100 101 Jordan extended Guttman's analysis to a five-facet six—level system. Data on a scale on attitudes toward the mentally retarded did generate simplex approximations. Jordan did not examine fully (a) his choice of specific facet permutations for each level; (b) the effect of order of administration on relative size of correlations among levels; and (c) the effect of invariant directionality in answer foils. Several related problems therefore remained: (a) identification of all possible facet permutations; (b) examination of effects from order of administration and from item directionality; (0) construction of an ex- perimental instrument dealing with the emotionally dis- turbed rather than the mentally retarded; and (d) ap- plication of appropriate statistical tests. Research Design Of 32 permutations of five dichotomized facets, only 12 appeared semantically possible--Jordan's six and an additional six. These 12 level members were hypothesized ‘to be ordered within seven semantic paths; each path comprised six or four of the 12 level members. Directions for items of each level member were simplified and phrased :in parallel form; all items were phrased so that all sets (3f answer foils were identical. A set of experimental instruments was developed, two for each of the semantic paths (one instrument with the 102 level members in a random order and one instrument with the level members in the hypothesized order). Items were written so that while content of answer foils re— mained identical, directionality of foils and gram- matical emphasis within items varied randomly. The 1U varying instruments were administered in random order, one to a subject, to 825 students in undergraduate psychology and education. Data from the 1A sub-groups were analyzed by Kaiser's Q2 procedure, which evaluates individual correlation matrices and orders the variables within the matrices. A Q2 value was determined for the hypothesized and best orderings of level members within each semantic path; for the random—order administrations of semantic paths, the g2 value was also determined for the particular random order of administration. Results For six of the seven paths analyzed, the Q2 value for the randomly administered, randomly ordered matrix was less than the g2 value for the randomly administered, hypothetically ordered matrix; in no case did the hypo- thesized ordering of level members generate the best simplex approximation. On the other hand, the lack of an ordering principle obviously better than the hypo- thesized one and the generally close correspondence 103 between hypothesized and best orders suggested that the hypothesized ordering principle, the level members identified, and the orders hypothesized among those level members, are useful extensions of the Guttman- Jordan formulations. Discussion of Results The results obtained generally supported the theory being examined. Of particular interest are: (a) se- mantic path B, (b) the ordering of all 12 level members, and (c) conjoint-disjoint struction interation. Semantic Path B Semantic path B (of. Figs. 5.03 and 5.0“) is of particular interest for several reasons. The Q2 values obtained for the hypothesized ordering of level members were, when averaged (0.883), the best of any semantic path (0.886 from the re-ordering of the random adminis— tration and 0.879 from the administration in hypothe— sized order). The differences between g2 values for the hypothesized and best orders were the smallest for any semantic path. The best order, obtained by Kaiser's procedures, of the level members involved a reversal of only levels IV and V in both administrations of the path. Semantic path B differs from the semantic path originally identified by Jordan (path 0) on levels II and III. In place of Jordan's "Societal interactive norm" IOU on level II, path B has "Personally—assigned group status"; in place of "Moral evaluation" on level III, path B has "Self concept." Since the difference be- tween 02 values (an average of 0.883 for path B and 0.792 for path 0) cannot be assigned statistical significance, and since examination of the level members involved, as well as of the other identified level members, is still tentative, neither path could presently be identified as "best"-—i.e., as having the most psychologically useful level members and as gen- erating the consistently best simplex approximations. Ordering of 12 Level Members Because some correlations among the 12 level mem- bers are missing--the level members do not appear to- gether in the same path-—a complete ordering of all 12 is impossible from the present data. Results obtained, however, would appear to Justify further exploratory research, using level members across paths to obtain the missing values. Some rough comparisons can however be made. An approximate order within levels can be estimated by averaging the available correlations for level members within the level and then re—arranging the level members to reduce the absolute number of simplex violations within each level. Figure 6.01 is a matrix of averaged 105 .mHm>mH pcopmadHU mcofim mcHLmULo mzu woesmmm 0cm o>Hpmocmp zHco mH who: ompcmmmpq mcHLmULo one .Lumq oHpcmEmm oflmm osp.cH Lmoaom ooc 00 0m>Ho>cH mLoDEoE Ho>oH 03» on» maps: Lzooo mpmo wchmHS .H2H.m IH0.m mozsaH .000 Umchpoo mcoHpmHobLoo oLoE no 020 0:0 00 mowmno>m mHoEHm mp0 nmumochH wcoHpmHoLLoo ”moo: H H> H > H >H m >H H HHH 2 HHH m HHH m HHH m HH H HH m HH H H nwmwww 00.H mm. 0m. nu m2. In mm. mm. 25. mm. mm. mm. H H> 00. mm. In 02. I: 02. mm. 0m. 0m. mm. mm. H > 00.H II mm. II mw. wt. H2. mm. mm. om. H >H 00.H In mm. In In 05. II In Hm. m >H 00.H -u I- .. 2m. om. u- cm. H HHH 00.H In In ON. I: In Hm. 2 HHH 00.H :1 mm. nu m2. m2. m HHH 00.H II m5. mm. mm. m HHH 00.H :1 ,II Hm. m HH 00.H II mm. H HH 00.H m0. m HH 00.H H H Ho.w mmDOHm umpmwpo mHm>HompcoE .mponEmE Ho>oq mH mcoe< mcoHpmHopLoo oommpm>< Ho prpmz 106 available correlations, arranged to reduce the number of simplex violations within levels II, III, and IV. Conjoint-Disjoint Struc— tion Interaction An a priori relation between conjoint and disjoint struction was assumed in the construction of the SAF scales. As indicated in Figure H.03, Status-descriptive items appear in level members I 1, II 1, II 3, III 2; Action-descriptive items in level members II 2, III 1, III 3, IV I, IV 2, VI 1; and Feeling-descriptive items in levels III A and V l. The effect of varying content on simplex ordering may be of interest. In the seven semantic paths, S, A, and F content items appear in mixed arrangements; se- mantic path A, for example, has S content on levels I and II, followed by A content on levels III and IV, followed by F content on level V and then A content on level VI; S, A, and F contents are similarly intermingled in all seven paths. The re-ordering of level members according to Kaiser's procedures, however, in 12 of IA cases, grouped all S content together and all A content together (only one F-content level member appears in any semantic path). In nine cases, level members were grouped S - F - A; in two case, S - A — F, and in only one case F - S - A .(path G). Order within such groupings did not appear 107 consistent; the general trend of such grouping there- fore, (a) did not indicate a consistent ordering principle for improving on the present conjoint-struction principle; (b) suggested that conjoint and disjoint struction in- teract; and (0) suggested that, within disjoint struction, S-content items are weaker than F-content items, which are in turn weaker than A-content items. Problems in Theory For subjects in the present research, no control was established over age or sex, although undergraduate students in introductory psychology and education courses were assumed to be homogeneous enough for present purposes. The possibly high degree of education and intelligence (compared to total population) of such subjects may have had an effect on their ability to discriminate among fairly similar items--hence, on the patterns of corre- lations among those items. Similarly, Rokeach (1968) has suggested that degree of discrimination varies with amount of open and closed—mindedness. Jordan's prelim- inary data indicate that simplex patterns hold up across cultures; the evidence that such patterns hold up even within cultures but across varying levels of sophisti- cation is less strong. Further research may be needed to establish that the discriminatory ability necessary to distinguish among varying level members is found in the less sophisticated. 108 As indicated in Chapter III, no claim is made that all possible phrasings of attitude items have been identi- fied--what, for example, of "We think we should . . ." or "I expect other persons to . . ."? Such items appear to have psychological meaning but are not presently assigned to any level member in the semantic map. Of the items presently identified, some may be considerably less ef— fective than others; a standard item analysis might be helpful. Finally, although Kaiser's g2 appeared sufficient for the purposes of the present study, a more rigorous test of simplex approximation is needed. Particular attention should be paid in such a test to such factors as general magnitude of correlations and the effect of equal and unequal spacings among entries. Suggestions for Further Study Some suggestions for further study have been made within the above discussion. The following suggestions, both for additional analyses of the present data and for the construction of new experiments, appear most useful. Present Data As indicated in Chapter IV, items were randomly assigned to one of four variants of directionality and grammatical emphasis. Such random assignment was performed 109 as a control; direct analysis of the effects and inter- action of these variables is possible. Procedures for an appropriate analysis of variance are indicated at the conclusion of Appendix F. Additional variables which might be of interest in an analysis of variance are level (separate analyses for each path) and SAF content (within and across paths). New Experiments An ordering of all 12 level members now appears possible. Such an ordering would require at least one correlation for all combinations of two level members. A set of instruments could therefore be prepared which would include level members across various paths; such instruments would not be directly subject to the criterion of simplex approximation (of. the conclusion of Chapter III) but would provide the data missing in a matrix such as Figure 6.01. Although the reasons for varying SAF content across levels appeared compelling in the present study (cf. Chapter IV), the apparent interaction of conjoint and disjoint struction noted above suggests that an attempt be made to reconstruct the 12 level—member sub-scales with a common content across all 12. Results from administration of such scales would be useful in determining the exact character of such interaction. 110 Finally, the validation of the experimental scales here proposed remains to be done. Jordan (1968a) has suggested a variety of predictor variables related to attitude; Rokeach (1968) described personality variables which may be directly related to patterns of conjoint struction. Conclusion Some general conclusions may be drawn from the re— sults of the present study: (a) Guttman facet design provides a workable approach to attitude-scale construc- tion; (b) the particular extensions suggested within the present study are useful interpretations of the Guttman— Jordan approach; and (c) additional research is needed, particularly on conjoint-disjoint struction interaction. Such research may clarify the relationships suggested among perceptions of self and others, of group expecta- tions and moral evaluations, of feelings and specific acts. APPENDICES APPENDIX A ATTITUDE—BEHAVIOR SCALE——MR I ATTITUDE—BEHAVIOR SCALE——MR The following directions and items are taken from the Attitude-Behavior Scale—-M. R. (ABsuMR). Because of the limited purposes of the present study, the only di- rections and items reproduced below are those from the six sections which correspond to Jordan's (1968) hypo— thesized six levels (Chapter II, Figure A). The complete research instrumentl contains ad- ditional items dealing with demographic data, with at- titudes toward man's sense of control over his environ- ment (efficacy), with the importance of religion, with amount and type of contact with the attitude object, and with knowledge of the attitude object. In addition each of the items within the six scale sections indicated be- low is matched, in the original scale, with a question of the form "How sure are you of your answer? (I) not sure; (2) fairly sure; (3) sure." This matching set of questions is designed to measure what Guttman refers to as "intensity." Such measures of intensity have not been included in the instrument designed for the present study. 1Several studies are scheduled for completion in 1969 using the complete ABS-MR (Gottlieb, 1969; Harrelson, 1969; Jordan, 1969; Morin, 1969). 113 114 Collection of demographic and similar data was also be— yond the scope of the present study, although comments about intensity analysis are included in Chapter VI, ”Summary and Conclusions.” Finally, only the first two items from each of the attitude-level sections are re— produced below: the purpose is only to indicate the general structure of Jordan's original instrument. The complete ABS—MR is contained in Gottlieb (1969), Horrelson (1969), and Morin (I969). ififii‘ ATTITUDE BEHAVIOR SCALE——MR by John E. Jordan DIRECTIONS This booklet contains statements of how people feel about certain things. In this section you are asked to indicate for each of these statements how most other people believe that mentally retarded peOple compared to people who are not retarded. Directions: Section I In the statements that follow you are to indicate on the answer sheet how other people compare mentally retarded persons to those who are not mentally retarded. Other people generally believe the following things about the mentally retarded as compared to those who are not retarded: '“*“““ 1. Energy and vitality 2. How sure are you of this answer? 1. less energetic 1. not sure 2. about the same 2. fairly sure 3. more energetic 3. sure 3. Ability to do school A. How sure are you of this an- work swer? 1. less ability 1. not sure 2. about the same 2. fairly sure 3. more ability 3. sure Directions: Section II This section contains statements of ways in which other people sometimes act toward people. You are asked to indicate for each of these statements what other people generally believe about interacting with the mentally re- tarded. Other people generally believe that mentally retarded persons ought: ‘ Al. To play on the school playground with other children who are not mentally retarded: I. usually not approved 2. undecided 3. usually approved ull6 43. To visit in the homes of other children who are not mentally retarded: 1. usually not approved 2. undecided 3. usually approved Directions: Section III This section contains statements of the "right“ or "moral" way of acting toward people. You are asked to in— dicate whether ypugyourself agree or disagree with each statement accordifig to how you personally believe you ought to behave. In respect to people who are mentally retarded, do you be- lieve that it is usually right or usually wrong: 81. To take a mentally retarded child on camping trips with normal children: 1. usually wrong 2. undecided 3. usually right 83. To permit a mentally retarded child to go to the movies with children who are not mentally retarded: 1. usually wrong 2. undecided 3. usually right Directions: Section IV This section contains statements of ways in which people sometimes act toward other people. You are asked to indicate for each of these statements whether you personal- l1 would act toward mentally retarded people. In respect to a mentally retarded person, would you 121. Share a seat on a train for a long trip: 1. no 2. don't know 3. yes 117- 123. Have such a person as a fellow worker: 1. no 2. don't know 3. yes Directions: Section V This section contains statements of actual feelings that people may hold toward the mentally retarded. You: are asked to indicate howpyou feel toward people who are mentally retarded compared to peOple who are not mentally retarded. —_— How do you actually feel toward persons who are mentally retarded compared to others who are not mentally retarded: 1. Disliking l. more 2. about the same 3. less 3. Fearful 1. more 2. about the same 3. less Directions: Section VI This section contains statements of different kinds of actual experiences you have had with mentally retarded persons. If the statement applies to you, mark ye_; ’If not you should mark pg. Experiences or contacts with the mentally retarded: 41. Shared a seat on a bus, train, or plane: 1. no 2. uncertain 3. yes 43. Eaten at the same table together in a restaurant: 1. no 2. uncertain 3. yes APPENDIX B ITEM VARIANTS FOR 12 LEVEL MEMBERS ITEM VARIANTS FOR 12 LEVEL MEMBERS As indicated in Chapter IV, two aspects of phrasing appear in these variations: directionality of the item——i.e., the determination of the more favorable response, relative to the attitude object: "agree" or "disagree"; and grammatical position of the "actor." An individual item may be phrased so that agreement is either favorable or unfavorable toward the attitude object; an item may also be phrased so that either the actor or the attitude object is grammatically emphasized as sub— ject of the clause. Each item can therefore be phrased four different ways without a change in specification of situation or "disjoint struction." The following variants of item phrasing indicate ways in which each of the items in the research instrument could be phrased. The variants have been assigned code numbers to correspond with data—processing requirements: variants 1.1 and 1.2 are phrased so that the response "agree" implies favorableness toward the attitude object; for variants 2.1 and 2.2, "agree" implies unfavorableness. In variants 1.1 and 2.1 the actor is placed in the position of grammatical emphasis; in variants 1.2 and 2.2 the at- titude object is placed in that position. 119 120 The particular variant used for each item in the res search instrument was picked by random assignment. Dis“ joint-struction specificity—vi.e., the particular situa— tional context of each item—-is discussed in the introduction to Appendix 0. Appendix D contains an explanation of various data—processing operations used to re-arrange and re—score items prior to correlational analyses. Finally, only one item from each of the twelve level members is presented here in all four variants. Appendix C includes the complete set of items actually used from each level member, coded as indicated above. Level I, Member 1 1.1 We think that we are less intelligent than most emotionally disturbed persons. 1.2 We think that most emotionally disturbed persons are more intelligent than we are. 2.1 We think that we are more intelligent than most emotionally disturbed persons. 2.2 We think that most emotionally disturbed persons are less intelligent than we are. Level II, Member 1 1.1 I think that we are less intelligent than most emotionally disturbed persons. 1.2 I think that most emotionally disturbed persons are more intelligent than we are. 2.1 I think that we are more intelligent than most emotionally disturbed persons. 2.2 I think that most emotionally disturbed persons are less intelligent than we are. 121 Level II, Member 2 Level 1.1 We think that we have some friends who are emotionally disturbed. We think that some of our friends are emotional— ly disturbed. We do not think that we have emotionally dis— turbed friends. We do not think that friends of ours are emotionally disturbed. II, Member 3 1.1 11;, Other people think that I am less intelligent than most emotionally disturbed persons. Other peOple think that most emotionally dis— turbed persons are more intelligent than I am. Other people think that I am more intelligent than most emotionally disturbed persons. Other peOple think that most emotionally dis— turbed persons are less intelligent than I am. Member 1 Level Level 1.1 1.2 We should be friends with people who are emo- tionally disturbed. Persons who are emotionally disturbed should be accepted as our friends. We should not be friends with persons who are emotionally disturbed. Persons who are emotionally disturbed should not be accepted as our friends. III,Member 2 1.1 1.2 I think that I am less intelligent than most emotionally disturbed persons. I think that most emotionally disturbed persons are more intelligent than I am. Level 122 2.1 I think that I am more intelligent than most emotionally disturbed persons. 2.2 I think that most emotionally disturbed persons are less intelligent than I am. III, Member 3 Level 1.1 Other people expect me to have friends who are emotionally disturbed. 1.2 Other people expect persons who are emotionally disturbed to be accepted as my friends. 2.1 Other people expect me not to have friends who are emotionally disturbed. 2.2 Other people expect persons who are emotionally disturbed not to be accepted as my friends. III,#Member A Level 1.1 We are not frightened by emotionally disturbed persons. 1.2 Emotionally disturbed persons do not frighten us. 2.1 We are frightened by emotionally disturbed persons. 2.2 Emotionally disturbed persons frighten us. IV, Member 1 Level 1.1 I would have.friends who are emotionally dis- turbed. 1.2 Persons who are emotionally disturbed would be accepted as my friends. 2.1 I would not have friends who are emotionally disturbed. 2.2 Persons who are emotionally disturbed would not be accepted as my friends. IV, Member 2 1.1 We have friends who are emotionally disturbed. 1.2 Some of our friends are emotionally disturbed. 123 2.1 We do not have friends who are emotionally disturbed. 2.2 None of our friends are emotionally disturbed. LevelfV, Member 1 1.1 I am not frightened by emotionally disturbed persons. 1.2 Emotionally disturbed persons do not frighten me. 2.1 I am frightened by emotionally disturbed persons. 2.2 Emotionally disturbed persons frighten me. LeveYleI3 Member 1 1.1 I have friends who are emotionally disturbed. 1.2 Some of my friends are emotionally disturbed. 2.1 I do not have any friends who are emotionally disturbed. 2.2 None of my friends are emotionally disturbed. APPENDIX C ATTITUDE—BEHAVIOR SCALES SAF SCALE: FORM ED—l ATTITUDEFBEHAVIOR SCALES SAF SCALE: FORM ED-l The twelve sets of 20 questions presented here are the actual sets of item variants used in the present re- search. Each of the sets of 20 questions represents a specific level number (see Chapters III and IV). No sub— ject was asked to respond to more than six sets of 20 questions; the particular sets of questions to be given as a unit have been specified in Chapters III and IV. Since, however, some sets of 20 questions appeared in several of the various 6—set and A-set combinations, corresponding to specified semantic paths, duplication of every set of questions from every combination appears unnecessarily repetitive. In the instruments presented to subjects, each item was numbered sequentially, depending upon the position of the item within the set of 20 questions and the position of that set within the entire group of four or six sets. An item might therefore be number 36 for one subject and number 96 for another subject. In this Appendix the items of each set (corresponding to one of the 12 level members) have been numbered 1-20 for identification purposes (see Appendix E). This identifying number is in parentheses be— low the phrasing code for each item. 125 126 The code number at the beginning of each set of 20 questions indicates the level (Roman numeral) and member (Arabic numeral) which that set of questions represents: thus SAF:III2 indicates that the questions belong to the second member of level III. A corresponding code is given at the end of the set of 20 items, with the number "2" in parentheses to indicate the second of two pages comprising the specified items. In the actual instruments, all sets of 20 questions were followed by the directions "Go on to the next page" except the final set. The final set would, of course, be either level-member 1V2 if semantic path G were presented to the subject level by level, or level— member VIl if any other semantic path were presented to the subject level by level. When levels are randomized, the appropriate directions must be placed at the end of each set of 20 items. In addition to the sheets containing items from various level members, each subject was presented a cover— sheet of directions; the cover sheet included a specified "series number"——a number which identified the particular semantic path and the order (random or level—by-level) in which the level members of that path were assembled. As indicated in Chapter IV, various semantic paths were ran- domly distributed to subjects, so that such identification was necessary; such identification preserved the anononymity of each subject. 127 Finally, administrators were provided with a stan- dardized set of administration instructions. A copy of these instructions is included in this Appendix. The number preceding each item is the code which in— dicates both directionality (numeral before the decimal point: "1" indicates an item for which the response "agree" implies favorableness toward the attitude object and "disagree" implies unfavorableness); and grammatical form (numeral after the decimal point: "1" indicates an item in which the actor is grammatically emphasized). Similarly, "2" before the decimal point indicates an item for which "agree" implies unfavorableness toward the attitude object; "2" after the decimal point indicates an item in which the attitude object is grammatically emphasized. Appendix B indicates representative items from all level members; the items differ there in both directionality and grammatical form. The specific content, or "disjoint struction," of the items used in the research instrument was suggested by items from the original scale of Jordan and from the later work of Jordan and Hamersma (1969). As indicated in Chapter VI, specification and analysis of disjoint struction is presently in a very early stage. Items used in the present instrument, therefore, may not be representative of the universe of disjoint-struction item specificity; the items used, however, do represent the hypothesized universe of 12 level members presently identified by conjoint struction specifications. 128 SAF Scale: Form ED-l Information for Administrators Thank you for your cooperation. Your interruption of regular class routines is very much appreciated. The following information may be helpful to you and to your students. The supply of scales which has been provided is of varying types randomly arranged. Please do not rearrange the scales before distributing them to your students; the order of distribution-—e.g., left to right or back to front of your c1ass--does not make any difference, since the types of scales have been mixed together in random order. Please distribute the scales, one to a person; then the IBM answer sheets and scoring pencils. Then read the following comments to your students: The questionnaires which you are being asked to fill out have questions about your own beliefs and actions and about the beliefs and actions of other persons. Some of the questionnaires have 120 items and some have only 80 items. The first page of your questionnaire contains a sample item and an explanation of the different types of items which you are being asked to mark. Please be careful to do these four things: (1) Use the IBM answer sheet; do not mark on the questionnaire itself. (2) Do not put your name either on the question— naire or on the answer sheet. (3) At the bottom of the first page of your question- naire is a "series number" underlined which has an arrow pointing to it; put that number on the upper right-hand corner of your answer sheet. This number does not identify you; it identifies only which particular set of questions you answered. (A) Please answer every question: note that the second-choice answer for every item is "uncertain.“ 129 These items have been prepared in accordance with the theoretical formulations of Louis Guttman about at? titude-item construction—vin particular, the Guttman con— cept of "facet design." The authors of the present scale will provide both theoretical explanations and summaries of actual data from the scale to interested faculty and students upon request. Paul Maierle John E. Jordan College of Education 130 ‘ SAF SCale: Form ED-l Directions You yourself and other persons like you often behave in the same way toward persons who are emotionally disturbed. You also have some general ideas about yourself, about others like you, and about persons who are emotionally disturbed. Sometimes you behave the same way toward everyone, and some- times you may behave differently toward persons who are emo— tionally disturbed. The words "emotionally disturbed" are used in the statements in this scale. Where these words are used, they include those children or adults whose behaviors, feelings, or emotions cause them to have difficulties with every-day problems which they are unable to solve. This scale has statements about ideas and about be— havior. Each section of this scale is different from every other section, although some of the statements in each section are similar. Your answers in one section, therefore, may be the same as answers in another section, or your answers may differ from section to section. Try to mark each set of statements as accurately as possible. If you agree with a particular statement, mark the answer number 1, "agree." If you disagree with a par- ticular statement, mark the answer number 3, "disagree." If you are uncertain whether you agree or disagree, mark the answer number 2, "uncertain." Here is a sample statement: I think that most emotionally disturbed persons are taller than I am: (1) agree (2) uncertain (3) disagree If you do in fact think that most emotionally dis- turbed persons are taller than you are, you would mark the answer number 1, "agree": If you are using an IBM answer sheet, you would mark a heavy dark line on the answer sheet as follows: (1) H <2) III: (3) III: (A) III: (5) :22: If you are told to mark your answers on this booklet, you would indicate the same answer, number 1, "agree," by making a circle around the number 1 as follows: @ agree (2) uncertain (3) disagree 131 Do not put your name on this booklet or on your answer sheet. Your booklet is one of a series; please mark that series number on the upper right—hand corner of your answer sheet: the series number of this booklet is Z:_Mark this number on . By: Paul Maierle your answer sheet. John E. Jordan College of Education Michigan State University 132 SAF:Il This section contains statementS‘about ideas which you may share with other peopleIabBEtrpErEBEB'Who'are emo« tionally disturbed. If a statement describes the ideas of most people like yourself, mark the answer number I, "agree." If the statement does not describe your ideas, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 2.2 We think that most emotionally disturbed persons are (l) lazier than we are. (1) agree (2) uncertain (3) disagree 2.2 We think that most emotionally disturbed persons are (2) less intelligent than we are. (1) agree (2) uncertain (3) disagree 2.1 We think that we are more sexually attractive than (3) most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.2 We think that most emotionally disturbed persons are (A) cleaner than we are. (1) agree (2) uncertain (3) disagree 1.1 We think that we are less COOperative than most emo- (5) tionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.1 We think that we are less friendly than most emotion— (6) ally disturbed persons. (1) agree (2) uncertain (3) disagree 2.1 We think that we are more mature than most emotionally (7) disturbed persons. (1) agree (2) uncertain (3) disagree 1.2 We think that most emotionally disturbed persons are (8) more helpful than we are. (1) agree .(2) uncertain (3) disagree 2.2 We think that most emotionally disturbed persons are (9) more cruel than we are. (1) agree (2) uncertain (3) disagree 2.1 We think that we are more loyal than most emotionally (10) disturbed persons. (1) agree (2) uncertain (3) disagree 1.1 We think that we are less energetic than most emo- (ll) tionally disturbed persons. (1) agree (2) uncertain (3) disagree 133 We think that most emotionally disturbed persons are better looking than we are. (1) agree (2) uncertain (3) disagree We think that we are less prejudiced than most emo- tionally disturbed persons. (1) agree (2) uncertain (3) disagree We think that most emotionally disturbed persons are more careless than we are. (1) agree (2) uncertain (3) disagree We think that most emotionally disturbed persons are less understanding than we are. (1) agree (2) uncertain (3) disagree We think that we are less reliable than most emo- tionally disturbed persons. (1) agree (2) uncertain (3) disagree We think that most emotionally disturbed persons are happier than we are. _ (1) agree (2) uncertain (3) disagree We think that most emotionally disturbed persons are more honest than we are. (1) agree (2) uncertain (3) disagree We think that we are more talented than most emo- tionally disturbed persons. (1) agree (2) uncertain (3) disagree We think that we are less healthy than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:Il(2) GO ON TO THE NEXT PAGE. 13A SAF:IIl This section contains statements about ideas which you may have about yourself and aboutfothers who are not emotionally disturbed. If a statement describes how ou yourself think, mark the answer number 1, "agree." If the statement does not describe how you think, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 2.2 I think that most emotionally disturbed persons are (l) lazier than peOple like myself. (1) agree (2) uncertain (3) disagree 2.2 I think that most emotionally disturbed persons are (2) less intelligent than people like myself. (1) agree (2) uncertain (3) disagree 2.1 I think that people like myself are more sexually (3) attractive than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.2 I think that most emotionally disturbed persons are (A) Cleaner than people like myself. (1) agree (2) uncertain (3) disagree 1.1 I think that persons like myself are less cooperative (5) than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.1 I think that people like myself are less friendly (6) than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 2.1 I think that peOple like myself are more mature than (7) most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.2 I think that most emotionally disturbed persons are (8) more helpful than people like myself. (1) agree (2) uncertain (3) disagree 2.2 I think that most emotionally disturbed persons are (9) more cruel than peOple like myself. (1) agree (2) uncertain (3) disagree 2.1 I think that persons like myself are more loyal than (10) most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.1 I think that peOple like myself are less energetic (11) than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 135 I think that most emotionally disturbed persons are better looking than people like myself. (1) agree (2) uncertain (3) disagree I think that people like myself are less prejudiced than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree I think that most emotionally disturbed persons are more careless than people like myself. (1) agree (2) uncertain (3) disagree I think that most emotionally disturbed persons are less understanding than peOple like myself. (1) agree (2) uncertain (3) disagree I think that people like myself are less reliable than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree I think that most emotionally disturbed persons are happier than people like myself. (1) agree (2) uncertain (3) disagree I think that most emotionally disturbed persons are more honest than people like myself. (1) agree (2) uncertain (3) disagree I think that people like myself are more talented than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree I think that people like myself are less healthy than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:II1(2) GO ON TO THE NEXT PAGE. 136 SAF:II2 This section contains statements about ways in which you and others may think ypp behave toward persons who are emotionally disturbed. If a statement describes how you think you behave, mark the answer number 1, "agree." If the statement does not describe how you think you behave, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 1.2 (1) We think that some people who eat with us are emo- tionally disturbed. (1) agree (2) uncertain (3) disagree We think that some of the people from whom we accept help are emotionally disturbed. (1) agree (2) uncertain (3) disagree We do not think that persons for whom we work are emotionally disturbed. (1) agree (2) uncertain (3) disagree We do not think that we live near emotionally dis- turbed persons. (1) agree (2) uncertain (3) disagree We do not think that anyone who is emotionally dis— turbed is invited to our home. (1) agree (2) uncertain (3) disagree We do not think that anyone who is under our authority is emotionally disturbed. (1) agree (2) uncertain (3) disagree We do not think that we rent things from persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree We do not think that we date emotionally disturbed persons. (1) agree (2) uncertain (3) disagree We do not think that we avoid persons who are emo- tionally disturbed. (1) agree (2) uncertain (3) disagree We do not think that we borrow money from persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree We think that persons with whom we fight are emo- tionally disturbed. (1) agree (2) uncertain (3) disagree 137 We do not think that anyone to whom we lend things is emotionally disturbed. (1) agree (2) uncertain (3) disagree We think that some of us are married to emotionally disturbed persons. (1) agree (2) uncertain (3) disagree We think that we rent things to persons who are emoe tionally disturbed. (1) agree (2) uncertain (3) disagree We think that we borrow things from some persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree We think that some of our friends are emotionally dis- turbed. (1) agree (2) uncertain (3) disagree We think that some of the people we help are emotionally disturbed. (1) agree (2) uncertain (3) disagree We think that we loan money to some persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree We think that some of the persons we enjoy sexually are emotionally disturbed. (1) agree (2) uncertain (3) disagree We do not think that we work with emotionally dis— turbed persons. (1) agree (2) uncertain (3) disagree SAF:II2(2) GO ON TO THE NEXT PAGE. 138 SAF:II3 This section contains statements about ideas which other people who are not emotionally disturbed may have about 19g. If a statement describes whatppeople like yourself think about you, mark the answer number 1,I"agree." If the statement does not describe what other people think about you, mark the answer number 3, "disagree." If you are un- certain about the statement, mark the answer number 2, "uncertain." 2.2 Other people think that most emotionally disturbed (1) persons are lazier than I am. (1) agree (2) uncertain (3) disagree 2.2 Other people think that most emotionally disturbed (2) persons are less intelligent than I am. (1) agree (2) uncertain (3) disagree 2.1 Other peOple think that I am more sexually attractive (3) than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.2 Other people think that most emotionally disturbed (A) persons are cleaner than I am. (1) agree (2) uncertain (3) disagree 1.1 Other people think that I am less cooperative than (5) most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.1 Other people think that I am less friendly than most (6) emotionally disturbed persons. ' (1) agree (2) uncertain (3) disagree 2.1 Other people think that I am more mature than most (7) emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.2 Other people think that most emotionally disturbed (8) persons are more helpful than I am. (1) agree (2) uncertain (3) disagree 2.2 Other people think that most emotionally disturbed (9) persons are more cruel than I am. (1) agree (2) uncertain (3) disagree 2.1 Other people think that I am more loyal than most emo- (10) tionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.1 Other people think that I am less energetic than most (11) emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 139 Other people think that most emotionally disturbed persons are better looking than I am. (1) agree (2) uncertain (3) disagree Other people think that I am less prejudiced than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree Other people think that most emotionally disturbed persons are more careless than I am. (1) agree (2) uncertain (3) disagree Other people think that most emotionally disturbed persons are less understanding than I am. (1) agree (2) uncertain (3) disagree Other people think that I am less reliable than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree Other people think that most emotionally disturbed persons are happier than I am. (1) agree (2) uncertain (3) disagree Other people think that most emotionally disturbed persons are more honest than I am. (1) agree (2) uncertain (3) disagree Other people think that I am more talented than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree Other people think that I am less healthy than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:II3(2) GO ON TO THE NEXT PAGE. 1A0 SAF:IIIl This section contains statements about ideas you may have about how you and others should act toward emotionally disturbed'pérsbns. If a statement describes how you think you and others should act, mark the answer number 1, "agree." If the statement does not describe how you think you and others should act, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 1.2 Persons who are emotionally disturbed should be al— (1) lowed to eat with us. (1) agree (2) uncertain (3) disagree 1.2 Persons who are emotionally disturbed should be asked (2) for help (1) agree (2) uncertain (3) disagree 2.2 Persons who are emotionally disturbed should not be (3) our bosses. (1) agree (2) uncertain (3) disagree 2.1 We should not live near persons who are emotionally (A) disturbed. (1) agree (2) uncertain (3) disagree 2.2 Persons who are emotionally disturbed should not be (5) invited to our home. (I) agree (2) uncertain (3) disagree 2.2 Persons who are emotionally disturbed should not be (6) under our authority. (1) agree (2) uncertain (3) disagree 2.1 We should not rent things from persons who are emo- (7) tionally disturbed. (1) agree (2) uncertain (3) disagree 2.1 We should not date persons who are emotionally dis- (8) turbed. (1) agree (2) uncertain (3) disagree 1.1 We should not avoid persons who are emotionally dis- (9) turbed. (1) agree (2) uncertain (3) disagree 2.1 We should not borrow money from persons who are (10) emotionally disturbed. (1) agree (2) uncertain (3) disagree 2.2 Persons who are emotionally disturbed should be fought. (ll) (1) agree (2) uncertain (3) disagree 1A1 Persons who are emotionally disturbed should not be loaned things. (1) agree (2) uncertain (3) disagree We should marry persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree We should rent things to persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree We should borrow things from persons who are emo- tionally disturbed. (1) agree (2) uncertain (3) disagree Persons who are emotionally disturbed should be accepted as our friends. (1) agree (2) uncertain (3) disagree Persons who are emotionally disturbed should be helped by us. (1) agree (2) uncertain (3) disagree We should loan money to persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Persons who are emotionally disturbed should be en- joyed sexually. (1) agree (2) uncertain (3) disagree We should not work with emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:III1(2) GO ON TO THE NEXT PAGE. 1A2 SAF:III2 This section contains statements about ideas you may have about yourself. If a statement‘describes your opinion ofpyourself, mark the answer number 1, "agree.' If the statement does not describe your opinion, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 2.2 I think that most emotionally disturbed persons are (l) lazier than I am. (1) agree (2) uncertain (3) disagree 2.2 I think that most emotionally disturbed persons are (2) less intelligent than I am. (1) agree (2) uncertain (3) disagree 2.1 I think that I am more sexually attractive than most (3) emotionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.2 I think that most emotionally disturbed persons are (A) cleaner than I am. (1) agree (2) uncertain (3) disagree 1.1 I think that I am less cooperative than most emo- (5) tionally disturbed persons. (1) agree (2) uncertain (3) disagree 1.1 I think that I am less friendly than most emotionally (6) disturbed persons. (1) agree (2) uncertain (3) disagree 2.1 I think that I am more mature than most emotionally (7) disturbed persons. (1) agree (2) uncertain (3) disagree 1.2 I think that most emotionally disturbed persons are (8) more helpful than I am. (1) agree (2) uncertain (3) disagree 2.2 I think that most emotionally disturbed persons are (9) are cruel than I am. (1) agree (2) uncertain (3) disagree 2.1 I think that I am more loyal than most emotionally (10) disturbed persons. (1) agree (2) uncertain (3) disagree 1.1 I think that I am less energetic than most emotionally (ll) disturbed persons. (1) agree (2) uncertain (3) disagree 1A3 I think that most emotionally disturbed persons are better looking than I am. (1) agree (2) uncertain (3) disagree I think that I am less prejudiced than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree I think that most emotionally disturbed persons are more careless than I am. (1) agree (2) uncertain (3) disagree I think that most emotionally disturbed persons are less understanding than I am. (1) agree (2) uncertain (3) disagree I think that I am less reliable than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree I think that most emotionally disturbed persons are happier than I am. (1) agree (2) uncertain (3) disagree I think that most emotionally disturbed persons are more honest than I am. . (1) agree (2) uncertain (3) disagree I think that I am more talented than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree I think that I am less healthy than most emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:II12(2) GO ON TO THE NEXT PAGE. 1AA SAF:III3 This section contains statements about ideas which people may have about how you will act toward emo- tionally disturbed persons. If a statement describes hpw_ others expect you to act toward emotionally disturbed persons, mark the answer number 1, "agree." If the state— ment does not describe how others expect you to act, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." (1) Other people expect that some of the persons with whom I eat will be emotionally disturbed. (1) agree (2) uncertain (3) disagree Other people expect that persons who are emotionally disturbed will be asked for help by me. (1) agree (2) uncertain (3) disagree Other people expect that persons for whom I work will not be emotionally disturbed. (1) agree (2) uncertain (3) disagree Other people expect me not to live near emotionally disturbed persons. (1) agree (2) uncertain (3) disagree Other people expect that no one who is emotionally disturbed will be invited to my home. (1) agree (2) uncertain (3) disagree Other people expect that no one who is emotionally disturbed will be under my authority. (1) agree (2) uncertain (3) disagree Other people expect me not to rent things from persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Other people expect me not to date persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Other people expect me not to avoid persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Other people expect me not to borrow money from persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Other people expect that persons who are emotionally disturbed will be fought by me. (1) agree (2) uncertain (3) disagree 1A5 Other people expect that no one who is emotionally disturbed will be loaned things by me. (1) agree (2) uncertain (3) disagree Other people expect me to marry someone who is emoe tionally disturbed. (1) agree (2) Other people expect uncertain (3) disagree me to rent things to persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Other people expect me to borrow things from persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Other people expect that persons who are emotionally disturbed will be accepted as my friends. (1) agree (2) uncertain (3) disagree Other peOple expect that persons who are emotionally disturbed will be helped by me. (1) agree (2) uncertain (3) disagree Other people expect me to loan money to persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Other people eXpect that persons who are emotionally disturbed will be enjoyed sexually by me. (1) agree (2) uncertain (3) disagree Other people expect me not to work with emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:III3C2) GO ON TO THE NEXT PAGE. 1A6 SAF:IIIA This section contains statements about feelings which you and people like yourself may have about'éthionally disturbed persons. If a statement describes your feelings about emotionally disturbed persons, mark the answer number 1, "agree." If the statement does not describe your feelings, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 2.2 Emotionally disturbed persons displease us. (1) (1) agree (2) uncertain (3) disagree 2.2 Emotionally disturbed persons make us feel negative. (2) (1) agree (2) uncertain (3) disagree 2.1 We feel dislike for emotionally disturbed persons. (3) (1) agree (2) uncertain (3) disagree 1.2 Emotionally disturbed persons do not make us feel (A) tense. (1) agree (2) uncertain (3) disagree 1.1 We are not frightened by emotionally disturbed persons. (5) (1) agree (2) uncertain (3) disagree 1.2 Emotionally disturbed persons do not anger us. (6) (1) agree (2) uncertain (3) disagree 1.1 We do not feel contempt for emotionally disturbed (7) persons. (1) agree (2) uncertain (3) disagree 1.1 We do not feel uncomfortable near emotionally dis- (8) turbed persons. (1) agree (2) uncertain (3) disagree 1.2 Emotionally disturbed persons do not make us feel (9) dismay. (1) agree (2) uncertain (3) disagree 2.1 We feel loathing for emotionally disturbed persons. (10) (1) agree (2) uncertain (3) disagree 1.1 We feel attracted to emotionally disturbed persons. (11) (1) agree (2) uncertain (3) disagree 2.2 Emotionally disturbed persons make us feel unsym— (12) pathetic. (1) agree (2) uncertain (3) disagree 2.1 We do not feel at ease with emotionally disturbed (13) persons. (1) agree (2) uncertain (3) disagree 1A7 Emotionally disturbed persons make us feel unhappy. (1) agree (2) uncertain (3) disagree Emotionally disturbed persons make us feel confusion. (1) agree (2) uncertain (3) disagree Emotionally disturbed persons do not repulse us. (1) agree (2) uncertain (3) disagree We feel secure near emotionally disturbed persons. (1) agree (2) uncertain (3) disagree Emotionally disturbed persons attract us sexually. (1) agree (2) uncertain (3) disagree We feel restless near emotionally disturbed persons. (1) agree (2) uncertain (3) disagree We are not calm near emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:IIIAC2) GO ON TO THE NEXT PAGE. 1A8 SAF:IVl This section contains statements about how you think you would act toward emotionally disturbed‘persons. If a statment describes how you think you would act, mark the answer number 1, "agree." If the statement does not describe how you would act, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 1.2 Persons who are emotionally disturbed would be al- (1) lowed to eat with me. (1) agree (2) uncertain (3) disagree 1.2 Persons who are emotionally disturbed would be asked (2) for help by me. (1) agree (2) uncertain (3) disagree 2.2 Persons who are emotionally disturbed would not be (3) accepted as my bosses. (1) agree (2) uncertain (3) disagree 2.1 I would not live near emotionally disturbed persons. (A) (1) agree (2) uncertain (3) disagree 2.2 Persons who are emotionally disturbed would not be (5) invited to my home. (1) agree (2) uncertain (3) disagree 2.2 Persons who are emotionally disturbed would not be (6) accepted under my authority. (1) agree (2) uncertain (3) disagree 2.1 I would not rent things from persons who are emo— (7) tionally disturbed. (1) agree (2) uncertain (3) disagree 2.1 I would not date persons who are emotionally disturbed. (8) (1) agree (2) uncertain (3) disagree 1.1 I would not avoid persons who are emotionally dis— (9) turbed. (1) agree (2) uncertain (3) disagree .1 I would not borrow money from persons who are emo- (10) tionally disturbed. (1) agree (2) uncertain (3) disagree 2.2 Persons who are emotionally disturbed would be fought (11) by me. (1) agree (2) uncertain (3) disagree 2.2 Persons who are emotionally disturbed would not be (12) loaned things by me. (1) agree . (2) uncertain (3) disagree 1A9 I would marry someone who is emotionally disturbed. (1) agree (2) uncertain (3) disagree I would rent things to persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree I would borrow things from persons who are emo- tionally disturbed. (1) agree (2) uncertain (3) disagree Persons who are emotionally disturbed would be ac— cepted as my friends. (1) agree (2) uncertain (3) disagree Persons who are emotionally disturbed would be helped by me. (1) agree (2) uncertain (3) disagree I would loan money to persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Persons who are emotionally disturbed would be enjoyed sexually by me. (1) agree (2) uncertain (3) disagree I would not work with emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:IV1(2) GO ON TO THE NEXT PAGE. 150 SAF:IV2 This section contains statements about how persons like yourself actuallylbehave toward persons who are emo- tionally disturbed. If a statement describes how you and persons like yourself actually behave, mark the answer number 1, "agree." If the statement does not describe how you and others behave, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 1.2 Some of the peOple who eat with us are emotionally (l) disturbed. (1) agree (2) uncertain (3) disagree 1.2 Some of the people from whom we accept help are emo— (2) tionally disturbed. (1) agree (2) uncertain (3) disagree 2.2 No one for whom we work is emotionally disturbed. (3) (1) agree (2) uncertain (3) disagree 2.1 We do not live near emotionally disturbed persons. (A) (1) agree (2) uncertain (3) disagree 2.2 No one who is emotionally disturbed is invited to (5) our home. (1) agree (2) uncertain (3) disagree 2.2 No one under our authority is emotionally disturbed. ) (1) agree (2) uncertain (3) disagree 1 We do not rent things from persons who are emotionally ) disturbed. (1) agree (2) uncertain (3) disagree 2.1 We do not date emotionally disturbed persons. (8) (1) agree (2) uncertain (3) disagree 1.1 We do not avoid persons who are emotionally disturbed. (9) (1) agree (2) uncertain (3) disagree 2.1 We do not borrow money from persons who are emotionally (10) disturbed. (1) agree (2) uncertain (3) disagree 2.2 Persons with whom we fight are emotionally disturbed. (ll) (1) agree (2) uncertain (3) disagree 2.2 No one to whom we lend things is emotionally dis- (12) turbed. (1) agree (2) uncertain (3) disagree 151 Some of us are married to emotionally disturbed persons. (1) agree (2) uncertain (3) disagree We rent things to some persons who are emotionally dis- turbed. (1) agree (2) uncertain (3) disagree We borrow things from persons who are emotionally dis- turbed. (1) agree (2) Some of our friends (1) agree (2) Some of the persons (1) agree (2) We loan money to persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree uncertain (3) disagree are emotionally disturbed. uncertain (3) disagree we help are emotionally disturbed. uncertain (3) disagree Some of the persons we enjoy sexually are emotionally disturbed. (1) agree (2) uncertain (3) disagree We do not work with emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:IV2 THANK YOU FOR YOUR COOPERATION. Please be sure that you have answered every question and that you have marked the series number of this booklet on the upper right—hand corner of your answer sheet: the series number is indicated on the bottom of the first page of this booklet. 152 SAF:V1 This section contains statements about feelings which you yourself may have abSOt'empiionally dieturbed persons. If a statement describes a feeling which you haveppersonally, mark the answer number 1, agree." If the statement does not describe a feeling which you have, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 2.2 Emotionally disturbed persons displease me. (1) (1) agree (2) uncertain (3) disagree 2.2 Emotionally disturbed persons make me feel negative. (2) (1) agree (2) uncertain (3) disagree 2.1 I feel dislike for emotionally disturbed persons. (3) (1) agree (2) uncertain (3) disagree 1.2 Emotionally disturbed persons do not make me feel (A) tense. (1) agree (2) uncertain (3) disagree 1.1 I am not frightened by emotionally disturbed persons. (5) (1) agree (2) uncertain (3) disagree 1.2 Emotionally disturbed persons do not anger me. (6) (1) agree (2) uncertain (3) disagree 1.1 I do not feel contempt for emotionally disturbed (7) persons. (1) agree (2) uncertain (3) disagree 2.1 I do not feel comfortable near emotionally disturbed (8) persons. (1) agree (2) uncertain (3) disagree 1.2 Emotionally disturbed persons do not make me feel (9) dismay. (1) agree (2) uncertain (3) disagree 2.1 I feel loathing for emotionally disturbed persons. (10) (1) agree (2) uncertain (3) disagree 1.1 I feel attracted to emotionally disturbed persons. (11) (1) agree (2) uncertain (3) disagree 2.2 Emotionally disturbed persons make me feel unsym— (12) pathetic. (1) agree (2) uncertain (3) disagree 2.1 I do not feel at ease with emotionally disturbed (13) persons. (1) agree (2) uncertain (3) disagree (i9) 2.1 (20) 153 Emotionally disturbed persons make me feel unhappy. (l) agree (2) uncertain (3) disagree Emotionally disturbed persons make me feel confusion. (1) agree (2) uncertain (3) disagree Emotionally disturbed persons do not repulse me. (1) agree (2) uncertain (3) disagree I feel secure near emotionally disturbed persons. (1) agree (2) uncertain (3) disagree Emotionally disturbed persons attract me sexually. (1) agree (2) uncertain (3) disagree I feel restless near emotionally disturbed persons. (1) agree (2) uncertain (3) disagree I am not calm near emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:V1(2) GO ON TO THE NEXT PAGE. 15A SAF:VIl This section contains statements about your own behavior toward persons who are emotionally disturbed. If a statement describes an action of yours, mark the answer number 1, "agree." If the statement does not describe your actions, mark the answer number 3, "disagree." If you are uncertain about the statement, mark the answer number 2, "uncertain." 1.2 (1) AH V|\_) Some of the peOple who eat with me are emotionally disturbed. (1) agree (2) uncertain (3) disagree Some of the persons from whom I accept help are emotionally disturbed. (1) agree (2) uncertain (3) disagree No one for whom I work is emotionally disturbed. (1) agree (2) uncertain (3) disagree I do not live near emotionally disturbed persons. (1) agree (2) uncertain (3) disagree No one who is emotionally disturbed is invited to my home. (1) agree (2) uncertain (3) disagree No one under my authority is emotionally disturbed. (1) agree (2) uncertain (3) disagree I do not rent things from persons who are emotion— ally disturbed. (1) agree (2) uncertain (3) disagree I do not date emotionally disturbed persons. (1) agree (2) uncertain (3) disagree I do not avoid emotionally disturbed persons. (1) agree (2) uncertain (3) disagree I do not borrow money from persons who are emotion- ally disturbed. (1) agree (2) uncertain (3) disagree Persons with whom I fight are emotionally disturbed. (1) agree (2) uncertain (3) disagree No one to whom I lend things is emotionally disturbed. (1) agree (2) uncertain (3) disagree I am married to an emotionally disturbed person. (1) agree (2) uncertain (3) disagree 1.1 (1A) 1.1 (15) 1.2 (16) 1.2 (17) 1.1 (18) 1.2 (19) (£0) 155 I rent things to persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree I borrow things from persons who are emotionally dis- turbed. (1) agree (2) uncertain (3) disagree Some of my friends are emotionally disturbed. (1) agree (2) uncertain (3) disagree Some of the people I help are emotionally disturbed. (1) agree (2) uncertain (3) disagree I loan money to persons who are emotionally disturbed. (1) agree (2) uncertain (3) disagree Some of the persons I enjoy sexually are emotionally disturbed. (1) agree (2) uncertain (3) disagree I do not work with emotionally disturbed persons. (1) agree (2) uncertain (3) disagree SAF:VIl(2) THANK YOU FOR YOUR COOPERATION. Please be sure that you have answered every question and that you have marked the series number of this booklet on the upper right—hand corner of your answer sheet: the series number is indicated on the bottom of the first page of this booklet. APPENDIX D DATA CODING AND RE—SCORING DATA CODING AND RE—SCORING As indicated in Chapters IV and V and in Appendices B, C, and E, original responses from subjects were to items randomly distributed in directionality and grammatical form, while the correlational analyses were performed on data so arranged that the larger the numerical score on any item (or total from a set of items, or total across all sets), the more positive the attitude of the subject. A procedure was needed to change some of the response data so that each item for every subject could be scored in exactly the same way-~i.e., so that "1" across all subjects and items would indicate the least favorable response. In addition to the original response of each subject to each item, entered on IBM answer sheets, each item could be identified by its position within a set of twenty, by the set (corresponding to level member) to which it belonged, by its general content (see Chapter IV), by the semantic path to which it belonged, and, of particular importance for appropriate coding, by its directionality and grammatical form. A complete code book follows in Appendix E; Appendices B and C indicate the specific coding adopted for direction— ality and grammatical form. 157 158 The purpose of this Appendix is to indicate, by pre- sentation of instructions written for the original scoring procedures (punching of cards directly from answer sheets) and for the coding and re—scoring procedures (addition of directionality and grammatical form data and re-scoring of appropriate items), how sets of 20-item responses were re- scored so that scores across all items for all subjects were analogous--i.e., so that "1" (indicating either "agree" or "disagree" in the original response, depending on the directionality of the item) could be universally interpreted as the least favorable response. The data were first transferred directly from the answer sheets, with additional subject—identifying data, and then re—scored. Separation of these operations was preferable to a one-step operation in which data would be transferred from answer sheets and re-scored simultaneously for two reasons: (a) the intermediate step provided an ad- ditional Opportunity for checking on the accuracy of data- recording procedures; (b) to enter item-coding data and to rescore response data would have required several "runs," or duplication of procedures in transferring of data from answer sheet to card, whereas the method adopted required only one "run" per card at the transfer point and one addi- tional "run" per card at the re-scoring point. The instructions presented below are therefore in two parts: (a) scoring instructions for transfer of data from 159 answer sheets to IBM cards; and (b) coding and re-scoring instructions. The latter are given here in their entirety, while only the instructions for one set of answer sheets are given. Answer sheets were sorted, prior to card-punching, into the 1A sets of data collected (7 semantic paths; each path administered once in random order and once in level— by-level order); the instructions for each of the other 13 sets of answer sheets paralleled those for the set given here as an example. The instructions presented below were intended for data—processing personnel. Interpretation of an entry in any column requires use of the code book (Appendix E). SCORING INSTRUCTIONS: SAF data, set 1 (n=69) General: 69 answer sheets; data on each sheet to be punched in 6 parts on 6 cards, one part to a card, as follows: Card Columnsl Data 1 l-A A-digit ID number on answer sheet (in area marked "student number") 5-8 1612_(punch this A-digit number) 10-29 Items 1-20 on answer sheet 30—end leave cpen 2 l-A ID 5-8 2513 10-29 Items 2l-A0 1Column 9 automatically used by scoring office to indicate number of cards to be punched to given specifi- cations. 160 Card Columns Data 3 l-A ID 5-8 311i 10—29 Items Al—60 A l-A ID 5-8 A;l§_ 10-29 Items 61-80 5 l—A ID 5-8 5211 10-29 Items 81-1001 6 1—A ID 5—8 6Al2 10-29 Items 101—120 1 Sets 7 and 1A (semantic path F randomly and level- by—level administered) have only 80 items—-hence, only A cards. 161 SAF Data-transfer Instructions1 1. Number of cards/subject a. sets 1—6, 8-13: 6 cards b. sets 7, 1A: A cards 2. Card number: col. 5 on all cards, all sets 3. Subject number: cols. l—A, all cards, all sets A. Set (1A) a. each set (n=50+) run separately b. within each set, data transferred by one of A variant instructions (specific variant to be used with each card of each set is indicated on master sheet accompanying that set; see following pages for A variants) 5. General instructions across all cards, sets, variants a. cols. 1-9 transferred as is b. cols. 12, 15, 18 ... 69 (1) add as indicated in variant instructions (2) only "1" or "2" legitimate c. cols. 10, l3, l6 ... 67 (1) add as indicated in variant instructions (2) only "1" or "2" legitimate (3) cols. contain coding information for im- mediately following cols.: "l"--following data reflected: "2"--following data OK: transfer as is d. cols. 11, 1A, 17 ... 68: item responses (1) only "1," "2," "3," or "0" legitimate on original (2) "A" on original = "3" on original (3) "5’" "6’" n7," "8’" and "9" on original = on original 6. Note that an item response of "A" on original may become "3" or "1" on new card, depending on preceding code (see 5, c, (3) above). fi 1The author is indebted to Mr. Marcus Vale, of the M. S. U. Computer Center, for the computer program which performed the operations described above. 162 SAF Data—transfer Instructions, Variant 1: all cards with "l" in col. 8 of source card. Source Source 001 Col ADD 001 Col ADD 1 1 A0 1 ,3 2 2 11 A1 20 3;; 3 3 A2 1 5cm A A A3 1 go 5 5 12 AA 21 a? 6 6 A5 2 arc 7 7 A6 2 :32. 8 8 13 A7 22 " 9 9 A8 1 10 2 A9 2 1 11 10 1A 50 23 12 2 51 2 13 2 52 2 2 1A 11 15 53 2A 15 2 5A 2 16 2 55 1 3 17 12 16 56 25 18 1 57 1 l9 1 58 l A 20 13 17 59 26 21 2 6o 2 22 1 61 1 5 23 1A 18 62 27 2A 1 63 2 25 1 6A 2 6 26 15 19 65 28 27 1 66 1 28 2 67 1 7 29 16 2o 68 29 3o 1 69 1 31 1 _,_d 8 32 17 33 2 3A 2 9 35 18 36 2 37 2 10 38 19 39 l 163 SAF Dataetransfer Instructions, Variant 2: "2" in col. 8 of source card. Source C01 C01 ADD 1 1 g: 2 2 H 3 3 Egg A A 5 5 5 s 2 '38 8 8 SE 9 9 10 1 1 11 10 12 2 13 1 2 1A 11 15 2 16 2 3 l7 12 18 2 l9 2 A 20 13 21 1 22 2 5 23 1A 2A 2 25 2 6 26 15 27 2 28 2 7 29 16 3O 1 31 2 8 32 17 33 1 3A 1 9 35 [ 18 36 1 37 2 10 38 19 39 1 all cards with 11 12 13 1A l5 l6 l7 l8 19 20 Source Col ADD 2O 21 22 23 2A 25 26 27 28 29 [UN FJH 54m FJN HHH HFJ HFJ 14nd [\JI'U 16A SAF Data—transfer Instructions, Variant 3: all cards with "2" in col. 8 of source card except cards from sets 7yflA Source Source 001 Col ADD 001 Col ADD 1 1 A0 1 g: 2 2 11 A1 20 28 3 E if 1 3 2 g: 5 5 12 AA 21 a: 9 3 32 3 £8 8 8 13 A7 22 2,5 9 9 A8 1 1o 2 A9 2 1 11 10 1A 50 23 12 2 51 2 13 2 52 2 2 1A 11 15 53 2A 15 2 5A 2 16 2 55 1 3 17 12 16 56 25 18 . 1 57 2 19 l 58 1 A 20 13 17 59 26 21 2 6o 1 22 1 61 1 5 23 1A 18 62 27 2A 1 63 2 25 1 6A/ 2 6 26 15 19 65 28 27 2 66 1 28 1 67 2 7 29 16 20 68 29 3o 1 69 . 1 31 8 32 17 33 1 3A 1 9 35 18 36 2 37 2 1o 38 19 ‘ 39 1 165 SAF Data-transfer Instructions, Variant A: all cards from sets 7 and 1A with a "3" in col. 8 of source card. Source Source 001 Col ADD 001 Col ADD 1 1 A0 1 ,3 2 2 11 31 2o 9" 3 3 2 l '353 A A A3 2 36 5 5 12 AA 21 8g 6 6 A5 2 010 7 7 A6 2 :32 8 8 13 A7 22 r’ 9 9 A8 1 10 2 A9 2 1 11 10 1A 50 23 12 2 51 2 13 2 52 2 2 1A 11 15 53 2A 15 2 5A 2 16 2 55 1 3 17 12 16 S6 25 18 1 57 2 19 1 58 1 A 20 13 17 59 26 21 2 60 1 22 1 61 1 5 23 1A 18 62 27 2A 1 63 2 25 6A 2 6 26 15 19 65 28 27 2 66 1 28 1 67 2 7 29 16 20 68 29 3o 1 69 1 31 1 8 32 17 33 1 3A 1 9 35 18 36 2 37 2 10 38 19 39 l 166 Note: although the general instructions and specific notes at the top of each variant—instuction listing indicated which variant was to be used with each card of each set, a listing like the following was also provided with each set of cards: SAF Data—transfer Instructions, Set 1 General: 69 subjects, 6 cards per subjects; see ac— companying sheets for variants indicated below: Card Variant to be Used 1 2 2 3 l A 2 5 1 O\ [\J APPENDIX E CODE BOOK CODE BOOK "IntrOduCtion As noted in Appendix C, many of the 12 sets of 20 items, corresponding to the 12 hypothesized level members, appear in more than one semantic path. To duplicate the coding for each repetition of the same items appears un— necessary. The appropriate interpretations for each 20- item set are therefore given below only once; as in Appendix 0, items are numbered within each 20—item set for identification purposes. The interpretative material given below is applicable only to the card decks now available for the research ad— ministration of the SAF instrument; as indicated in Appendix D, data on that set of cards have been recoded from original responses. This code book, therefore, applies to the card decks available for analysis; it does not apply directly to the original answer seets or to intermediate decks of cards used in recoding procedures. The author is available to assist in defining re-coding procedures for paralled instruments using other item or order—of—level variations. 1.4 m ’I) 169 Coding A "O" in any column, any card, any set, any subject indicates "missing data"-—i.e., either the item is not applicable or the subject did not respond. The following coding interpretations are given in two parts: (a) col- umns 1-9, applicable to all cards, all sets, all subjects; and (b) columns 10-69, applicable to the specified sets of cards for all cards and subjects within that set. An explanation of the brief coding information for some columns may be helpful: Col. Cols. Cols. Cols. 8: "Content SAF"--see Chapter A; "Scoring variant"--see Appendix D 10, 13, 16 . . . 67 (see Appendix D): "l"--response data in following column reflected from original response: 1=3 2=2 3=l "2"--response data in following column as in original 11, 1A, 17 . . . 68: response data If preceded by "1," then a response code of "1," indicating "agree" on the answer sheet, is here scored as "3" (see level member I1, item A): the largest numerical value is therefore here scored for the response most favorable to the attitude object. If preceded by "2," then a response code of "1," indicating "agree" on the answer sheet, is here scored as "1" (see level member I1, item 1): the smallest numerical value is therefore here scored for the response least favorable to the attitude object. 12, 15, 18 . . . 69: see Appendices B and C 170 Code: All cards, subjects, SETS in columns 1-9 Col. Scale/Item Content Code 1 SET IDl Semantic path path path path path path path path Q’IJB'JUOWID 2 SET ID Order of path ad- ministration random level-by-level NH NOW-C'UUMH 3—A Subject ID Subject number with- in SET 0 }_I I 99 first set2 second set third set fourth set fifth set sixth set 5 Level ID and Position in sequence Card # of sets adminis- tered mmthH level level level level level level 6 Level ID Semantic position3 (Guttman level) OUT-EWNH O\U'IJ:'UU|\)|—‘ lSET comprises a group of six or four 20—item sets; SETS are: 11, 21, 31, A1, 51, 61, 71, 12, 22, 32, A2, 52, 62, 72, and correspond to a random or level-by-level administration of one semantic path. 2Set here indicates any 20—item group of questions, cor- responding to a level member, within any 6—set SET or A—set SET. 3When a semantic path is administered level-by-level, semantic position of level and order of administration are identical. 171 Col. Scale/Item Content Code 7 Member ID Semantic position within level member member member member .1:me DWNH 8 Content Content and scoring 8, variant 1 variant A, variant 2 F; SETS 11-61, 12-62: vari- ant 3; SETS 71, 72: vari- ant A [UH LA) 9 Reproduction control column--no significance for analysis. 172 Code: Level member Il (all cards, all SETS with "l" in col. 6 and "l" in col. 7) Col. Item Content Code 10 l Directionality 2 11 1 Response: edp lazier 1 agree 2 uncertain 3 disagree l2 1 Grammatical form 2 l3 2 D 2 1A 2 R: edp less intelligent 1 agree 2 uncertain 3 disagree 15 2 G 2 16 3 D 2 l7 3 R: we more sexually 1 agree attractive 2 uncertain 3 disagree 18 3 G 1 19 A D l 20 A R: edp cleaner 1 disagree 2 uncertain 3 agree 21 A G 2 22 5 D l 23 5 R: we less cooperative 1 disagree 2 uncertain 3 agree 2A 5 G l 25 6 D l 26 6 R: we less friendly 1 disagree 2 uncertain 3 agree 27 6 G l 28 7 D 2 29 7 R: we more mature 1 agree 2 uncertain 3 disagree 3O 7 G l 173 Col. Item Content Code 31 8 l 32 8 edp more helpful 1 disagree 2 uncertain 3 agree 33 8 2 3A 9 2 35 9 edp more cruel 1 agree 2 uncertain 3 disagree 36 9 2 37 10 2 38 10 we more loyal 1 agree 2 uncertain 3 disagree 39 10 1 A0 11 1 Al 11 we less energetic 1 disagree 2 uncertain 3 agree A2 11 1 A3 12 1 AA 12 edp better looking 1 disagree 2 uncertain 3 agree A5 12 2 A6 13 2 A7 13 we less prejudiced 1 agree 2 uncertain 3 disagree A8 13 1 A9 1A 2 50 1A edp more careless 1 agree 2 uncertain 3 disagree 51 1A 2 52 15 2 53 15 edp less understanding 1 agree 2 uncertain 3 disagree 5A 15 2 17A Col Item Content Code 55 16 l 56 16 we less reliable 1 disagree 2 uncertain 3 agree 57 16 l 58 17 l 59 17 edp happier 1 disagree 2 uncertain 3 agree 60 17 2 61 18 l 62 18 edp more honest 1 disagree 2 uncertain 3 agree 63 18 2 6A 19 2 65 19 we more talented 1 agree 2 uncertain 3 disagree 66 19 1 67 20 l 68 20 we less healthy 1 disagree 2 uncertain 3 agree 69‘ 20 1 175 Code: Level member IIl (all cards, all SETS with "2" in col. 6 and "l" in col. 7) Col. Item Content Code 10 1 Directionality 2 11 1 Response: edp lazier 1 agree 2 uncertain 3 disagree 12 l Grammatical form 2 l3 2 D 2 1A 2 R: edp less intelligent 1 agree 2 uncertain 3 disagree 15 2 G 2 16 3 D 2 l7 3 R: I more sexually 1 agree attractive 2 uncertain 3 disagree l8 3 G 1 19 A D l 20 A R: edp cleaner 1 disagree 2 uncertain 3 agree 21 A G 2 22 5 D l 23 5 R: I less cooperative 1 disagree 2 uncertain 3 agree 2A 5 G l 25 6 D l 26 6 R: I less friendly 1 disagree 2 uncertain 3 agree 27 6 G l 28 7 D 2 29 7 R: I more mature 1 agree 2 uncertain 3 disagree 30 7 G l 176 Col. Item Content Code 31 8 l 32 8 edp more helpful 1 disagree 2 uncertain 3 agree 33 8 2 3A 9 2 35 9 edp more cruel 1 agree 2 uncertain 3 disagree 36 9 2 37 10 2 38 10 I more loyal 1 agree 2 uncertain 3 disagree 39 10 1 A0 11 1 Al 11 I less energetic 1 disagree 2 uncertain 3 agree A2 11 1 A3 12 1 AA 12 edp better looking 1 disagree 2 uncertain 3 agree A5 12 2 A6 13 2 A7 13 I less prejudiced 1 agree 2 uncertain 3 disagree A8 13 1 A9 1A 2 50 1A edp more careless 1 agree 2 uncertain 3 disagree 51 1A 2 177 Col. Item Content Code 52 15 2 53 15 edp less understanding 1 agree 2 uncertain 3 disagree 5A 15 2 55 16 l 56 16 I less reliable I disagree 2 uncertain 3 agree 57 16 l 58 17 l 59 17 edp happier 1 disagree 2 uncertain 3 agree 6O 17 2 61 18 l 62 18 edp more honest 1 disagree 2 uncertain 3 agree 63 18 2 6A 19 2 65 19 I more talented 1 agree 2 uncertain 3 disagree 66 19 l 67 20 1 68 20 I less healthy 1 disagree 2 uncertain 3 agree 69 20 l 178 Code: Level member II2 (all cards, all SETS with "2" in col. 6 and "2" in col. 7) Col. Item Content Code 10 l Rirectionality 1 ll 1 Response: edp eat with us 1 disagree 2 uncertain 3 agree l2 1 grammatical form 2 13 2 D 1 1A 2 R: helpers are edp 1 disagree 2 uncertain 3 agree 15 2 G 2 l6 3 D 2 l7 3 R: superiors not edp 1 agree 2 uncertain 3 disagree l8 3 G 2 19 A D 2 20 A R: we do not live near 1 agree 2 uncertain 3 disagree 21 A G 1 22 5 D 2 23 5 R: edp not invited 1 agree 2 uncertain 3 disagree 2A 5 G 2 25 6 D 2 26 6 R: subordinates not edp 1 agree 2 uncertain 3 disagree 27 6 G 2 28 7 D . 2 29 7 R: renters not edp 1 agree 2 uncertain 3 disagree 30 7 G l 179 Col. Item Content Code 31 8 2 32 8 we do not date 1 agree 2 uncertain 3 disagree 33 8 1 3A 9 1 35 9 we do not avoid 1 disagree 2 uncertain 3 agree 36 9 1 37 10 2 38 10 we do not borrow $ 1 agree 2 uncertain 3 disagree 39 10 1 A0 11 2 Al 11 persons fought are edp 1 agree 2 uncertain 3 disagree A2 11 2 A3 12 2 AA 12 borrowers not edp 1 agree 2 uncertain 3 disagree A5 12 2 A6 13 1 A7 13 we are married to edp 1 disagree 2 uncertain 3 agree A8 13 1 A9 1A 1 50 1A we rent to edp 1 disagree 2 uncertain 3 agree 51 1A 1 180 Col. Item Content Code 52 15 l 53 15 : we borrow things from 1 disagree 2 uncertain 3 agree 5A 15 1 55 16 1 56 16 friends are edp 1 disagree 2 uncertain 3 agree 57 16 2 58 17 1 59 17 people helped are edp 1 disagree 2 uncertain 3 agree 6o 17 2 61 18 l 62 18 we loan money to 1 disagree 2 uncertain 3 agree 63 18 1 6A 19 l 65 19 we enjoy edp sexually 1 disagree 2 uncertain 3 agree 66 19 2 67 20 2 68 20 we do not work with 1 agree 2 uncertain 3 disagree 69 20 1 181 Code: Level member II3 (all cards, all SETS with "2" in col. 6 and "3" in col. 7) Col. Item Content Code 10 1 Rirectionality 2 ll 1 Response: edp lazier 1 agree 2 uncertain 3 disagree 12 l grammatical form 2 l3 2 D 2 1A 2 R: edp less intelligent 1 agree 2 uncertain 3 disagree 15 2 G 2 l6 3 D 2 17 3 R: I more sexually 1 agree attractive 2 uncertain 3 disagree l8 3 G l 19 A D l 20 A R: edp cleaner 1 disagree 2 uncertain 3 agree 21 A G 2 22 5 D l 23 5 R: I less cooperative 1 disagree 2 uncertain 3 agree 2A 5 G l 25 6 D 1 26 6 R: I less friendly 1 disagree 2 uncertain 3 agree 27 6 G l 28 7 D 2 29 7 R: I more mature 1 agree 2 uncertain 3 disagree 30 7 G l 182 Col. Item Content Code 31 8 l 32 8 edp more helpful 1 disagree 2 uncertain 3 agree 33 8 2 3A 9 2 35 9 edp more cruel 1 agree 2 uncertain 3 disagree 36 9 2 37 10 2 38 10 I more loyal 1 agree 2 uncertain 3 disagree 39 10 1 A0 11 1 A1 11 I less energetic 1 disagree 2 uncertain 3 agree A2 11 1 A3 12 1 AA 12 edp better looking 1 disagree 2 uncertain 3 agree A5 ' l2 2 A6 13 2 A7 13 I less prejudiced 1 agree 2 uncertain 3 disagree A8 13 1 A9 1A 2 50 1A edp more careless 1 agree 2 uncertain 3 disagree 51 1A 2 183 Col. Item Content Code 52 15 D 2 53 15 R: edp less understanding 1 agree 2 uncertain 3 disagree 5A 15 G 2 55 16 D 1 56 16 R: I less reliable I disagree 2 uncertain 3 agree 57 16 G l 58 17 D l 59 17 R: edp happier 1 disagree 2 uncertain 3 agree 60 17 G 2 61 18 D l 62 18 R: edp more honest 1 disagree 2 uncertain 3 agree 63 18 G 2 6A 19 D 2 65 19 Re: I more talented 1 agree 2 uncertain 3 disagree 66 19 G l 67 20 D l 68 20 R: I less healthy 1 disagree 2 uncertain 3 agree 69 20 G l 18A Code: Level member IIIl (all cards, all SETS with "3" in col. 6 and "l" in col. 7) Col. Item Content Code 10 1 girectionality l 11 1 Response: edp allowed 1 disagree to eat 2 uncertain 3 agree 12 l grammatical form 2 l3 2 D 1 1A 2 Re: edp asked for help 1 disagree 2 uncertain 3 agree 15 2 G 2 l6 3 D 2 l7 3 R: edp not superiors 1 agree 2 uncertain 3 disagree l8 3 G 2 19 A D 2 20 A R: we should not live near 1 agree 2 uncertain 3 disagree 21 A G 1 22 5 D 2 23 5 R: edp not invited 1 agree 2 uncertain 3 disagree 2A 5 G 2 25 6 D 2 26 6 R: edp not subordinates 1 agree 2 uncertain 3 disagree 27 6 G 2 28 7 D 2 29 7 R: we should not rent from 1 agree 2 uncertain 3 disagree 30 7 G 1 185 Col. Item Content Code 31 8 D 2 32 8 R: we should not date 1 agree 2 uncertain 3 disagree 33 8 G 1 3A 9 D 1 35 9 R: we should not avoid 1 disagree 2 uncertain 3 agree 36 9 G l 37 10 D 2 38 10 R: we should not borrow $ 1 agree from 2 uncertain 3 disagree 39 10 G 1 A0 11 D 2 Al 11 R: edp should be fought 1 agree 2 uncertain 3 disagree A2 11 G 2 A3 12 D 2 AA 12 R: edp not loaned things 1 agree 2 uncertain 3 disagree A5 12 G 2 A6 13 D 1 A7 13 R: we should marry edp 1 disagree 2 uncertain 3 agree A8 13 G 1 A9 1A D l 50 1A R: we should rent to edp 1 disagree 2 uncertain 3 agree 51 1A G 1 186 Col. Item Content Code 52 15 1 53 15 : we should borrow from 1 disagree 2 uncertain 3 agree 5A 15 l 55 16 l 56 16 edp accepted as friends 1 disagree 2 uncertain 3 agree 57 16 2 58 17 l 59 17 edp helped by us 1 disagree 2 uncertain 3 agree 60 17 2 61 18 1 62 18 we should loan $ to 1 disagree 2 uncertain 3 agree 63 18 1 6A 19 1 65 19 edp enjoyed sexually 1 disagree 2 uncertain 3 agree 66 19 2 67 20 2 68 20 we should not work with 1 agree 2 uncertain 3 disagree 69 2o 1 187 Code: Level member III2 (all cards, all SETS with "3" in col. 6 and "2" in col. 7) Col. Item Content Code 10 l girectionality 2 ll 1 Response: edp lazier 1 agree 2 uncertain 3 disagree 12 l grammatical form 2 l3 2 D 2 1A 2 R: edp less intelligent 1 agree 2 uncertain 3 disagree 15 2 G 2 l6 3 D 2 17 3 R: I more sexually 1 agree attractive 2 uncertain 3 disagree 18 3 G 1 19 A D l 20 A R: edp cleaner 1 disagree 2 uncertain 3 agree 21 A G 2 22 5 D l 23 5 R: I less cooperative 1 disagree 2 uncertain 3 agree 2A 5 G 1 25 6 D l 26 6 R: I less friendly 1 disagree 2 uncertain 3 agree 27 6 G l 28 7 D 2 29 7 R: I more mature 1 agree 2 uncertain 3 disagree 3O 7 G l 188 Col. Item Content Code 31 8 l 32 8 edp more helpful 1 disagree 2 uncertain 3 agree 33 8 2 3A 9 2 35 9 edp more cruel 1 agree 2 uncertain 3 disagree 36 9 2 37 10 2 38 10 I more loyal 1 agree 2 uncertain 3 disagree 39 10 1 A0 11 1 Al 11 I less energetic 1 disagree 2 uncertain 3 agree A2 11 1 A3 12 1 AA 12 edp better looking 1 disagree 2 uncertain 3 agree A5 12 2 A6 13 2 A7 13 I less prejudiced 1 agree 2 uncertain 3 disagree A8 13 1 A9 1A 2 50 1A edp more careless 1 agree 2 uncertain 3 disagree 51 1A 2 189 Col. Item Content Code 52 15 2 53 15 edp less understanding 1 agree 2 uncertain 3 disagree 5A 15 2 55 16 1 56 16 I less reliable 1 disagree 2 uncertain 3 agree 57 16 1 58 17 l 59 17 edp happier 1 disagree 2 uncertain 3 agree 60 17 2 ~ 61 18 1 62 18 edp more honest 1 disagree 2 uncertain 3 agree 63 18 2 6A 19 2 65 19 I more talented 1 agree 2 uncertain 3 disagree 66 19 1 67 2O 1 68 20 I less healthy 1 disagree 2 uncertain 3 agree 69 20 l 190 Code: Level member III3 (all cards, all SETS with "3" in col. 6 and "3" in col. 7) Col. Item Content Code 10 l girectionality 1 11 1 Response: edp eat with 1 disagree 2 uncertain 3 agree 12 l grammatical form 2 l3 2 D 1 1A 2 R: edp asked for help 1 disagree 2 uncertain 3 agree 15 2 G 2 l6 3 D 2 l7 3 R: superiors not edp 1 agree 2 uncertain 3 disagree 18 3 G 2 19 A D 2 20 A R: I do not live near 1 agree 2 uncertain 3 disagree 21 A G 1 22 5 D 2 23 5 R: edp not invited 1 agree 2 uncertain 3 disagree 2A 5 G 2 25 6 D 2 26 6 R: subordinates not edp 1 agree 2 uncertain 3 disagree 27 6 G 2 28 7 D 2 29 7 R: I do not rent from 1 agree 2 uncertain 3 disagree 3O 7 G l 191 Col. Item Content Code 31 8 D 2 32 8 R: I do not date 1 agree 2 uncertain 3 disagree 33 8 G 1 3A 9 D 1 35 9 R: I do not avoid 1 disagree 2 uncertain 3 agree 36 9 G l 37 10 D 2 38 10 R: I do not borrow $ from 1 agree 2 uncertain 3 disagree 39 10 G 1 A0 11 D 2 Al 11 R: edp fought 1 agree 2 uncertain 3 disagree A2 11 G 2 A3 12 D 2 AA 12 R: edp not loaned things 1 agree 2 uncertain 3 disagree A5 12 G 2 A6 13 D 1 A7 13 R: I marry edp 1 disagree 2 uncertain 3 agree A8 13 G 1 A9 lA D l 50 1A R: I rent to 1 disagree 2 uncertain 3 agree 51 1A G l 52 15 D l 53 15 R: I borrow things from 1 disagree 2 uncertain 3 agree 5A 15 G l 192 Col. Item Content Code 55 16 l 56 16 edp accepted as friends 1 disagree 2 uncertain 3 agree 57 16 2 58 17 1 59 17 edp helped 1 disagree 2 uncertain 3 agree 6O 17 2 61 18 l 62 18 I loan $ to 1 disagree 2 uncertain 3 agree 63 18 1 6A 19 l 65 19 edp enjoyed sexually 1 disagree 2 uncertain 3 agree 66 19 2 67 2O 2 68 20 I do not work with 1 agree 2 uncertain 3 disagree 69 20 1 193 Code: Level member IIIA (all cards, all SETS with "3" in col. 6 and "A" in col. 7) Col. Item Content Code 10 l girectionality 2 ll 1 Response: edp displease 1 agree 2 uncertain 3 disagree 12 l Grammatical form 2 l3 2 D 2 1A 2 R: we negative 1 agree 2 uncertain 3 disagree 15 2 G 2 l6 3 D 2 17 3 R: we dislike 1 agree 2 uncertain 3 disagree 18 3 G l 19 A D l 20 A R: edp do not make us 1 disagree tense 2 uncertain 3 agree 21 A G 2 22 5 D l 23 5 R: we are not frightened 1 disagree by 2 uncertain 3 agree 2A 5 G l 25 6 D l 26 6 R: edp do not anger us 1 disagree 2 uncertain 3 agree 27 6 G 2 28 7 D l 29 7 R: we do not feel contempt 1 disagree 2 uncertain 3 agree 3O 7 G l 19A Col. Item Content Code 31 8 l 32 8 : we are not uncomfor- 1 disagree table near 2 uncertain 3 agree 33 8 1 3A 9 l 35 9 edp do not dismay us 1 disagree 2 uncertain 3 agree 36 9 2 37 10 2 38 10 : we loath 1 agree 2 uncertain 3 disagree 39 10 1 A0 11 1 Al 11 : we are attracted 1 disagree 2 uncertain 3 agree A2 11 1 A3 12 2 AA 12 edp make us unsym- 1 agree pathetic 2 uncertain 3 disagree A5 12 2 A6 13 2 A7 13 : we are not at ease near 1 agree 2 uncertain 3 disagree A8 13 1 A9 1A 2 50 1A edp make us unhappy 1 agree 2 uncertain 3 disagree 51 1A 2 195 Col. Item Content Code 52 15 2 53 15 edp confuse us 1 agree 2 uncertain 3 disagree 5A 15 2 55 16 1 56 16 edp do not repulse 1 disagree 2 uncertain 3 agree 57 16 2 58 17 l 59 17 ° we feel secure near 1 disagree 2 uncertain 3 agree 60 17 l 61 18 l 62 18 edp attract us 1 disagree sexually 2 uncertain 3 agree 63 18 2 6A 19 2 65 19 we are restless near 1 agree 2 uncertain 3 disagree 66 19 1 67 20 2 68 20 : we are not calm near 1 agree 2 uncertain 3 disagree 69 2O 1 196 Code: Level member IVl (all cards, all SETS with "A" in col. 6 and "l" in col. 7) Col. Item Content Code 10 l girectionality 1 11 1 Response: edp eat with 1 disagree 2 uncertain 3 agree l2 1 grammatical form 2 l3 2 D 1 1A 2 R: edp asked for help 1 disagree 2 uncertain 3 agree 15 2 G 2 l6 3 D 2 17 3 R: superiors not edp 1 agree 2 uncertain 3 disagree l8 3 G 2 19 A D 2 20 A R: I would not live near 1 agree 2 uncertain 3 disagree 21 A G l 22 5 D 2 23 5 R: edp not invited 1 agree 2 uncertain 3 disagree 2A 5 G 2 25 6 D 2 26 6 R: subordinates not edp 1 agree 2 uncertain 3 disagree 27 6 G 2 28 7 D 2 29 7 R: I would not rent from 1 agree 2 uncertain 3 disagree 3O 7 G l 197 Col. Item Content Code 31 8 2 32 8 I would not date 1 agree 2 uncertain 3 disagree 33 8 1 3A 9 1 35 9 I would not avoid 1 disagree 2 uncertain 3 agree 36 9 1 37 10 2 38 10 I would not borrow $ 1 agree from 2 uncertain 3 disagree 39 10 1 A0 11 2 Al 11 edp fought 1 agree 2 uncertain 3 disagree A2 11 2 A3 12 2 AA 12 edp not loaned things 1 agree 2 uncertain 3 disagree A5 12 2 A6 13 1 A7 13 I would marry edp 1 disagree 2 uncertain 3 agree A8 13 1 A9 1A 1 50 1A I would rent to 1 disagree 2 uncertain 3 agree 51 1A 1 198 Col. Item Content Code 52 15 D l 53 15 R: I would borrow things 1 disagree from 2 uncertain 3 agree 5A 15 G l 55 16 D l 56 16 R: edp accepted as friends 1 disagree 2 uncertain 3 agree 57 16 G 2 58 17 D l 59 17 R: edp helped 1 disagree 2 uncertain 3 agree 60 17 G 2 61 18 D l 62 18 R: I would loan $ to 1 disagree 2 uncertain 3 agree 63 18 G 1 6A 19 D l 55 19 R: edp enjoyed sexually 1 disagree 2 uncertain 3 agree 66 19 G 2 67 20 D 2 68 20 R: I would not work with 1 agree 2 uncertain 3 disagree 69 20 G 1 199 Code: Level member IV2 (all cards, all SETS with "A" in col. 6 and "2" in col. 7) Col. Item Content Code 10 l girectionality 1 ll 1 Response: edp eat with 1 disagree 2 uncertain 3 agree l2 1 grammatical form 2 l3 2 D 1 1A 2 R: edp asked for help 1 disagree 2 uncertain 3 agree l5 2 G 2 l6 3 D 2 l7 3 R: superiors not edp 1 agree 2 uncertain 3 disagree l8 3 G 2 19 A D 2 20 A R: we do not live near 1 agree 2 uncertain 3 disagree 21 A G l 22 5 D 2 23 5 R' edp not invited 1 agree 2 uncertain 3 disagree 2A 5 G 2 25 6 D 2 26 6 R: subordinates not edp 1 agree 2 uncertain 3 disagree 27 6 G 2 28 7 D 2 29 7 R: We do not rent from 1 agree 2 uncertain 3 disagree 3O 7 G 1 200 Col. Item Content Code 31 8 D 2 32 8 R: we do not date 1 agree 2 uncertain 3 disagree 33 8 G 1 3A 9 D 1 35 9 R: we do not avoid 1 disagree 2 uncertain 3 agree 36 9 G 1 37 10 D 2 38 10 R: we do not borrow $ 1 agree from 2 uncertain 3 disagree 39 10 G 1 A0 11 D 2 Al 11 R: edp fought 1 agree 2 uncertain 3 disagree A2 11 G 2 A3 12 D 2 AA 12 R: edp not loaned things 1 agree 2 uncertain 3 disagree A5 12 G 2 A6 13 D 1 A7 13 R: we marry edp 1 disagree 2 uncertain 3 agree A8 13 G 1 A9 1A D l 50 1A R: we rent to 1 disagree 2 uncertain 3 agree 51 1A G l 52 15 D 1 53 15 R: we borrow things from 1 disagree 2 uncertain 3 agree 5A 15 G l 201 Col. Item Content Code 55 16 l 56 16 edp accepted as 1 disagree friends 2 uncertain 3 agree 57 16 2 58 17 1 59 17 edp helped 1 disagree 2 uncertain 3 agree 6O 17 2 61 18 1 62 18 we loan $ to 1 disagree 2 uncertain 3 agree 63 18 1 6A 19 l 65 19 edp enjoyed sexually 1 disagree 2 uncertain 3 agree 66 19 2 67 20 2 68 20 : we do not work with 1 agree 2 uncertain 3 disagree 69 20 l 202 Code: Level member V1 (all cards, all SETS with "5" in col. 6 and "l" in col. 7) Col. Item Content Code 10 l girectionality 2 ll 1 Response: edp displease 1 agree 2 uncertain 3 disagree l2 1 grammatical form 2 l3 2 D 2 1A 2 R: I negative 1 agree 2 uncertain 3 disagree 15 2 G 2 16 3 D 2 17 3 R' I dislike 1 agree 2 uncertain 3 disagree 18 3 G l 19 A D 1 20 A R: edp do not make me tense 1 disagree 2 uncertain 3 agree 21 A G 2 22 5 D l 23 5 R: I am not frightened by 1 disagree 2 uncertain 3 agree 2A 5 G 1 25 6 D l 26 6 R: edp do not anger me 1 disagree 2 uncertain 3 agree 27 6 G 2 28 7 D l 29 7 R: I do not feel contempt 1 disagree 2 uncertain 3 agree 30 7 G 1 203 Col. Item Content Code 31 8 2 32 8 I am not comfortable 1 agree near 2 uncertain 3 disagree 33 8 1 3A 9 l 35 9 edp do not dismay me 1 disagree 2 uncertain 3 agree 36 9 2 37 10 2 38 10 I loath 1 agree 2 uncertain 3 disagree 39 10 1 A0 11 1 Al 11 I am attracted 1 disagree 2 uncertain 3 agree A2 11 1 A3 12 2 AA 12 edp make me unsym- 1 agree pathetic 2 uncertain 3 disagree A5 12 2 A6 13 2 A7 13 I am not at ease near 1 agree 2 uncertain 3 disagree A8 13 1 A9 1A 2 50 1A edp make me unhappy 1 agree 2 uncertain 3 disagree 51 1A 2 20A Col. Item Content Code 52 15 2 53 15 edp confuse me 1 agree 2 uncertain 3 disagree 5A 15 2 55 16 1 56 16 edp do not repulse me 1 disagree 2 uncertain 3 agree 57 16 2 58 17 1 59 17 I feel secure near 1 disagree 2 uncertain 3 agree 6O 17 1 61 18 l 62 18 edp attract me sexually 1 disagree 2 uncertain 3 agree 63 18 2 6A 19 2 65 19 I am restless near 1 agree 2 uncertain 3 disagree 66 19 1 67 20 2 68 20 I am not calm near 1 agree 2 uncertain 3 disagree 69 20 l 205 Code: Level member VIl (all cards, all SETS with "6" in col. 6 and "l" in col. 7) Col. Item Content Code 10 1 girectionality l 11 1 Response: edp eat with 1 disagree 2 uncertain 3 agree 12 l grammatical form 2 l3 2 D 1 1A 2 R: edp asked for help 1 disagree 2 uncertain 3 agree l5 2 G 2 l6 3 D 2 l7 3 R: superiors not edp 1 agree 2 uncertain 3 disagree 18 3 G 2 19 A D 2 20 A R: I do not live near 1 agree 2 uncertain 3 disagree 21 A G l 22 5 D 2 23 5 R: edp not invited 1 agree 2 uncertain 3 disagree 2A 5 G 2 25 6 D 2 26 6 R: subordinates not edp 1 agree 2 uncertain 3 disagree 27 6 G 2 28 7 D 2 29 7 R: I do not rent from 1 agree 2 uncertain 3 disagree 30 7 G l 206 Col. Item Content Code 31 8 D 2 32 8 R: I do not date 1 agree 2 uncertain 3 disagree 33 8 G 1 3A 9 D l 35 9 R: I do not avoid 1 disagree 2 uncertain 3 agree 36 9 G l 37 10 D 2 38 10 R: I do not borrow $ from 1 agree 2 uncertain 3 disagree 39 10 G 1 A0 11 D 2 A1 11 R: edp fought 1 agree 2 uncertain 3 disagree A2 11 G 2 A3 12 D 2 AA 12 R: edp not loaned things 1 agree 2 uncertain 3 disagree A5 12 G 2 A6 13 D 1 A7 13 R: I marry edp 1 disagree 2 uncertain 3 agree A8 13 G 1 A9 1A D l 50 1A R: I rent to 1 disagree 2 uncertain 3 agree 51 1A G l 207 Col. Item Content Code 52 15 1 53 15 I borrow things from 1 disagree 2 uncertain 3 agree 5A 15 l 55 16 l 56 16 edp accepted as friends 1 disagree 2 uncertain 3 agree 57 16 2 58 17 l 59 17 edp helped 1 disagree 2 uncertain 3 agree 60 17 2 61 18 l 62 18 I loan $ to 1 disagree 2 uncertain 3 agree 63 18 1 6A 19 l 65 19 edp enjoyed sexually 1 disagree 2 uncertain 3 agree 66 19 2 67 20 2 68 20 I do not work with 1 agree 2 uncertain 3 disagree 69 20 l APPENDIX F IDENTIFICATION OF VARIABLES IDENTIFICATION OF VARIABLES . Introduction As indicated in Chapters III and IV, the SCOpe of the present study does not include complete analysis of individual items or of the varying effects upon individ- ual scores from such variables as directionality of items, grammatical form of items, or general content of items. The data.have however, been identified and coded to make 53 such analyses possible. By a combination of card-sorting and programming procedures, response data (in columns 11, 1A, 17 . . . 68) of all cards can be identified in various ways. Among the variables of interest are: (a) conjoint—struction specifi— city—-how do varying level members affect scoring patterns? (b) directionality——does item directionality affect rev sponse? (c) grammatical form——does grammatical emphasis affect response? (d) general content (see Chapter IV)-- do response patterns vary across different types of dis— joint struction? (e) sematic path——does the SET of level members as a whole affect responses to individual level— member items? (f) order of administration of semantic path--are some orders of administration more effective in producing discriminating scores? (g) position within 209 210 administered sequence-—do scores reflect how early or late in a particular instrument sets of items are placed? and (h) SET--comparisons across SETS must be done under the limitation that subjects varied, in the original data, from SET to SET; the analyses which comprise the main part of Chapter V have been done within rather than across SETS. The Code Book (Appendix E) contains all information necessary for the identification of data according to such variable requirements. The following examples may be of use in such identification. Variable Identification Conjoint—struction specificity may be of interest: 12 level members have been identified; level number is given in column 6 and member number in column 7. For example, Level member Il appears on varying cards in all 1A SETS; is identified in column 6 ("1" and column 7 ("l"); and comprises SET Cardl SET Card 11 3 12 l 21 3 22 l 31 3 32 1 Al 3 A2 1 51 5 52 l 61 A 62 1 71 2 72 l 1See Appendix G 211 Directionality of items may affect responses: agree— ment with some items indicates a favorable attitude toward the attitude object; agreement with other items indicates an unfavorable attitude. For example, those items for which "agree" indicates a favorable attitude in SET 11, card 1 (level member V11) fin are coded in columns 10, l3, l6 . . . 67 (code "1"); appear in columns 11, 1A, 17 . . . 68; and comprise: Coding Column Data Column (Item) 5A 10 11 1 13 1A 2 3A 35 9 A6 A7 13 A9 50 1A 52 53 15 55 56 16 58 59 17 61 62 18 6A 65 19 A computer routine to identify and analyse such data would (a) read coding column (10, l3, 16, etc.): if "1" record data in following column; if "2" skip next two col— umns; (b) read next coding column . . . etc. Grammatical form of items may affect responses: the "actor" is grammatically emphasized in some items; in other items the attitude object is grammatically emphasized. For example, those items in which the actor is grammatically emphasized in SET 11, card 1 are coded in columns 12, 15, 18 . . . 69; 212 appear in columns 11, 1A, 17 . . . 68; and comprise: Coding Column Data Column (Item) 21 20 )4 3g 32 8 3 35 9 39 38 10 h “8 A7 13 . 5“ 53 15 63 62 18 69 68 20 General content (S, A, or F: see Chapter IV) can g be identified by sorting on column 8. S — content, coded "1," is found in level members 11, 111, 113, III2. A — content, coded "2," is found in level members II2, IIIl, III3, IVl, IV2, VIl. Content F, coded "3," is found in level members IIIA and V1. An individual item appears in parallel forms in identical position across all level mem- bers of the same general content. For example, item 1: 11: We think that most emotionally disturbed persons are lazier than we are. Ill: 1 think that most emotionally disturbed persons are lazier than people like myself. 113: Other people think that most emotionally dis— turbed persons are lazier than I am. III2: I think that most emotionally disturbed persons are lazier than I am. Semantic paths may have varying characteristics. For example, path A (code "1" in column 1, all cards) comprises SET 11 all cards (path A, random order), SET 12 all cards (path A, level-by-level order). Al I {fag-1111‘ 1| Ill. 213 Order of administration, whether level-by-level or random, may affect the relationship of level members within a semantic path. For example, all randomly-ordered se- mantic paths are coded in column 2 (code "1"), and comprise SET All cards 11 N l! 2 l n H 31 H II [4 l I! u 51 I! ll 6 l n It 71 n I! Position within a given sequence of level members may affect response. For example, the level member administered fourth in any semantic path, any order, is coded in column 5 (same as card number), and comprises SET Card Level member 11 A 1111 21 A V11 31 A VIl A1 A V11 51 A V1 61 A 11 71 A IIIA 12 A IVl 22 A IVl 32 A IVl A2 A IVl 52 A IVl 62 A 1V1 72 A IV2 Mi 21A §§g§, corresponding to random or level-by—level ad— ministrations of semantic paths, constituted the basic units for the analyses described in Chapters IV and V. As indicated in the Code Book, any semantic path can be identi- fied in column 1 (path) and column 2 (order of administration) of all cards. Fa Conclusion Combinations of sorting and programming procedures can be used to isolate and analyse the variables indicated g above. Further analyses, such as a two—by-two analysis of variance utilizing directionality of item and grammatical form of item might also be of interest. For such an analysis, a routine could be devised which would sum in four different totals, from columns 11, 1A, 17 . . . 68, those items which are: (a) preceded by "l" and followed by "1"; (b) preceded by "l" and followed by "2"; (c) preceded by "2" and followed by "l"; and (d) preceded by "2" and followed by "2." The resulting data would belong to the following cells of an ANOVA table: (a) agree=favorable, actor emphasized; (b) agree=favorable, object emphasized; (c) agree=unfavorable, actor emphasized; and (d) agree=unfavorable, object em- phasized. APPENDIX G SEQUENCE OF LEVEL MEMBERS IN SEMANTIC PATHS SEQUENCE OF LEVEL MEMBERS IN SEMANTIC PATHS As indicated in Chapter IV, each semantic path was administered in two orders: (a) a random order; and (b) level-by-level. The following summary may be of help in conjunction with Appendices E ("Code Book") and F ("Identification of Variables"). Path Level Members Sequence Sequence Random Level—by-level A - II III 1111 IV1 V1 VIl B Il IIl III2 IV1 VI VII C I1 . II2 IIIl IV1 V1 VIl D II II2 III3 IV1 V1 VIl tMUTl—‘Chw tUTl—‘NONUO .IZ'I-J O\U1 HUD Hmox-zmuu- oxmzme O\U'|£-'U)I\JI—' O\U'l-I=WN|—’ Chm-Emmi” 216 Path Level Members Sequence Sequence Random Level—byvlevel E I1 II3 III2 IV1 V1 VII mmzme F I1 II3 III3 IV1 V1 VIl Chm-DUONH G I1 II2 IIIA IV2 .1:me GLOSSARY GLOSSARY Appro ximat ion--see "simplex approximation . " Attitude—-"delimited totality of behavior with respect tc> scnnething" (Guttman, 1950, p. 51). Conjoint struction-~see also "struction," "disjoint strunrtion"--"operationally defined as the ordered sets of . five facets from low to high across all five facets simultaneously" (Jordan, 1968a, p. 76); that part of the semantic structure of attitude items which can be determined indepen- dently of specific response situations. Content-~situation (action, feeling, comparison, cir— cnmmtances) indicated in an attitude item; gen- enelly corresponds to "disjoint struction." Defhfltimnfl.statement--specification of characteristics pmnmr to an item of a given level member, typi- cany stated in phrase or clause form. kahntimmd system--ordered group of definitional state- mmmscu'of’the corresponding level members; typi- cfliyeither the group constituting a "semantic pmm"or the complete group of 12 level members in Hm'temantic map." 219 220 Direct10nality--characteristic of an item, sometimes called positive or negative, determining agreement with the item as indicating favorableness or unfavorableness toward the attitude object. Disjoin‘t struction--see also "struction," "conjoint struc- tion"--that part of the semantic structure of attitude F‘L items which is directly dependent on specification .-m~.. VP'- ' MEI- of situation and object; a more precise term than "content." Element--one of two or more ways in which a facet may be ”A expressed; in the present system, all conjoint facets are dichotomous, expressed in one of two ordered elements. Emotionally disturbed--"those children or adults whose be- haviors, feelings, or emotions cause them to have difficulties with every-day problems which they are unable to solve (SAF Scale, Form ED-l; Appendix 0)." Facet--one of several semantic units distinguishable in the verbal expression of an attitude; in the present sys- tem, five dichotomous facets are noted within the conjoint struction; see Chapter VI for discussion of disjoint struction. Facet profile-wee "struction profile." Leveludegree of attitude strength specified by the number of strong and weak facets in the member(s) of that level; in the present system, six ordered levels are 221 identified: level I is characterized by the unique member having five weak facets; level II, by members having four weak and one strong facet . . . level VI, by the unique member having five strong facets. Level member—~one of one or more permutation(s) of strong and weak facets which are common to a given level; in the present system, 12 level members have been identified: three on level II, four on level III, two on level IV, and one each on levels I, V, and VI. Map--see "semantic map." Member--see "level member." Path--see "semantic path." .Profile--see "struction profile." Reversal--change in a specified order of levels or of cor- relations, involving only the two indicated levels or correlations. Semantic--pertaining to or arising from the varying meanings, grammatical forms, or stylistic emphasis of words, phrases, or clauses. Semantic map—-two—dimensional representation of hypothesized relationships among six levels and among 12 level members (of. Chapter III). Semantic path--ordered set of level members, typically six, such that each member has one more strong facet than the immediately preceding member and one less strong facet than the immediately following member. 222 Semantic possibility analysis-~linguistic discussion of the implications of the five dichotomous conjoint facets identified in the present system; of 32 permutations, only 12 are considered logically consistent. Simplex—~specific form of (correlation) matrix, diagonally dominated and decreasing in magnitude away from the main diagonal; see Chapter VI for comparison of equally spaced and unequally-spaced diagonals. Simplex approximation--matrix which approaches more or less perfectly the simplex form; existing tests (Kaiser, 1962; Mukherjee, 1966) reflect both ordering of in— dividual entries and sizes of differences between entries and between diagonals. Strong(er)--opposite of weak(er)--term functionally assigned to one of two elements, to a facet expressed by its strong element, or to a level member characterized by more strong facets than another level member; the strong - weakcxnfidxnnnnis presently examined as un- idimensional; see Chapters II and III for identi- fication of strong and weak elements. Struction--see also "conjoint struction," "disjoint struc- tion"--semantic pattern identifiable in any attitude item, or the system of such identifications. Struction profile-~specification, typically indicated by small letters and numerical subscripts, of the per— mutation(s) of weak and strong elements or facets in 223 a level member or a set of level members; or of permutations of disjoint elements or facets. Transposition--change in a specified order of levels or of correlations involving a change in position of one level or correlation and the corresponding one- place shift in the position of following or preceding levels or correlations. Weak-—opposite of "strong" (which see). REFERENCES REFERENCES Adis-Castro, G., & Waisenen, F. B. Attitudes toward mental illness: Some socio-economic and modernization cor- relatives. San Jose, Costa Rica: University of Costa Rica, 1965. (a) Adis—Castro, G., & Waisenen, F. B. Modernity and tolerance: The case of attitudes toward mental illness. San Jose, Costa Rica: University of Costa Rica, 1965. (b) Adis-Castro, G., & Waisenen, F. B. The socio-economic con- text of attitudes toward mental illness. San Jose, Costa Rica: University of Costa Rica, 1965. (c) Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., & Sanford, R. N. The authoritarian personality. New York: Wiley, 196A. Allport, G. w. Attitudes. In C. Murchison (Ed.), Handbook of social psychology. Worchester, Mass.: Clark University Press, 1935. Anderson, H. E. The effects of response sets in question- naire studies. American Journal of Occupational Therapy, 1965, 19, 3A8-35o. Anderson, L. R. Some personality correlates of the strength of belief and strength of affect dimensions of the summation theory of attitude. Journal of Social Psychology, 1968, 7A, 25-38. Askenasy, A. R. Attitudes toward mental patients and their rehabilitation: A study of mental health personnel in an Atlantic boarder state in Hawaii and in England. Part one of a two part final report, Office of V0- cational Rehabilitation,Project RD A26,LWorld Federa- tion for Mental Health: U. S. Committee, New York, 1963. Bolles, M. M., Metzger, H. F., & Pitts, M. W. -Early home background and personality adjustment. American Journal of Orthopsychiatry, l9Al, 20, 530-535. 225 226 Boyd, G. R. Classroom adjustment of the underchosen child thropgh changes in teachers' attitudes and behavior. Troy, Alabama: Troy State College, 19A3. Cohen, J., & Struenning, E. L. Opinions about mental il— lness in the personnel of two large mental hospitals. Journal of Abnormal and Social Psychology, 1962, 6A, 3A9-360. Cohen, J., & Struenning, E. L. Opinions about mental il- lness: Mental hospital occupational profiles and profile clusters. Psychological Reprints, 1963, 12 (1), 111-12A. Crano, W. D., & Schroder, H. M. Complexity of attitude structure and processes of conflict reduction. Jour- nal of Personality and Social Psychology, 1965, 5, 110-llA. DeBaty, P. La mesure des attitudes. Paris: Presses Universitaires de France, 1967. DuBois, P. H. An analysis of Guttman's simplex. Psychome- trika, 1960, 25, 173-182. Edwards, A. L. Techniques of attitude scale construction. New York: Appleton-Century Crofts, 1957. Edwards, A. L. Experimental design inppsychological re- search. New York: Holt, Rinehart and Winston, 196A. Erb, D. L. Racial attitudes and empathy: a Guttman facet theory examination of their relationships and de- terminants. Unpublished doctoral dissertation, Michigan State University, 1969. Farina, A., & Holzberg, J. D. Attitudes and behaviors of fathers and mothers of male schizophrenic patients. Journal of Abnormal Psychology, 1967, 72, 381-387. Farina, A., & Ring, K. The influence of perceived mental illness on interpersonal relations. Journal of Abnormal Psychology, 1965, 70, A7-51. Field, M. Maternal attitudes found in twenty-five cases of children with primary behavior disorders. American Journal of Orthopsychiatry, l9AO, 10, 293-296. Fishbein, M. (Ed.) Readings in attitude theopy-and mea- surement. New York: Wiley, 1967. 227 Foa, U. G. The contiguity principle in the structure of interpersonal relations. 'Human Relations, 1958, A, 229-239- Gatherer, A., & Reid, J. J. A. Public attitudes and mental health education. Northamptonshire, England: North- amptonshire County Council, 1963. Gilbert, D. Ideologies concerning mental illness: A socio- psychological study of mental hospital personnel. Un- published doctoral dissertation, Radcliffe University, 1r~ 195 . Gottlieb, K. A Guttman facet analysis of attitudes toward mental retardation in Colombia. Content, structure and determinants. Unpublished doctoral dissertation, Michigan State University, Scheduled Dec., 1969. Greenbaum, J. J., & Wang, D. D. A semantic-differential study of the concepts of mental retardation.. Journal of General Psychology, 1965, 73, 257-272. “.7 w! “'- Guilford, J. P. Response biases and response sets. In Psychometric methods. New York: McGraw-Hill, 195A. Gulliksen, H. A least squares solution for paired compari- sons with incomplete data. Psychometrika, 1956, 21,. 125-13A. Guttman, L. A basis for scaling qualitative data. American Sociological Review, 19AA, 9, 139-150. Guttman, L. The Cornell technique for scale and intensity analysis. Educational and Psychological Measurement, 19A7, 7, 2A7-250. Guttman, L. The problem of attitude and opinion measureh ment. In S. A. Stouffer (Ed.), Measurement and ‘ prediction. Princeton: Princeton University, 1950. Guttman, L. A new approach to factor analysis: The radex. In P. Lazarsfeld (Ed.), Mathematical thinkigg in the social sciences. Glencoe, Illinois: Free Press, 195A. Guttman, L. An outline of some new methodology for social research. Public Opinion Quarterly, 1955, 18, 395- AOA. Guttman, L. A structural theory for intergroup beliefs and action. American Sociolpgical Review, 1959, 2A, 318— 328. 228 Guttman, L., 8 Foa, U. G. Social contact and an inter- group attitude. Public Opinion Quarterly, 1951, 15, A5—53. Guttman, L., & Suchman, E. A. Intensity and a zero point . for attitude analysis. American Sociological Review, 19A7, 12, 57-67. Halloran, J. D. Attitude formation and change. Leicester: Leicester University Press, 1967. Hamersma, P. J., 8 Jordan, J. E. Attitude — Behavior Scale: BW/WN-G. East Lansing, Michigan: Michigan State University, 1969, available from authors. Hamersma, R. J. Construction of an attitude—behavior scale of Negroes and Whites toward each other using Guttman facet design and analysis. Unpublished doctoral dis- . sertation, Michigan State University, 1969. g; Harrelson, L. E. A Guttman facet analysis of attitudes to— ward the mentally retarded in the Federal Republic of Germany: Content, structure, and determinants. Un- published doctoral dissertation, Michigan State University, Scheduled Dec., 1969. Humphreys, L. G. Investigations of the simplex. Psych— ometrika, 1960, 25, 313—323. Insko, C. A. Verbal reinforcement of attitude. Journal of Personality and Social Psychology, 1965, 2, 62 - f, (1 2 ‘2. j o :5 I Insko, C. A. Theories of attitude change. New York: Ap— pleton—Century—Crofts, 1967. Jordan, J. E. Attitude-Behavior scale--MR (ABS-MR). East Lansing, Michigan: Michigan State University, 1967, available from author. Jordan J. E. Attitudes toward education and physically disabled persons in eleven nations. East Lansing, Michigan: Latin American Studies Center, Michigan State University, 1968. (a) Jordan, J. E. A Guttman facet theory analysis of attitudes and behaviors of blacks (Negroes) and whites toward each other. East Lansing, Michigan: Michigan State University, 1968, available from author. (b) Jordan, J. E., A Hamersma, R. J. Attitude-behavior scale: ck/white (ABS:BW and ABS:WN), U. S. 112268 version. t Lansing, Michigan: Michigan State University, 9, available from authors. I C )‘ p_._4 H F: G 0‘) DJ "(3 >17 229 Kaiser, H. F. Scaling a simplex. Psychometrika, 1962, 27, 155-162. Kasanin, J., Knight, E., & Sage, P. The parent-child relationship in schizophrenia. Journal of Nervous and Mental Disorders, 193A, 79, 2A9—263. Katz, D., & Stotland, E. A preliminary statement to a theory of attitude structure and change. In S. Koch, Psychology;, A stpgy of a science. New York: McGraw-Hill, 1959. Kerlinger, F. N. The attitude structure of the individual: A Q-study of the educational attitudes of profes- sionals and laymen. Genetic Psychology Monographs, 1956, 53, 285-239. Kerlinger, F. N. Foundations of behaviora} research. New York: Holt, Rinehart and Winston, 1966. Klebanoff, L. W. Parents of schizophrenic children: I. Parental attitudes of mothers of schizophrenic, brain-injured.and retarded, and normal children. American Journal of Orthopsychiatry, 1959, 29, AAS- Lazarsfeld, P. F. Latent structure analysis. In S. Koch (Ed.) Psychology: A study of a science. New York: McGraw-Hill, 1959. Likert, R. A technique for the measurement of attitudes. Archives of Psychology, 1932, 22, 5-A3. Loew, C. A. Acquisition of a hostile attitude and its re- lationship to agressive behavior. Journal of Per- sonality and Social Psychology, 1967, 5, 335+3Al. Mark, J. C. Attitudes of mothers of male schizophrenics toward child behavior. Journal of Abnormal and Social Psychology, 1953, A8, 185-189. McKeown, J. E. The behavior of parents of schizophrenics, neurotic, and.normal children. American Journal of Sociology, 1950, 56, 175-179. Morin, K. N. Attitudes of Texas Mexican-Americans toward mental retardation: A Guttman facet analysis. Un- published doctoral dissertation, Michigan State University, Scheduled Dec., 1969. Mosteller, F. Remarks on the method of paired comparisons: I. The least squares situation assuming equal stan- dard deviations and equal correlations. Psychome- trika, 1951, 16, 3-9. 230 Mukherjee, B. N. Derivations of likelihood-ratio tests for Guttman quasisimplex covariance structures. Psychometrika, 1966, 31, 97-123. Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. The measurement of meaning. Urbana: University of Illinois Press, 1957. Osgood, C. E., & Tannenbaum, P. H. The principle of con— gruity in attitude change. ’Psychological Review, 1955, 62, A2-A5. Price, K. O., Harburg, E., & Newcomb, T. M. Psychological balance in situations of negative interpersonal attitudes. Journal of Personality_and Social Psy- chology, 1966, 3, 265-270. Proctor, D. I. An investigation of the relationships be- tween knowledge of exceptional children, kind and amount of experience, and attitudes toward their classroom integration. Unpublished doctoral dis- sertation, Michigan State University, 1967. Prout, C. T., & White, M. A. A controlled study of per- sonality relationships in mothers of schizophrenic male patients. American Journal of Psychiatry, 1951, 107, 251-256. Reichard, S., & Tillman, C. Patterns of parent-child re- lationships in schizophrenia. Psychiatry, 1950, 13, 2A7-257. Rokeach, M. The open and closed mind. New York: Basic Books, 1961. Rokeach, M. Beliefs, attitudes, and values. San Francisco: Jossey-Bass, 1968. Shaefer, E. S., & Bell, R. Q. Development of a parental attitude research instrument. Child Development, 1958, 29, 339-351- Shaw, M. E., & Wright, J. M. Scales for measurement of at- titudes. New York: McGraw-Hill, 1967. Shepherd, I. L., & Guthrie, G. M. Attitudes of mothers of schizophrenic patients. Journal of Clinical Psychology, 1959, 15, 212-215. Sherif, C. W., & Sherif, M. (Eds.) Attitude; ego-involve- ment, and changy. New York: Wiley, 1967. 231 Shoben, E. J. The assessment of parental attitudes in re- lation to child adjustment. Genetic Psychology Monographs, l9A9, 39, 101-1A8. Sinha, B. K. Maternal attitudes and values in respect to emotionally disturbed and physically disabled persons. Unpublished doctoral dissertation, Michigan State University, 1966. Stephenson, W. The study of behavior. Chicago: University of Chicago, 1953. ' ”a Stotsky, B. A., & Rhetts, J. E. Functional significance of attitudes toward the mentally ill among nursing home personnel. Journal of Social Psychology, 1967, 71, 79-85. Stouffer, S. A. (Ed.) Measurement and prediction. Prince- 3 ton: Princeton University, 1950. Struenning, E. L., & Cohen, J. Factorial invariance and other psychometric characteristics of five opinions about mental illness factors. Educational and Psy- chological Measurement, 1963, 23, 289-298. Suchman, E. A., & Guttman, L. A solution to the problem of question bias. Public Opinion Quarterly, l9A7, 11, AA5-A55. Symonds, P. M. The nature of conduct. New York: MacMillan, 1928. Swarte, J. H. The mentally ill, stereopypes and attitudes. Amsterdam, the Netherlands: National Federation for Mental Health, 1968. Taylor, J. B. The organization of physician attitudes to- ward the emotionally disturbed patient. Journal of Health and Human Behavior, 1965, 2, 99-10A. Thurstone, J. R. A procedure for evaluating parental at- tides toward the handicapped. American Journal of Mental Deficiency, 1959, 6A, lA8-155. Thurstone, L. L. Attitudes can be measured. American Journal of Sociology, 1928, 33, 529-55“- Tietze, T. A study of mothers of schizophrenic patients. ngchiatry, l9A9, 12, 55-65. 232 Tittle, C. R., & Hill, R. J. Attitude measurement and pre- diction of behavior: An evaluation of conditions and measurement techniques. SOciometry, 1967, 30, 199-213. Torgerson, W. S. Theory and methods of scaling. New York: Wiley, 1958. Vinacke, W. E. Theypsychology of thinking. New York: McGraw-Hill, 1952. Wright, F. H., & Klein, R. A. Attitudes of hospital per- rs sonnel and the community regarding mental illness. Journal of Counseling Psychology, 1966, 13, 106-107. Zavallone, M., & Askenasy, A. Attitudes toward mental illness: A cross-cultural study.. New York: World Federation for Mental Health: U. S. Committee, 1963. Zinnes, J. L. Scaling. Annual Review of Psychology, 1969, g! 20, AA7-A78. . Zuckerman, M., Oltean, M., & Monashkin, I. The parental attitudes of mothers of schizophrenics. Journal of Consulting'PsyChology, 1958, 22, 307—310.