SELF {DENTIFECATIO‘N WWHEN A WECEFEC CGNTE??? OF EXPERIENCE, AND BEHAVIOR Thesis for ”10 Degree of DH. D. MICHIGAN STATE UNIVERSITY S. Clark McPhail 1965 'fLIBRARY Michigan State University This is to certify that the thesis entitled SELF IDENTIFICATION WITHIN A SPECIFIC CONTEXT OF EXPERIENCE AND BEHAVIOR presented by S . Clark McPhail has been accepted towards fulfillment of the requirements for Ph. D. degree inSociology if §7EMAT Major professor Date 74% 5/ 0-169 ABSTRACT SELF IDENTIFICATION WITHIN A SPECIFIC CONTEXT OF EXPERIENCE AND BEHAVIOR by S. Clark McPhail This study was conducted to obtain information regarding a fundamental social psychological question: "What are the basic processes which produce and maintain persons' conceptions and identi- fications of themselves?" The inquiry was derived from the theoretical orientation of George H. Mead and focused on a single class of self identifications and the experiential base on which that class of behaviors should rest: {religious self identification in relation to persons' impli- cation in a religious context of experience and behavior. Examination of previous studies indicated no consistent relationship between group affiliation and self identification in terms of that affiliation. This discrepancy was attributed to previous investi- gators' failure to attend to the classes of activities which should produce self identification in terms of Mead's theoretical orientation. Mean's orientation was examined and a tentative set of propositions was de- rived for the purpose of empirical test. Mead's major proposition holds that the self, defined as the acts of naming one's own activities, is produced by the behaviors which others address to the person relevant to the context of experience and behavior in which the person and those others are involved. Context of experience and behavior is defined as the activities in which the person is implicated, directly or indirectly, with others. No data were available to the investigator to directly test the major proposition. In the absence of observations of the behaviors directed by others toward the person within some context of experience and behavior, use was made of observations of the extent of persons' implication in a religious context of experience and behavior. Consis- tent with the theoretical orientation, implication in such activities should place the person in a position where others can address identifi- cation behaviors toward him in terms of those activities . Therefore. the major hypothesis submitted to test was: The greater the extent of the person's implication in a religious context of experience and be- havior the greater the likelihood that person will identify himself as religious. Four sets of observations of religious activities or cate- gories of religious interaction were presented and discussed as logically consistent referents for the construct of religious context of experience and behavior--RCEB: the extent of religious preference homogeneity between persons and their close friends--RHOMOG; persons' definition of their most important group as religious or non-religious--MIG; the extent of persons' subscription to statements of religious belief--RBEL; and, the extent of persons' participation in the formal and informal activities of religious groups and organizations--RPART. Persons' statements in response to the Self Identification Problem, the question "Who Am 1?", were taken as logically consistent referents for self as defined. These self designations were examined for the presence or absence of statements categorized as religious self identification state- ments. Three analytical procedures were employed to test the major and contingent sub -hypotheses. ( l) The association between RCEB components, separately and in various combinations, and the dependent variable of self identification was examined by means of the chi square statistic. The extent of association, where present, was determined by means of Cramer's V. (2) An examination was made of the per cent of cases accounted for in the predicted cells of MAXRCEB and MINRCEB. The SI % Index was computed to indicate the over-all per cent of cases accounted for in both cells in the test of the major hypo- thesis and each of the sub -hypotheses. (3) Finally, an examination of the relative consequence or importance of the RCEB component variables in relation to self identification was made by means of Guttman's formal model for scale analysis. No association was found for the bivariate relationship of RHOMOG and religious self identification. Associations of .16, .25, and .42 were found for the bivariate relationships between MIG, RBEL, RPART, and religious self identification, respectively. The SI % Index indicated that 55, 59, and 72 per cent of all cases were accounted for in the predicted cells for the three respective bivariate relationships. The RCEB component variables were then examined in various combinations in relation to the dependent variable of self identi- fication. The major hypothesis asserted a direct relationship between the extent of persons' implication in a religious context of experience and behavior and the liklihood of those persons identifying themselves as religious. The association between RBEL-together-with-MIG and religious self identification was V = .28 and the SI % Index was 61 per cent. The association between RPART -together-with-MIG and religious self identification was V = . 36 and the SI % Index was 73 per cent. The association between RBEL-together-with -RPART and reli- gious self identification was V = . 39 and the SI % Index was 74 per cent. The association between REEL-together -with- RPART -together-with- MIG and religious self identification was V = . 41 and the SI % Index was 77 per cent. By use of Guttman's scale analysis technique RPART was determined to be the most consequential of all the RCEB components considered simultaneously. Consistent with the obtained pattern of relationship in both bivariate and multivariate cases as presented above, RPART was followed in importance by RBEL and MIG in terms of placing persons in a position where others could address religious identifying behaviors toward them resulting in their identification of themselves as religious. Both the deviant cases and the lack of relationship between RHOMOG and religious self identification were examined and discussed. While sufficient additional data were not available to account for the deviant cases, an exploratory line of analysis is suggested and discussed which is consistent with both the theoretical orientation and with the pattern of relationship consistently observed in the entire study. Modifications are made in the theoretical orientation to take into account the lack of observed relationship between RHOMOG and religious self identification SELF IDENTIFICATION WITHIN A SPECIFIC CONTEXT OF EXPERIENCE AND BEHAVIOR by ‘ l f . . I' \ a l S . Clark Mc Phail A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOC TOR OF PHILOSOPHY Department of Sociology 1965 9 AC KNOW LE DGEMEN TS Several academicians have played a significant part in my development as a student of human behavior. Charles Browning, then of Pasadena College, suggested the sociological enterprise to me as a potentially rewarding challenge. Norman Jackman and Jack Dodson. then of the University of Oklahoma, initiated the fulfillment of that challenge with their sociological perspective and their encouragement of my efforts. Fred Waisanen played a crucial role in the early stages of my learning experience at Michigan State University. Carl Couch and Bob Stewart were especially significant with their challenging questions and penetrating criticism of the ideas I was fortunate to discuss with them. Professorial figures, however, cannot begin to exhaust the list of persons to whom I owe a debt of gratitude at this point in my career. I owe a great deal to the friendship, challenge and con- structive criticism of several peers in my educational experience: 'to an undergraduate friend, Tom Whitsett; to a fellow M.A. candidate, Frank Sampson; and especially to my fellow sociologists at Michigan State University--Dick Brymer, Eugenio Fonseca, and Chuck Tucker. The debts incurred in the selection, construction and com- pletion of a doctorial thesis project are many. I wish to thank Charles P. Loomis for the use of the data from which the study was constructed. I am grateful to my committee members, Professors Artis, Couch, Faunce, Kumata, McKee and Stewart for their participation in a variety of ways in my doctoral studies and particularly for their participation in the examination process. I am grateful to my wife for her typing and editorial assistance; for her patient and understanding cooperation in ii deflecting her own and our children's activities during the lengthy period required to complete the thesis; and, for her encouragement and support during both the bleak and the bright days of the thesis preparation. Finally, I wish to thank Bob Stewart, mentioned above as professor and committee member, whose able criticism, unlimited patience and friendship played a crucial role in my own intellectual development over the past two years. As chairman of my committee his penetrating criticism revealed many flaws in my work. The consequence of this was the reformulation or resolution of many issues for the both of us and the construction of more questions than I shall be able to resolve in my sociological career. Fortunately, his critical activity and its consequence is matched by his activities of laughing at those erroneous tacks which may be corrected with profit, setting aside those issues which are irrelevant to the enterprise at hand, and, having the patience to allow his students to make the distinction between error and irrelevance. To be consistent with the theoretical orientation within which I have developed this thesis I must assert that both the merit and error of the ideas in this work are products of my relations with those persons named above. I have attempted to take their activities into my own conduct and while the final translation of their ideas rests ’with my constructed activities, the responsibility remains social. iii TABLE OF CONTENTS Chapter I . THE PROBLEM: SELF IDENTIFICATION AND THE CONTEXT OF EXPERIENCE AND BEHAVIOR Introduction . . An Overview of G. N. Mead's Position on Self Other Relationship . . Specific Focus of the Present Study A Review of Previous Investigations Relevant to the Present Study . Proposal of Research Focii to Resolve Inconsistencies in Previous Studies Summary of Chapter I . Outline of Remainder of Dissertation II . THEORETICAL ORIENTATION . Introduction . . G. H. Mead's Position on the Social Self A Formalized Statement of the Median Theory of the Social Self . , Proposition I ‘ Proposition II Proposition III. Proposition IV. . . . An Explication of the Construct--Context of Experience and Behavior Summary of Chapter II III. METHODS OF INQUIRY: A niisCUSSION AND SPECIFICATION OF THE OBSERVATICNAL AND ANALYTICAL PROCEDURES . Introduction . . Statement of Hypothesis . . . . Discussion and Definition of Major Concepts in the Hypothesis iv Page 20 23 24 25 25 25 36 4O 4O 42 42 43 44 66 68 68 68 69 Chapte r Page The Specification of Operational Procedures Versus Operational Definitions . . . . 70 Preliminary Description of Operational Procedures for Making Observations of Referents for Concepts in the Major Hypothesis . . . . 72 Restatement of Major Hypothesis and Statement of Sub— -Hypotheses . . . . 75 Discussion of the Issue of the Validity and Reliability of Procedure of Inquiry . . . 76 Specification and Discussion of Operational Procedures for Observing Referents of the Major Concepts . . . . 96 Discussion of the Sample and the Issue of Sample- -Universe Relationships . . . . 119 Analysis Procedures . . . . 124 Operational Restatement of Hypotheses . . 138 Summary of Chapter III . . . . . . . 141 IV. TEST OF HYPOTHESES . . . . . . . 144 Introduction . . . . . . . 144 Statement of Hypotheses in Operational From 144 Test of Hypotheses on the Original Sample . 145 Test of Hypotheses on Reduced RCEB Sample 152 A Re -examination of the Theory with Respect to RHOMOG. . . . . . 162 Test of Hypotheses on RCEB' Sample . . . 167 Test of Major Hypothesis . . 173 Imposition of the Three RCEB Components on Guttman' 3 Formal Model for Scale Analysis . . 187 Examination of Deviant Cases in MAXRCEB and MINRCEB . . . . . . 193 Summary and Conclusions for Chapter IV . 198 V. REVIEW, CONCLUSIONS AND PROPOSALS FOR FUTURE INVESTIGATION . . . . 207 Introduction . . . . . . . . . . 207 The Problem . . . . . . . . . . 207 The Theoretical Orientation . . . 207 Relation of Observations to the Hypotheses . 215 Proposals for Future Research . . . . 221 Table 10, ll. 12. l3. 14. LIST OF TAB LES Rank Order of Denomination Religious Preference by Religious Self Identification Statements in Three Studies RHOMOG and Self Identification . MIG and Self Identification RBEL and Self Identification RPART and Self Identification Summary of Bivariate Associations between RCEB Components and Self Identification for the General Sample and Interassociations for the RCEB Components . :RHOMOG and Self Identification MIG and Self Identification RBEL and Self Identification RPART and Self Identification Summary of Bivariate Associations between RCEB Components and Self Identification and Inter- associations for the RCEB Components for the Reduced RCEB Sample MIG and Self Identification RBEL and Self Identification RPART and Self Identification vi Page 16 146 147 148 149 157 153 154 155 156 158 168 169 170 Table 15. l6. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. Summary of Bivariate Associations between RCEB Components and Self Identification, and, Inter- associations for RCEB Components for the RCEB' Sample RBEL-together-with-MIG, and, Self Identification RPART-together-with-MIG, and, Self Identification RBEL-together-with-RPART, and, Self Identification RBEL -togethe r -with -RPART -togethe r -with -MIG, and, Self Identification Summary of Bivariate and Multivariate Association Between RCEB Components and Self Identification for the RCEB' Sample Rank Order of Importance for Six Order Combinations of RCEB Components Based on C.R. and I.O.C. Values Response Patterns for the RPART -RBEL-MIG Order Combinations RCEB Scale Types for Order Arrangement RPART- RBEL-MIG and Self Identification Deviant Cases in MINRCEB: Upper Six Values of RPART Index and Self Identification . Deviant Cases in MAXRCEB: Upper Six Values of RPART Index and Self Identification . Deviant Cases in MAXRCEB: Median Split of Upper Six Values of RPART Index and Self Identification . Page 172 174 176 177 180 183 188 190 191 195 195 197 LIST OF FIGURES Figure Page 1. Hypothetical Contingency Table for RHOMOG and Self Identification with Designation of MAXCREB and MINRCEB Cells . . . . . . . . . 126 2. Hypothetical Contingency Table for RHOMOG- together-with-MIG, and, Self Identification with Designation of MAXCREB and MINRCEB Cells . 130 3. Half-Parallelogram Pattern for Guttman Formal Scale Model . . . . . . . . . . . 133 4. Half-Parallelogram Pattern for Guttman Model with Various Item Response Patterns and Error Scores............. 135 viii LIST OF APPENDIC IES Appendix Page A. Self Identification Problem: Interviewer Instructions for Administering Problem and Recording Obser- vations, and, General Coding Instructions for Statements Made in Response to the Question, "Who Am I?" . . . . . . . . . . . 225 B. The Use of the Counter-Sorter for Scalogram Analysis . . . . . . . . . . . . . 233 ix CHAPTER I THE PROBLEM: SELF IDENTIFICATION AND THE CONTEXT OF EXPERIENCE AND BEHAVIOR Introduction This study has been designed to obtain crucial information regarding a fundamental social psychological question: "What are the basic processes which produce and maintain persons' conceptions and identifications of themselves?" The points of departure for the study were two: first, the inquiry was derived from the theoretical orienta- tion of George H. Mead who sets the question of self conception and identification within the larger context of self-other relations;1 and. second, the inquiry specifically focused on a single class of self iden- tifications and the experiential base on which that class of behaviors should, theoretically, re st . An Overview of G. H. Mead's Position on Self-Other Relationship Briefly considered, Mead views the self as the class of activities or behaviors which the person directs toward himself as an . Z ‘ object. Conception of self or self identification is that behavior which the person takes with respect to himself as an object. Mead argues that 1 George Herbert Mead. Mind, Self and Society: From the Standpoint of a Social Behaviorist, ed. by Charles W. Morris. University of Chicago Press, Chicago, 1954. 21bid., pp. 113-142. conceptions of an object or identifications of an object are behaviors directed toward that object, i.e. , conceptions ar_ebehaviors. 3 Objects have no intrinsic definition or meaning. The meaning which we say objects “have, " and consequently the conceptions that are "held" of those objects, stem from identifying behaviors directed toward those objects by persons. When similar identifying behaviors are directed toward an object by persons, we say that those persons have common meanings for or conceptions of that object. Thus, the conception of an object is the identifying behavior directed toward it by the person or persons in question. Mead asserts that the self is an object like all other objects. The conception of that object must, then, come about in the same manner as with all other objects; viz. through the behaviors of others which identify that object.4 Thus, with conceptions viewed as behaviors, the person will conceive of himself and identify himself as others con- ceive of and identify him; the person will behave toward himself as others behave toward him. It is in this sense that Mead's oft -quoted statement about self and others must be interpreted: The individual experiences himself as such, not directly, but only indirectly, from the particular standpoints of other indi- vidual members of the same social group, or from the general- ized standpoints of the social group as a whole to which he belongs. For he enters into his own experience as a self or individual, not directly or immediately, by becoming a subject to himself, but only in so far as he first becomes an object to himself just as other individuals are objects to him or in his experience; and he becomes an object to himself only by taking the attitudes of other individuals toward himself within a social environment or context of experience and behavior in which both he and they are involved. 3 Ibid., pp. l-18, and, p. 105. Thus, according to Mead, all objects must be social before they can be physical. 5Ibid. , p. 138. Self conceptions and identifications, then, are created, main- tained, and changed by the experiences produced through the activities others address to the person. To whatever extent these activities are repeated and consistent, the person's conception of self should be con- sistent. To whatever extent the activities of others addressed to a person change over time, changes in self-identification should occur. To whatever extent there is variety in the behaviors others address to the person, variety in self conceptions should be found. Mead states: We carry on a whole series of different relationships to different people. We are one thing to one man and another thing to another . . . .We divide ourselves up in all sorts of different selves with reference to our acquaintances. We discuss politics with one and religion with another. There are all sorts of different social reactions. It is the social process itself that is responsible for the appearance of the self; it is not there as a self apart from this type of experience. 6 The person's conception of self, which is his self, can have as many facets or aspects as there are consistent relationships with and iden- tifications by others. Thus the person's general conception and identification of self is viewed as the result of the different identifica- tion behaviors which are directed toward the person by others. On this latter point Mead states: The unity and structure of the complete self reflects the unity and structure of the social process as a whole; and each of the elemen- tary selves of which it is composed reflects the unity and structure of one of the various aspects of that [social] process in which the individual is implicated. 7 Consequently a person may conceive of and identify himself as "a father, " "a husband, " "a junk dealer. " "a Catholic, " "a friend, " and "a man with very little chance for success in life. " Mead's orien- tation would suggest that each of these various self identifications stem from the kinds of identification behaviors which are directed toward the 6Ibid., p. 142. 7 Ibid., p. 144. person by those with whom he is implicated in the various contexts or arenas of interaction. In outlining the above, only the central and most relevant elements of the theoretical orientation have been noted. I would not hold, nor would Mead, that self conceptions are mere images of other's actions. 8Nonetheless, the point must be made without quali- fications that self conceptions and identifications are contingent upon and systematically related to the activities of others. The preceding discussion has attempted to sketch in the outlines of Mead's position with respect to the relationship between the activities of others, and the activities which the person directs toward himself; the latter set of activities referring to Mead's conceptualization of the self. Specific Focus of the Present Study The second point of departure for the present study, men- tioned above, was the focus on a single class of self conceptions and identifications, and, the specific experiential base on which that class of behaviors should rest. I have previously indicated that various con- texts of interaction should, according to the orientation, result in cor- responding classes of self conceptions. This study will deal with only one context of interaction; viz. , that of religion. By virtue of this delimitation, the general question of the processes which produce and maintain persons' conceptions of them- selves is reduced to the specific question of the processes which pro- duce and maintain persons' conceptions or identifications of themselves as religious. Accordingly, the focus will be on persons' religious Mead views all self conscious conduct as interpretive response and not as the automatic reproductions of behavior in response to some stimulus object. See his discussion of the stages in the social act, pp. 3-25; his discussion of consciousness, p. 75; and his comments on the process of reflection, p. 89--all in Charles W. Morris (ed. ), The PhilosophLof the Act, Chicago: University of Chicago Press, 1938. activities and experiences with others in relation to their self concep- tions or identifications of self as religious. Thus while religious activities and a "religious context of experience and behavior" will be used to investigate the general theo- retical question, the concern of the study is n_otwith religion or reli- gious self identification per se. This particular set of activities and behaviors merely furnishes a convenient context within which the investigation of self identification can take place. In the course of that investigation, obviously, considerable information about religious activities should be forthcoming. Therefore, the present study will attempt to deal with three issues: (1) it will be an attempt to test a portion of a theoretical orien- tation and in so doing to answer, at least in part, a question about the basic processes which produce and maintain persons' conceptions and identifications of themselves; (2) it will provide empirical informa- tion about self identification; (3) it will provide empirical information about religious activity and experience. A Review of Previous Investigations Relevant to the Present Study Introduction The Meadian orientation, within which the present investi- gation is couched, has been frequently chastized for generating more criticism, and often more mysticism, than empiricism. Robert Merton once asserted that Mead "was not exposed to systematic empir- ical evidence. " He further went on to imply that neither Mead nor Meadians were interested in exposing themselves to such evidence with the following comment: "Mead and those of his followers who also eschew empirical research, had little occasion to move ahead to the Tie stion of conditions under which non -membership groups may also constitute a significant frame of reference" with respect to the development of the social self.H9 Such remarks may have been an appro- priate commentary 15 years ago. The commentary on empirical pro- ductivity is both inappropriate and inaccurate today. A number of studies have been made of self-other relationships; a central concern in Mead's consideration of human conduct. Many of these investigations have been explicitly formulated within the framework of Mead's orien- tation and have consistently supported the fundamental assertions of the orientation; viz., the self is not independent of others and self con- ceptions are systematically related to the activities of others and the activities in which the person is implicated with others . 9 R. K. Merton, Social Theory and Social Structure, Free Press, Glencoe, revised edition, 1957, p. 239. 10See here a number of doctoral dissertations in the Depart- ment of Sociology at the State University of Iowa under the direction of Manford H. Kuhn: T. S. McPartland, "The Self and Social Structure, “ 1953; F. B. Waisanen, "The Prejudice Variable: A Social Psychologi- cal and Methodological Study, " 1952; R. L. Stewart, "The Self and Other Objects: Their Measurement and Interrelationship, " 1955; C. J. Couch, "A Study of the Relationships between Self-views and Role -taking Accuracy, " 1955; W. S. Garretson, "College as a Social Object: A Study in Consensus, " 1961. A number of other studies have been published stemming from the "Iowa School” and these include: M. H. Kuhn and T. S. McPartland, ”An Investigation of Self Attitudes, " American Sociological Review, Vol. 19 (February, 1954), pp. 68-76; C. J. Couch, “Self Atti- tudes and Degree of Agreement with Immediate Others, " American Journal of Sociology, Vol. 63 (March, 1958). pp. 491-496; T. s. McPart- land, "Self Conception, Social Class and Mental Health, " Human Organ- ization, Vol. 17 (1958), pp. 24-29; M. H. Kuhn, "Self Attitudes by Age, Sex, and Professional Training, " Sociological Quarterly, Vol. 1 (January 1960), pp. 39-55; T. S. McPartland, J. Cumming, and W. Garretson, "Self Conception and Ward Behavior in Two Psychiatric Hos- pitals, " Sociometry, Vol. 24 (June, 1961), pp. 11-24; G. Vernon, "Religious Self Identification, " Pacific Sociological Review, Vol. 5 (Spring, 1962), pp. 40-43; W. Garretson, "The Consensual Definition of Social Objects, " Sociological Quarterly, Vol. 3 (April, 1962), pp. 107 -113; H. Mulford and W. Salisbury, "Self-Conceptions in a General Population, " Sociological Quarterly, Vol. 5 No. 1 (Winter, 1964), pp. 35-46. These studies have demonstrated that the general orienta- tion is researchable, that the specific problem of self-other relatiOn- ships can be researched, and that there is some general confirmation for the asserted relationship between self conception and the activities of and with others. Nonetheless, the specifics of the relationships between self conception and the activities of and with others have not consistently been drawn into sharp relief. The present study will attempt to approach that end. Among the studies dealing with self conception, three are of particular relevance to the present investigation. They deal with the explicit relationship of self conception, religious affiliation, and reli- gious activities. The treatment of this relationship, the results, and the interpretation of the results in these three studies provided the basis for initiating the investigation under discussion. I am referring here to Kuhn and McPartland's research on the relationship between reli- gious "self attitudes” and religious affiliation.11 Vernon's research on the relationship between religious self identification and religious . . . 12 . . . . affiliation; and, the present writer's preliminary Investigatlon of the Additional studies, falling within the Meadian orientation and dealing with self-other relationships include: Frank Miyamoto and San- ford Dornbusch, "A Test of Interactionist Hypotheses of Self-Conception, " American Journal of SociologL Vol._ 61 (March, 1956), pp. 399-403; S. Dinitz, A. R. Mangus, and B. Pasamanick, "Integration and Con- flict in Self-Other Conceptions as Factors in Mental Illness, " Sociometry, Vol. 22 (March, 1959), pp. 44-55; R. Videbeck, "Self- Conception and the Reactions of Others, " Sociometry, Vol. 23, No. 4 (December, 1960), pp. 351-359. 1 1Kuhn and McPartland, op. cit., 1954. 12Vernon, o . cit., 1962. relationship between religious self identification, religious affiliation, and religious activities . All three of these investigators utilized a technique, devel- oped by Kuhn, for observing self identifying behaviors.14 This tech- nique is known as the ”Twenty Statements Test" (hereafter referred to as the TST) and consists of asking the respondent to make twenty statements in response to the question, "Who Am 1?". This technique will be discussed at length in Chapter III but the above description will suffice for our present purposes. It should be noted, however. that the responses to the question "Who Am I?" yield such statements as "I am a man, " "I am a Christian, " "I am a husband, " "I am a junk dealer, " "I am a Catholic, " "I am a poor person, " "I feel all alone in this world, ” etc. A brief comment regarding these statements is appro- priate at this point. Whether they are written or verbal replies to the question "Who Am I?" they are considered by the present investi- gator as behaviors. Furthermore, replies to the "Who Am I?" ques- tion may be legitimately classified as identifying behaviors. The legitimacy of the classification will be considered at length in a sub- sequent section of this dissertation. These identifying behaviors, or responses to the question "Who Am I?",were treated by the several investigators mentioned above in a variety of ways. Despite this differential treatment, each employed the simple technique of noting the presence or absence of religious statements, i.e., statements about religious affiliation, reli- gion or religious activities. This similarity in the technique of classi- fying these behaviors is directly relevant to the present discussion of 13 Clark McPhail, "Religious Preference, Religious Acti- vity and Religious Self Identification, " unpublished study, Department of Sociology, Michigan State University, 1963. 4 Kuhn and McPartland, op. cit., 1954. the relationship between a specific context of experience and behavior, and, self identification. The relevance will become clear below as I discuss each of the studies and the theoretical interpretation imposed on the relationships obtained in those studies. The Kuhn -McPartland Stug Kuhn and McPartland hypothesized that persons belonging to minority religious groups would make reference to their religious affiliation earlier in the course of their twenty responses to the TST than would members of majority religious groups.15 They based this argument on the notion that minority religious group affiliates have in common the charaCteristic that religion is "important", "of conse- quence", and "differentiating. " Conversely, they suggested that ma- jority religion affiliates have in common the characteristic that reli- gion is "of little consequence, " “not important, " and "taken for granted. " To test their hypotheses, they took the mean rank position (in the twenty responses to the TST) of statements mentioning reli- gious affiliation, for each religious preference group. They found that minority religious group affiliates identified themselves in terms of . . . . . . . 16 . their religious affiliation more saliently, i.e., earlier in the course 15 Ibid. They classified Catholics, Lutherans, "Small Sects" members, Congregationalists, and "Christians" as minority religious groups, and Methodists and Presbyterians as majority religious groups. Minority religious groups were defined as those whose"”. . . subcul- tures contain norms which set their members at odds with the norms of the larger society"" p. 73. 16 . . Kuhn and McPartland discuss salience as follows: ""The salience of a reference may be understood as the relative spontaneity with which a particular reference will be used as an orientation in the organization of behavior. "" Ibid. , p. 74. Salience of religious reference in "self attitudes" was measured by the rank of religious reference (if made) in the list of twenty responses. If a religious reference was made in the first statement the respondent was given a score of 20; if men- tioned in the last statement a score of 1; if no reference to religious 10 of the twenty responses, than did majority religious group affiliates. Kuhn and McPartland argued that the religious group serves as a reference group for the minority religion affiliates but does not serve that function for majority religion affiliates. To further support this notion of "religious reference group identification" Kuhn and McPartland examined the relationship between minority-majority religious group affiliation and the mention or non-mention of such affiliation in response to the question, "With what groups do you feel most closely identified?" They found that minority religion affiliates mentioned their religious group affiliation in responding to this "reference group question" more frequently than did majority religious group members.18 They concluded their remarks on these findings with this interpretation: affiliation was made a score of zero was arbitrarily as- signed. Ibid. 17 . These authors took the mean salience score of respon- dents by denominational preference and then ranked the denominational groups by their mean score. Using the Catholic group mean as a base, no significant differences were found between that base and Lutherans, "Sects, " Jewish, or Congregationalists, (all minority religious groups by their classification.) Significant differences were found (p. . 01) between Roman Catholics, as a base, and Presbyterians, Methodists, and "Christians, " using what the authors report as an "analysis of variance" test. (Presbyterians and Methodists were classified as ma- jority religious groups.) I assume that for “analysis of variance" these authors used a simple "t test" for significance of difference between group means, with Catholics as the base mean group. With a dichotomous division of religious affiliation, Cath- olics-Jews, and, All Others, against presence or absence of religious affiliation mentioned, they obtained a x2 = 17.03, p. . 0001. With a trichotomous separation of religious affiliation, Catholics-Jews, "Small Sects, " and "Large Denominations, " they obtained a x2 = 19.45, p. .0001. 11 Religious affiliation references are significantly more salient among the self-attitudes of members of "differentisticH religious groups than among members of "majority" or conventional reli- gious groups. Corroboratively, the religious group as a reference group appears far more frequently as an answer to a direct, reference-group type of question among those made by members of "differentistic" religious groups.19 The Ve rnon Study Vernon's study was a modified replication of the Kuhn- McPartland study reported above. He attempted to extend their notion that all religious groups are not equally productive of religious self identifications.20 Vernon, like the previous investigators, contended that minority religious group affiliates would have a higher degree of "identification" with their religious groups and religion and would thus identify themselves on the TST in terms of that reference group orientation. In his sample he found that minority religious groups (Mormons and Roman Catholics), had a higher percentage of affiliates who made religious self identification statements in response to the TST than did majority religious groups, (Methodists, Baptists, Disciples of Christ, Congregationalists, Lutherans, Presbyterians and Episcopa- lians) . Vernon suggested that both Mormonism and Catholicism are "demanding" religious orientations, requiring strong religious commitments from their affiliates. He contended that both of these groups are more ”differentistic" and less tolerant of other religious points of view than are majority religions. Consequently he concluded that the ”significant others" of minority religious group (Mormon and 19Ibid. , pp. 75-76. 20 Vernon. ML 12 Catholic) respondents ". . .probably emphasize religious identification more strongly than in the case of the other boys (majority religious group respondents). " These findings, then, would seem to reflect the fact that groups which differ in some significant manner from the mainstream of society, if they are to persist, must have members for whom the group is a sufficiently strong reference group that these dif- ferences beco e incorporated into the members self structure. The results offhis study provide evidence of the relative effec- tiveness of the indoctrination efforts of these two high-identity groups.22 Previous Investigation by Present Writer The present writer also examined this asserted relationship between religious affiliation and religious self identification?3 The rank order of religious groups, in terms of per cent of affiliates making religious self identifications, provided no corroborative support for the studies of Kuhn-McPartland and Vernon. The present writer's data indicated that "Small‘Sects" and Roman Catholic affiliates (minority religious groups in the two previous studies) ranked fifth and sixth, respectively, in terms of per cent of members making religious self identification statements in response to the TST. Lutherans, Episco- palians, Baptists and Presbyterians ranked first through fourth, in that order, in terms of the per cent of affiliates making religious self identification statements . The present writer collected additional data on the reli- gious activities of respondents in this preliminary study. Respondents were asked to state the frequency with which they asked themselves "what God would have them to do" when faced with everyday decisions. The chi square test for independence between those who did and did not 2 1Ibid.. PP. 42. zzlbid. 2 3McPhail, op . cit . 13 make religious self identification statements yielded a value of x2 = 17.48, p. . 001, and thus precludes accepting the null hypothesis of no association between religious self identification and the frequency of this class of religious behaviors. Cramer's V, a measure of asso- ciation based on x2, indicates an association between religious self identification and the frequency of this religious activity of .33 Here, then, is preliminary data indicating a relationship between one class of religious activities and religious self identification. In addition, respondents were asked to state the extent of their subscription to five statements of religious belief.25 The chi 2 square test for independence yielded a value of x = 12. 94, p. . 01. 4The null hypothesis states there is no association, i.e.. independence between the two groups (those who did and who did not make religious self identification statements) with respect to the fre- quency of their performance of this particular class of religious acti- vity. The "frequency" categories were "often, " "seldom, " and ”never. " Cramer's V is discussed in Hubert M. Blalock Jr.'s, Social Statistics, N.Y.: McGraw Hill, 1960, p. 230. Blalock states: ""Although V is not commonly used in the social science literature, it seems to be preferable to Tschuprow's T and also to Pearson's con- tingency coefficient C in that it can attain unity even when the numbers of rows and columns are not equal. ”" Ibid. . p. 230. 25These five items, which will subsequently be referred to as the Statements of Religious Belief Index (RBEL), were originally developed by Snell Putney and Russel Middleton and reported in their "Dimensions and Correlates of Religious Ideologies, " Social Forces, Vol. 39, No. 4 (May, 1961), pp. 285-290. The present writer submitted the items to Guttman Scalogram Analysis and determined that the five items formed a Guttman Scale with a C. R. = . 97. The five items are: “I believe there is a real hell where men are punished for their sins;” "I believe God has a plan for every person's life;" "I believe in a life after death;" "I believe there is a devil who tries to make men sin;" "To me the most important work of the church is saving people's souls . " Each of the items had five answer categories; "strongly agree, " "slightly agree, " "don't know, " "slightly disagree," and ”strongly disagree. " Thus, with a score ranging from one to five on each item, the total index Score ranged from 5 to 25. The reader will note, however, that the items Were scored slightly different for the study to be reported in this disser- tation. A discussion of the scoring arrangement will be reported in Chap- te 1' III of the present report. l4 Cramer's V indicates an association between the extent of the person‘s subscription to statements of religious belief and religious self identi- fication of .-2_4.26 The data therefore indicated a relationship between this additional class of religious activities and religious self identifica- A tion. The null hypothesis states there is no association, i.e., independence, between the two groups (those who did and who did not make religious self identification statements) with respect to the extent of their subscription to the items in the Statement of Religious Belief Index (RBEL) . The null hypothesis was rejected, at the . 01 level, and the association reported is based on the contingency table association between quartile rank scores on the total score range of 5 to 25 for the five items, and, presence or absence of religious self identification statements. It is extremely important to indicate at this juncture that these "statements of religious beliefs" and "consideration of God's will ineveryday decisions" are conceptualized here as religious activities. They are not viewed as indirect measures of varying degrees of reli- giosity attitudes, inside the person, which predispose the respondent to behave "religiously" in the data collection situation or any other sit- uation. Rather, they are viewed as observations of religious behaviors taken by the respondent during the data collection situation and /or reports by the respondent of religious activities in which he engages. This is a critical issue in terms of this earlier study as well as the study to be discussed in the succeeding pages. I will maintain the position, throughout the dissertation, that the scientist, in testing hypotheses, is engaged in the process of submit- ting to test a particular kind of propositional statement; viz., one con- sisting of an asserted relationship between data statements or observa- tions of behaviors. Central to this position is the contention that a verbal statement, made by a respondent in a data collection situation, or any other situation, is as much an observable behavior, and one which is amenable to various counting procedures, as is kicking a football or blowing one's nose. Thus the verbal statements made by respondents in the data reported above, and in the data to be reported in the present investigation, will be treated as observable behavioral units or classes of activities. In the case of both studies they happen to be classes of reli- gious activities or behaviors. 15 A Comparison of Findings from Previous Studies In summary, the three studies discussed above report data on relationships between religious affiliation, religious activities, and religious self identification. All the investigators either couched their study or interpreted their data in terms of a Meadian orientation. Kuhn -McPartland and Vernon attempted to relate their data to that orien- tation contending that the relationships which persons have with others, in this case group affiliation and group “identification" produce their self conceptions. In both of these studies data were interpreted as supporting the argument that "minority religious group" affiliates make religious self identification statements (or have religious "self- attitudes") due to the fact that such groups are "differentistic" in our society and that, consequently, their members view their religion as important and consequential, and in turn identify themselves as reli- gious. These persons are contrasted with "majority religious group" affiliates for whom religion is seen as being of little consequence or im- portance. On the other hand, data collected by the present writer failed to confirm the notion of differential religious self identification based on affiliation with minority or majority religious groups. The discrepancies between religious preference groups for the three studies are reported in Table 1. Examination of the table readily reveals differences in the rank order of religious groups in the three studies with respect to affil- 28 , iates making religious self identifications. While there is some 28R should be pointed out that the rank order of religious groups are strictly comparable only in the case of the Vernon and the McPhail data. In those two studies rank order was based on the percen- tage of religious group preferents who made religious self identification statements. It should also be pointed out that religious self identification included statements referring to religion or religious activity as an object as well as statements of religious affiliation. 16 TABLE 1. --Rank order of denomination religious preference by religious self identification statements in three studies. Kuhn- Denomination McPartland Vernon Mc Phail Rank‘i‘ N Rank’i‘ * N Rank”< * N Roman Catholic 1 (38) 2 (258) 6 (39) Lutheran 2 (33) 7 (93) l (14) Small Sects 3a (20) - -- 5b (11) Jewish 4 (l9) - -- 8 (16) Congregational 5 (l3) 6 (58) 7 ( 15) Presbyterian 6 (32) 8 (153) 4 (20) Methodist 7 (73) 3 (148) 10 (39) "Christian" 8 (11) 5 (51) -- -- Mormon - -— l (44) -- -- Baptist - -- 4 (49) 3 (8) Episcopalian - -- 9 ( 36) 2 ( 12) Interdenomination - - - - - - 1 1 ( 15) None - - - - -- 9 ( 16) Agnostic - -- - -- 12 (7) Total - 239 - 890 212 *Rank order in the Kuhn-McPartland study was based on the denominational group mean salience of mention of religious affiliation; e. g. , "I am a Roman Catholic. " **Rank order in both the Vernon and McPhail studies was based on the percentage of denominational preferents making religious self identification statements of any kind. This would include ”I am a Roman Catholic, " as well as "I am a good Christian, " "I go to church regularly, " and ”I am a religious person. " a"Small Sects" in the Kuhn -McPartland study included Bap- tists, Episcopalians, Evangelicals, Mennonites, Reorganized Latter Day Saints (Mormons) and Unitarians. b”Small Sects" in the McPhail study included: Evangelical, Christian, Christian Science, Church of Christ, Church of God, Free Mfithodist, United Missionary Alliance, Christian Reform, and Nazarene. Wlth perhaps the exception of Christian, the "Small Sects" classification W38 based on a national numerical minority and not on a doctrinal minor- 1t¥ position or a numerical minority in the sample. National numerical mlr'Ilority was determined by the total national denominational membership has ed on U. S. Census and NatIOnal Council of Churches figures. l7 discrepancy between the Kuhn-McPartland and Vernon studies as to what constitutes a minority religious group, Vernon maintains that his data bear out the argument and data in the earlier study; viz. , that In the Kuhn-McPartland study rank order of religious groupings was based on the mean salience score of making a religious affiliation reference. See footnotes 15 and 16 above. Despite the slight differences in the nature of the data ob- tained between the Kuhn-McPartland and Vernon studies they are still comparable in terms of the argument which both sets of investigators used to interpret their data. Their argument is that persons affiliated with minority religious groups will make religious self identification statements because their group serves as an important and consequential reference group for them as contrasted with majority religious group affiliates. Consequently the slight discrepancy in the nature of the data on which the religious groups were ranked does not make a comparison of those rankings irrelevant or invalid. The complete data on which the rankings were established is given below in Table l-A. 18 - -- NS - -- cow - -- omm :38. NH o.oo C; I II II I II II owumoawafl o n.2, $3 - -- -- - -- -- 282 HH 0 .ON Amm v I I I I I I I I I I COMHNGMEOQOUHQHGH m ooo 3: o Ag: 33 - -- -- donoaoooaam m m.~o 2: a. mom a: - -- -- oasaom .. -- -- a 5:. $1 - -- -- 8882 II II II m ©..—N AHmv w Nw.H Amav :Gwmumwhfinoz S mime an. m 0.2 So: A -.m at oaeoofioz a. 9mm 8.: m of an: o Ste 32 cocooaoooom e n.3, 3: o new 83 m $5 3: Hoeosomoomooo w e on $3 - .. .: o So 8: sensed, m elem :3 - -- -- m we; 33 :oooom :38: 2 wine 3.: A atom 83 m dog 32 aeoofioq o 93 33 m nos 3...“; 2 am.» 82 osofioo oefiom W No % Z W No % Z woo W X. Z coflumfiaocofl u H N u H N u U. S X I e N. I e y. on m P M P M m. x m u W U n 9 n... as e m. mm mm mm mm. Edie: men. doooo> m. m exam 0 m. o m. a a u n u n u o S 8 MW T: .ther" freeing the person from the rule of a situational chameleon. 3 . . . . I would not attempt to minimize the importance of other per- SOns with whom the person must contend in the social situations in which they find themselves. Thus I do not mean to imply in the text above or in the previous footnote that the others with whom the person must contend in the situation are of no importance. Nothing would be less consistent with Mead's orientation. However, Mead would argue that the others taken iIlto account in the person's construction of his activities in a situation are not limited to just the persons in the situation although they are, obviously, of consequence in this construction. It would seem to me that Mead's use of the notion of "generalized other" and his emphasis on the relationships between people rather than "situations, " per se, are more useful approaches than the extreme situational emphasis that is found in some of the more I‘ecent dramaturgical models. Perhaps the most well known of these situa- tiOnal models is Erving Goffman's, The Presentation of Self in Evepydai $3 Garden City, N. Y.: Doubleday Anchor Book, 1959. Of course Gof- fl'nan admits that his dramaturgical model must emphasize the situation and situational others, almost to an extreme, due to the fact that it is pri- 1“harily applicable to the interaction between persons who are relative Strangers. See p. 222. 33 The self-conscious human individual, then, takes or assumes the organized social attitudes of the given social group or community (or of some one section thereof) (to which he belongs . . . [and] . he governs his own conduct accordingly. Thus, the self emerges or arises and develops only in rela- tion to other persons, by assuming the organized behaviors of others and . by thus becoming an individual reflection of the general systematic pattern of social group behavior in which it and the others are all involved. "15 Given this orientation, it is not surprising that Mead could not conceive of the self as a permanent or independent entity. He con- ceived of the self as developing and continuing only in relation to the other' 5 with whom the person is implicated in the various social processes or different "contexts or experience and behavior. " Selves can only exist in definite relationships to other selves . . . The individual possesses a self only in relation to the selves of other members of his social group; and the structure of his self expresses or reflects the general behavior pattern of this social group to which he belongs, just as does the structure of the self of every other individual belonging to this social group. 19 Mead specified two categories of "socially functional classes of subgroups" to which the person could belong. Some of them are concrete social classes or subgroups such as political parties, clubs, corporations, which are all actually functional social units, in terms of which their individual mem- bers are directly related to one another. The others are abstract social classes or subgroups, such as the class of debtors and the class of creditors, in terms of which their individual members are related to one another only more or less indirectly, and which only more or less indirectly function as social units, but which afforded or present unlimited possibilities . . . The given individual's membership in several of these abstract social classes or subgroups makes possible his entrance into definite social relations (however indirect) with an almost indefinite number of o‘the r individuals who also belong to or are 14 . Mead, op. c1t., p. 156. 1 5Ibid., p. 158. 16 Ibid. , p. 164. 34 included within one or another of these abstract social classes or subgroups cutting across functional lines of demarcation which divide different human social communities from one another, and including individual members from several (in some cases from all) such communities. 17 This lengthy statement from Mead is included to indicate the categories of membership 'and interaction with others that he was referring to when he stated that the person . . . Becomes an object to himself only by taking the attitudes of other individuals toward himself within a social environment or context of experience and behavior in which both he and they are involved.18 Thus, context of experience and behavior (referring as it does to the activities in which the person is involved, directly or in- directly, with other persons) can refer to the activities in which the person is implicated with other persons in both "membership groups" (e . g. , political party, church organization, work group) and in "abstract Social classes or sub groups" (i.e., categories such as debtor, Christian, religious person) which do not involve formal membership. Mead sug- 8e sts that both categories of interaction ("membership" and "abstract sOcial groups or subclasses") are of consequence in terms of the per- s<'.>n's participation in the social process in which the self is produced and develops. This is so because in both categories of interaction the Person is engaged in activities which place him in the position where others may direct identifying behaviors toward him corresponding with the particular category of activity. It should be pointed out, however, that neither the mere Presence of others around the person nor the direction of activities 1:0Vvard the person by those others, is sufficient to produce the self. Certainly the behavior of others toward the person is necessary, but 17 , Ibid., p. 157. 18 Ibid., p. 138. 35 the person must "take the attitude of the other" toward'himself; he must behave toward himself as the others behave toward him. The self comes into existence . . . not simply because one is in a social group and affected by others and affects them, but because (and this is a point I have been emphasizing) his own experience as a self is one which he takes over from his action upon others. He becomes a self in so far as he can take the attitude of another and act toward him- self as others act.19 (emphasis supplied) 192333., p. 171. This point is a crucial one in Mead's con- sideration of the consequences of the acts of others for the person's own behavior. It is perhaps the most important differentiation between tra- ditional behaviorism and the social behaviorism which Mead developed. Mead's point was that the person engages in conscious activity toward Obje cts, the self included, and that we must pay attention to the manner in which the person attends to his own activities in relation to the activi- ties which others direct toward him. Perhaps the only student to give focal attention to this idea of Mead was Ellsworth Faris. See his article, "The Retrospective Act, " Journal of Educational Sociology, Vol. 14 (October, 1949), pp. 79-91. Therein Faris deals with the dimension of time in relation to the social act involving both the activities of the person and the activities of others. He argues that the social act provides the basis for the emergence and evelopment of the social self. The self is a Product of the behaviors which others direct toward the person, but the 8°<3ia1 act is a "retrospective act"; i.e. , an act where the object of the a"iti-Vity is the consideration of a previous act. Thus the person behaves towBird himself as others have behaved toward him; he names his actiVities as others have named his activities, etc. Manford Kuhn points Out that "Faris extended Mead's basic idea that it is in the social act that self and meaning arise. He also extended Mead's general notions athOut the relation of time to human behavior. It is time which has proved most troublesome to the proponents of the S-R model of the act, for that fnodel is appropriate only to "immediate acts" [non-self conscious acts], In Faris' terms. Most human acts extend considerably through time. " ee Kuhn's discussion of "The Social Act, " in Julius Gould and William Kolb (eds.) A Dictionary of the Social Sciences, N. Y.: Free Press of MaCmillan, 1964, p. 644. 36 A Formalized Statement of the Meadian Theory of the Social Self The central issue now arises of the connection between the theoretical orientation just stated and the question posed‘at the first of this chapter; viz. , "What is the relationship between the religious activities and behaviors in which a person is implicated with others and that person's self identification?" It is possible, I believe, 'to demon- strate how this question may be logically derived from the theoretical orientation which has been stated in the preceeding pages. I will proceed in the following manner: first, attention will be directed to some specific definitional issues concerning the self and self identification; second, I will attempt to specify the logical relationship between the self and self identification; and third, I will state a limited and tentative series of propositions which will logically connect the discussion of self and self identification to my central question and its tentative answer--the major hypothesis to be dealt with in this study. nginition of the Self The self is an object to which behaviors are addressed. In this respect it is like all other objects.20 The characteristics or mean- ings which this object takes on are those which are imposed upon it by the naming activities directed toward it. The use of the words "it" and ”Object" should not be taken to suggest that the self is a "thing" or "entity." The concept "self object" has as referents collapsed series of activities in which the person engages and to which he addresses additional activi- . 21 . hes - Therefore, the self refers to a class of observable activities \ at 0th 2C)Perhaps the unique way in which the self does differ from 1 er objects is that the self is the one object which can be present in a 1 Situations. th 21Thus this concept, like all of our concepts, simply serves the purpose of collapsing a series of events, attributes, or activities and ereby provides a means of referring to that series of phenomena. 37 which take on their meanings or characteristics through an additional set of acts --naming acts directed by the person toward the activities in which he is engaged.22 The self, therefore, is defined as the acts of naming one's own activities. Neither the above characteristics or definition provide an explanation of how the self comes about. They do, however, set limi- tations within which explanation must be developed. Consistent with 221 believe it is in this context that Mead stated that the person can be both subject and object to himself. The person must, obviously, be an object to himself before he can be a subject. He becomes an object to himself through the activities which others ad- dress to his own activities. This is stated below as a crucial aspect of Mead's central proposition regarding the production of a social self. I am including at this point two rather lengthy quotes from Mead which illustrate his position. (emphasis supplied) "1 have already pointed out that the self appears in the social act and is a derivative of the gesture, that is, the indication by one individual in a cooperative act to another of some thing or character which is of mutual interest. When, as is the case in the human individual, hgcan address himself as he addresses another, and so take both at- 1iiildes, that of the one who indicates and that of the one who is indicated [EL the one, who is indicated to, becomes a social object on the same Pla&with the other. When the memory of the indication associates itself With this object, the self has appeared. " Mead, op. cit., p. 75. "Inasmuch as the physical object which is implied by the phy- pica-1,1 sciences includes the physiological organism of the observer, it 1.3 evident that the individual is not included in this world of physical ob- Jects . He stands outside and presents it. This is made possible by the Social character of the self, through which the individual addresses and obServes himself. The self that one addresses and observes may be the 8001211 individual perceived as other individuals are perceived, and in an énvironment of objects which have the same qualities of immediate exper- was those which belong to other social individuals. In this case the Inchvidual observed is identified with the psysiological self, especially its Vocal gestures, tones of voice, facial expressions, attitudes of body and all that calls out social response in others. The distinction between the Self as speaker and the self as an individual addressed [i.e., between the $3M as subject and the self as object, or, between the "I" and the "me"] Is the distinction which is doing something and the self which is the result of Past conduct. This former self, [the self as subject] is at the threshold of aCition and has as its function the statement and reordering of the world 38 Mead's social conception of human conduct is the assertion that the activi- ties in which the person is engaged and to which he addresses himself are not independent of others and the activities which they address to the per- son. The self, as a class of activities which the person directs toward himself as an object, is also dependent uponthe activities which others direct toward the person and his activities as objects. Relationship Between the Self and Self Identification A clearly relevant and crucial aspect of the self, defined as above, is the class of self identifying or self designating acts in whi'ch the person engages. A large class of readily observable behaviors is composed of those activities wherein the person tells others or himself who he is. Obvious and clear examples of this class of activities are such verbal assertions as: "I am a man"; "I am a Catholic"; "I am a re- ligious persont'; "I am a woman"; "I an a Negro"; "I am lonely"; etc. These are simply identifying or defining behaviors. Therefore we can Speak of this aspect of the self as the class of identification behaviors which the person takes toward himself as an object. —_h as Past and present in the interest of the conduct which is in hand. Out Of this conduct grows a changed situation in which the action of the indi- Vidual has its place. He is creative. This is evident in the social world. It WOuld not be what it is without his conduct. " £113. , p. 89. Z3Cf., Nelson Foote, "Identification as a Basis for a Theory 3f Motivation, " AER, Vol. 16 (February, 1951), p. 17. Foote states: _ We mean by identification appropriation of a commitment to a particular Tdentity or series of identities. As a process, it proceeds bynaming: Its Products are ever-evolving self-conceptions--with emphasis on the c°\n~, that is, upon ratification by significant others. " Orrin E. Klapp Iflakes a similar point in his discussion of "social typing" or naming be- ?‘Viors, the labelling of others and the labelling of oneself. "It is impos- 81ble, except conceptually, to separate this typing of others from the typing of OI1eself.. . . the other as we type him is the key to our construct of our self. We find outselves by the responses of others (Harry Stack Sullivan calls it consensual validation); but it is not simply the response of the‘ other that makes our self but the way we see ourselves in his eyes as we 39 I have suggested that: any object acquires definition or mean- ing through the activities or responses which are directed to it; the person may address activities toward himself "publicly" or "privately" and thus the activities can occur in any and all situations; and, the activities are not independent of other persons' activities of a similar or identical nature . Statement of Propositions Based on the above considerations, we can state the following proposition as central to Mead's implicit theoretical consideration of the self: have typed him. Without knowing what kind of fellow the other is, we can- not know what significance to attach to his approval or disapproval, how to rate ourselves by him. We cannot, then, without knowing him, know who we are." Klapp, Heroes, Villains, and Fools, Englewood Cliffs, N. J.: Prentice-Hall, Inc., 1962, pp. 4-5. 24Additional important aspects of the self would include activi- ties taken toward the acts of self identification: statements of evaluation; statements of desire to change; statements of importance; statements of Perceived consensus with respect to others who would agree or disagree With statements of self identification made by the person; and, the salience Of Self identification statements made by the person. A general discussion Of these dimensions and some preliminary empirical data relevant to the dirrlensions and their interrelationships is found in the paper by Charles W. TuCker, "The Dimensions of Self-Attitudes: A Working Paper, " mimeo- graphed, Department of Sociology, Michigan State University, 1965. A 'report of empirical investigation of the dimension of perceived consensus ls fOund in the paper by the present author. Clark McPhail, "Perceived C_°nsensus Regarding Statements about the Self in Response to the Ques- tlon. 'Who Am I', " paper read at the Ohio Valley Sociological Society Meetings, Columbus, Ohio, Spring, 1964. Kuhn reports data on the re- 1atiOnship between the salience of self-identification statements and conE’aensus; Manford Kuhn and Thomas McPartland, "An Empirical InVestigation of Self Attitudes, .. ASR, Vol. 19 (February, 1954), PP- 68-76. 40 I. The self, defined as acts of naming one's activities, is produced by the behaviors which others direct toward the person. The behaviors others direct toward the person which are crucial to the development of the self are primarily naming, labeling or identification activities.25 Examples of this class of activities are such designations as: "You are a man now"; "Good Catholics don't do things like that"; "Your religion must be very important to you"; "All you women may now be seated"; "You are a Negro and don't ever forget it"; "You must be 26 lonely without your wife, family, job and friends"; etc. These 2.5 The behaviors which have naming consequences are, perhaps, many. These may include both explicit naming behaviors, other verbal behaviors, and, non-verbal behaviors. Goffman, for example, would place considerable emphasis on such non-verbal behaviors as eye con- tact and body contact as well as verbal behaviors. See his Behavior in Public Places, N. Y.: Free Press of Glencoe: Macmillan, 1963, especially Chapters 2 and 9. Professor Robert L. Stewart has suggested to me that perhaps all activities directed toward the person have naming consequences and not just explicit naming behaviors alone. Perhaps if one took a frequency distribution of the variety of behaviors directed toward a person during the course of one day it might be shown that the explicit naming behaviors (e.g. , "You are a man") are less frequent in occurrence than other ver- val, and non-verbal, behaviors which could and do have naming conse- quences. Here, of course, one is attempting to pay attention to the per son's interpretation of all the activities which others direct toward him. For example, "Sit down and shut up, " if "interpreted" in one way by the person has naming consequences. The problem with pursuing this line of reason- ing is that in emphasizing the nonexplicit naming behaviors and the person's interpretation of same, you run the risk of imputing incorrect interpreta- tions or you run the risk of "getting inside the person's head" which is of course impossible and unnecessary. One way of avoiding these common traps would be to record all behaviors directed toward the person and then ask for his interpretation of these behaviors. Such a procedure would in— volve a very time-consuming procedure and although it might prove to be of great value it was not pursued in the current investigation. Rather, I have assumed that both explicit and implicit naming behaviors are frequent in day-to —day social intercourse and that both sets of activities have naming consequences for the person's identification of self. 26The correspondence between these designations and the self identi- fication statements cited earlier, as examples, is not coincidental. Both sets of identification statements are illustrative of the kinds of behaviors 41 designations not only name or identify the person(s) as a social object but, as in the case with the naming of any object, also define and specify ways of acting toward and with respect to the object so designated. Consequently identification behaviors directed toward a person also carry directives for and evaluations of that person and his activities. The first proposition, then, states a relationship between two sets of activities. One set of activities includes those identification behaviors directed toward the person by the others with whom he is engaged in interaction. The second set includes identification behaviors which the person directs toward himself as an object. The first proposition, as well‘as the discussion preceding it, implicates the person in social interaction with others. The various activities in which the person engages with others constitute what Mead called different "social environments" or "contexts of experience and 28 behavior" in which both the person and the others are involved. easily observable in everyday experience. As I stated earlier, it is not expected, due to the ability of the human animal as a symbol manipulator, that designating or naming activities of others and the subsequent identify- ing behaviors of the person would necessarily have a relationship of iden- tical equivalence. They should, however, correspond. The extent of their correspondence is an empirical question. 27 . . . . . This notion has been taken as a given by most soc1ologists with a symbolic interaction orientation. The evidence for the assertion-- that labels not only name objects but suggest ways of behaving toward these objects--demands rigorous and systematic test before it can be accepted as a valid, universal generalization. Such rigor has, to date, been lacking. Relevant and somewhat systematic observations which lend support to the assertion, however, have been provided. See John Kitsuse, "Societal Reaction to Deviant Behavior: Problems of Theory and Method" in Howard S. Becker (ed.) The Other Side, Press of Glencoe -Macmillan, 1964, pp. 87-102. Kitsuse reports data on the identification of homo- sexuals and subsequent behavior toward persons so identified. 28 , . . I am u51ng "context of experience and behav1or" as a con- struct referring to the activities in which the person is implicated (directly or indirectly) with others. This construct will be discussed extensively below. 42 For example, the person may be implicated with the same and/or dif- ferent sets of other persons in different kinds of activities (marrying, raising children, working, "politiking, " worshipping, etc. ) . This common sense observation is stated here merely to acknowledge the variation in the kinds of activities a person may engage in with a variety of others. These various activities may be considered as different referents for the construct "contexts of experience and behavior. " The consequence of this must be that the self, as defined above, can have many facets of aspects. Mead states: We carry on a whole series of different relationships to different people. We are one thing to one man and another thing to another . . . We divide ourselves up in all sorts of different selves with reference to out acquaintances. We discuss politics with one and religion with another. There are all sorts of different selves answering to all sorts of different social reactions.‘29 Thus we can state a second proposition: II. "Selves" are produced in the various "contexts of experience and behavior" by the corresponding activities others direct toward the person. A person, due to the various kinds of activities in which he is involved with other persons, may well behave toward himself as "a husband, " "a father, " "a republican, " "a junk dealer, " "a religious man, " "a person with very little chance for success in this world, " etc. The theoretical orientation under discussion here would suggest that each of these identification behaviors, directed by the person toward himself, corresponds with identical or similar designating behaviors directed toward the person by the others with whom he is implicated in various contexts of experience and behavior. Thus we can state a third proposition: III. The self corresponds with the behaviors directed toward the per- son by others in the various contexts of experience and behavior in which theperson and the others are implicated. 2 9Mead, Mine, Self and Society, p. 142. 43 The present study does not focus upon the issue of the general relationship between all activities and behaviors in which the person is implicated with others, and, all aspects of his self identification. Rather, attention is directed to the specific but representative concern of the rela- tionship between the religious activities and behaviors in which the person is implicated with others and that person's corresponding self identification. Thus the focus is a specific context of experience and behavior and, a specific class of activities--a religious context of experience and behavior and, the class of activities which make it up. Thus, following from Proposition III, we would designate a person who is implicated in reli- gious activities and behaviors with others as Operating within a "religious context of experience and behavior. " If those others direct religious identifying activities toward the person in that religious context of experience and behavior in which both he and they are implicated we may, in line with the above, state the fourth proposition. IV‘. The person who is implicated in a religious context of experience and behavior will identify himself as religious. This proposition constitutes the major hypothesis of the present study. As such it is the tentative answer to the question posed at the outset of the present study concerning the relationship between religious activities and self identification. The operational specifica- tion of terms in the hypothesis and the relevant procedures of observa- tion and analysis will be presented in Chapters III and IV. An additional task remains in the present chapter. Consi- deration must be given to the remaining, and as yet not specifically defined, concept in the major hypothesis--"religious context of exper- ience and behavior . " 44 An Explication of the Construct-- Context of Experience and Behavior Religious Context of Experience and Behavior What is meant by "religious context of experience and be- havior"? I have taken "context of experience and behavior" to be a construct referring to activities in which the person is implicated, directly or indirectly, with other persons. Religious context of exper- ience and behavior, therefore, refers to activities in which the person is implicated (directly or indirectly) with others where the object of the activities is religious. There are two important qualifications in the above defi- nition; directly- indirectly, and others. The former qualification is included to suggest that no act is ever taken completely independent of the acts of other persons even though those persons may not be physically present at the time the act occurs. The latter qualification, "others, " is used frequently throughout my discussion. When I speak of "others" who are implicated in the person's activities, I will have reference to Mead's notion of the consequences of the "generalized other" although I will suggest certain modifications of this concept as my discussion proceeds. In the succeeding pages I wish to discuss both of these aspects of "context of experience and behavior", i.e. , the direct and indirect implication of other persons in the person's activities . Relationship Between the Person's Activities, Objects, and the Activities of Others If one examines the activities directed toward a particular category of objects, it immediately becomes apparent that several dis- tinctions must be made regarding the different classes of activities that can be taken with respect to that particular category of objects . In general terms we say that a person is engaged in a religious act when 45 his behavior is addressed toward an object which he defines, with others, as religious. His meaning for or definition of that object is both pro- duced and sustained by the behaviors which other persons direct toward the object. This is consistent with the assertion made earlier that the meaning of any object, activity, or attribute is determined by the re- sponses made toward it. The "others, " whose responses produce and sustain the definition or meaning of any class of objects, activities or attributes, are not always present when the person in question may direct behaviors toward that object. I have suggested above, however, that their conse- quence or significance in the person's conduct is not limited by space- time boundaries. The fact that such phenomena are frequent, moreover, requires that we specify the different classes of activities which persons can direct toward any particular category of objects. Three such dis- tinctions, or classes, can be specified which are relevant to the present discussion of the activities directed toward religious objects: Class 1: Religious activities taken by the person in the presence of or with other persons who provide, share, and sustain the ongoing difinition of the activities as religious. Class 2: Religious activities taken by the person in the presence of persons who did not provide the person's definition of the activities but who may share and thus confirm that "meaning, " or persons who may not share the per- son's definition of the activities. Class 3: Religious activities taken by the person outside the "immediate physical presence" of other persons. 30 This qualifyer--immediate "physical" presence--is not used to suggest that the other persons involved in the act are any less real because they are not physically present. The qualifier is used simply to indicate a "different" category of person—other relationships; different in the sense that neither the act of the person in question nor the consequences of non—physically present other persons for that act can be directly observed. We are dependent on the reports of the per- son in question for observations of the total phenomena. 46 Class 2 and Class 3 "religious activities" are contingent upon Class 1 "religious activities" and attention will first be directed to a discussion of the latter. The neophyte comes to learn what is and what is not religious activity through the responses of other persons to his participation in the activity. In like manner the responses of others will continue to sustain his definition of the activity and thus may also be relevant to the possible disconfirmation of the activity as religious. An important issue here is the question of "what other persons are relevant in this process?" Mead makes it very clear that it is the response of those others with whom the person is engaged in cooperative or reciprocated activity. He states that the person must not only take over into his own conduct the behavior of others ("take the attitude of the other") toward himself as an object but toward all other objects or activities in which he is involved with those persons. Their response provides him with meaning for the ob- ject or activity and by virtue of sharing the response with them (by taking over their conduct into his own), he has shared meaning with them for the activity. It is upon this basis that all involved or relevant persons can coordinate their activities together with respect to the object or activity in question. Mead speaks of this unit of actors as a "community of others" or "the generalized other. " In discussing the relationship of this unit of persons to the self he comments on" . . . the organized response of the individual to the group in cooperative activities, in which the individual, taking the roles of different members of the group involved in the coopera- tive activity, can address himself as a self. This organized group reaction 3 of the individual over against himself I have termed the generalized other." For the present writer, then, the "generalized other" simply becomes the 31 Mead, loc. cit., pp. 154-156. 32 Mead, Philosophy of the Act, p. 432. 47 organized response of all those persons (with whom the person is im- plicated in cooperative or reciprocated activities of a particular class) toward the person's own activities. And it turns out, for Mead, that this organized or shared response is as universal--as widely shared-- as the size of the community which makes up or constitutes the meaning of the response to the particular activity in question. Mead states: The attitude of the community toward our own response is im- ported into ourselves in terms of the meaning of what we are doing. This occurs in its widest extent in universal discourse, in the reply which the rational world makes to our remark . The meaning is as universal as the community; it is the response that the world made up out of rational beings inevitably makes to our own statement. 33 (emphasis supplied) Mead asserts that "meaning is as universal as the com— munity. " We must therefore consider the possibility of some variation in the comprehensiveness of meaning for activities, objects, and attri- butes, contingent upon the "size" of the community whose members' responses make up the meaning for those activities, objects, and attri- butes. Mead suggests that the meaning of some activities or objects may be produced and sustained by the shared responses of all persons. Perhaps equally so is the notion that the meaning or definition of other activities and objects _may be produced and sustained by the shared re - sponses of a much more limited group of persons. I will suggest that a consideration of activities in both the former and latter cases will present one possibility of resolving the question of "what others are relevant, " and will require a reconceptualization of Mead's notion of the "generalized other. " The "Generalized Other" vs. Specific Sets of Others It is probably a simple minded truism that the observed activity of an erect biped placing one foot in front of the other in a con- tinuous pattern of forward movement is universally responded to, i.e. , 33Mead, Mind, Self and Society, p. 195. 48 defined as, "walking. " "Walking" may be considered as a "universal response that a world of rational beings" can make to this activity. In one sense we can say that everyone is in a position of recognizing, knowing, and re sponsing to the activity as "walking. " Due to this shared response there is an ongoing confirmation or consunsual validation of this definition of the activity as "walking" whenever it takes place in the presence of "rational beings." 34 Such "universal" recognition and consensual validation, however, may not in fact be the case with all activities . 34 . The response to this activity is that of "walking." It is a response that occurs frequently and one that can be consensually vali- dated by anyone and everyone. It is a tenable hypothesis that those activities which can be validated or confirmed by everyone are also those which can be most easily disconfirmed by anyone. This is a corol- lary of an idea suggested to me in conversation with Professor Robert L. Stewart concerning the confirmation and disconfirmation of self identities. "Those identities which are basic (i.e. , those which are most frequently confirmed by everyone--man, teacher, husband, father) are those iden- tities which can most easily be disconfirmed by anyone. Anyone who challenges by identity as a male or as a professor can disconfirm those identities or at least I am disturbed when their behavior does disconfirm these identities. " 35 . . . . . . . Such in fact is the case With all "deV1ant act1v1t1es" and there is excellent evidence, provided by Howard Becker's studies of marihuana users, to support this line of argumentation. The reader will remember that I am discussing the general category of relationships between a person, his activities directed toward an object, and, the activi- ties which other persons direct toward both the object and the activities directed by the person toward the object. In the immediately preceding paragraph I suggested what might be called the hypothesis of universal consensual validation; i.e., "those activities or identities which are most frequently confirmed by everyone are the activities and identities which can most easily be disconfirmed by anyone. " The logical alternative to this we might call the hypothesis of non —universa1 consensual validation; i.e., "those activities or identities which are least frequently confirmed by everyone are the activities and identities which least easily dis- confirmed by anyone. " The evidence which I wish to bring to bear on this line of argumentation stems from Becker's intensive studies of the sequence of activities which leads to the consistent use of marihuana for pleasure. His articles on the marihuana user are collected in Howard 5. Becker, 49 Take, for example, the statement, "I am a religious person." To what extent is there consensual response to this statement; what de- gree of consensual validation exists with respect to the definition of the activity? Some would argue that it has so many "meanings" (or calls out so many responses) as to be almost meaningless. It would be diffi- cult to establish a "universality" of meaning for such an activity beyond those persons whose responses produce and continue to confirm the defi- nition of the activity. This, then, would limit the number of others, or The Outsiders, N. Y.: Free Press of Glencoe-Macmillan, 1963, pp. 41-78. Becker arrives, by means of analytical induction, at four sets or sequences of activities in which novice users must engage, with experienced users, in order to become consistent marihuana users. I will re-state these sequences of activities in my own words although I do not believe I am violating Becker's position by so doing. (1) The novice must learn the activities one must direct toward the object (marihuana) in order to produce the "desired consequences" (effects) of object use. (2) The novice must learn to recognize those consequences, by virtue of the responses of experienced users to his report of the conse- quences, and to connect them with the activities which he directs toward the object. (3) The novice must learn to respond to the consequences of object use in the same manner as those others respond with whom he is engaged in the activities; i.e. , he must learn to define the effects of marihuana use as pleasurable. (4) Most important, the "novice" must continue to interact with these others who, by virtue of the common activities, make up a particular context of experience and behavior, and who are considered by the novice to be in a legitimate position to define the activities and the consequences of the activities. 323. , pp. 46-58. Becker emphasizes the fact that the definition of marihuana usage, and the consequence of that activity, is not one for which there is universal consensual validation. Everyone doesn't engage in the activity and everyone doesn't share the same definition of the activity nor do they share the same definition or identification of persons who participate in the activity. There is, however, a near -universal community of re- sponse and it is this response which is so effective in maintaining social control over participation in the activity. Becker states that this control is achieved " . . . by affecting the conceptions persons have of the to -be -controlled activity, and of the possibility or feasibility of engaging in it. These conceptions i.e., definitional responses arise in social situations in which they are communicated by persons regarded as reputable, and, validated in experience. Such situations may be so ordered that individuals come to conceive of the activity as distasteful, 50 the size of the community, to those with whom the person is involved in the activity and whose responses are relevant to the production and con- firmation (and therefore possible disconfirmation) of that particular Y inexpedient or immoral and therefore do not engage in it." Ibid. , p. 60. Becker finds that all novice users hold this near -universal d-e-f-i-nition at one time; i.e., prior to their persistent use of the drug. However, as the novice begins to engage in the activity of marihuana usage with others, " . . . he is likely to acquire a more 'emancipated' view of the moral standards implicit in the usual characterization of the drug user, at least to the point that we will not reject the activities out of hand simply because they are conventionally condemned. " I_b_i_d. , p. 73. The response of this set of others [i.e., fellow users] is crucial in another sense as well. Becker finds that with continued drug use the novice may have new and unanticipated experiences with the con- sequences of drug use; e. g. , using a "better" or "higher" grade of marihuana on subsequent occasions may produce new, different, frighten- ing (and thus temporarily disconfirming) experiences or consequences. At this point the responses of this "special" set of others is extremely important. With the new and unanticipated experience the novice's defi- nitions of marihuana usage are temporarily disconfirmed. However, if he is participating in the activity with other users they are in a position to re -define these new experiences by indicating to the novice that the experiences are "appropriate, " "predictable, " and "pleasant, " vs. "un- comfortable, " "unpredictable, "'and "frightening. " Thus Becker states: "The likelihood of such a redefinition occuring depends on the degree of the individual's participation with other users. Where this participation is intensive the individual is quickly talked out of his feeling against marihuana use." £13., p. 56. The important point here is that the marihuana user learns to define his activities and the consequence of his activities through the responses of those others with whom he engages in this particular set of activities. All others are not relevant in the definition of the activi- ties. All others are not in a position to consensually validate, and thus to potentially disconfirm, the person's definitional response to the ac- tivities. The person acquires this definition of and meaning for the activity by virtue of .the responses of those particular others within this particular context of experience and behavior and not a generalized com- munity of others in a general context of experience and behavior. Becker concludes: "In short, a person will feel free to use marihuana to the '~degree that he comes to regard conventional conceptions of it as the un- informed views of outsiders and replaces those conceptions with the "inside" view he has acquired through his experience with the drugs in the company ofTother users. " I_b_i_c_1., p. 78. 51 activity. The relevant others for this activity'would therefore be con- siderably smaller in number than the relevant others for the production and confirmation of the definition of more universal activities, objects, and attributes . I have attempted to indicate the need for a reconsideration of the "others" who are relevant in the definition and confirmation of one's activities. This discussion also suggests the necessity for a reconsideration of the appropriate applicability of Mead's concept of the "generalized other. " Some students of Mead have pointed to a logically relevant referent for this concept with their suggestion that one can equate "generalized other" with the third person plural, "they". For example, when you ask the question, "who counts with respect to the organization of the person's behavior, " or "who counts with respect to the definition, evaluation, and expectations for his behavior, " the response invariably turns out to be something like, "they count, " or "they say so," etc. Depending on the particular class of activity the "they" can be a large and quite generalized set of other persons: e. g. , those who would expect males to wear trousers in public; and those who would agree on whether or not a person is wearing trousers. Or "they" may refer to a more limited and particular set of other persons for other kinds of activities: e. g. , those who expect a person to pray in his every spare moment of time in order to be a religious person; those who would agree among themselves in defining this activity as religious activity; and those who would define a person engaged in such activity as a "re- ligious person. " Thus for certain activities the community of others is in- deed almost universal; they "they" is all inclusive and merits the use of the term "generalized other. " For other activities, however, the sets of relevant others are more limited and perhaps different (i.e. , non-overlapping) . A person may interact with, and thus have his activity defined by the responses of, different groups of persons in his 52 occupational, religious, and recreational contexts of experience and behavior. The three groups of persons in these three "contexts," however, may make up a universal community or "generalized other" with respect to the definition, evaluation, and expectation for activity referred to as that of a husband, male, etc. Based on the above considerations I find the concept of "generalized other" misleading when we are dealing with the sets of other persons who are theoretically relevant to a particular class of activities in a particular context of experience and behavior. Rather than refer to the numerous sets of others as "religious generalized other, " "political generalized other, " etc. , which is not only clumsy but misleading, I will refer to these various clusters of others as the person's "religious others, " "political others, " etc. The reader will bear in mind, however, that I conceive of the consequence of these others in the person's conduct as analogous to Mead's concep- tion of the consequence of the generalized other. The distinction is made between "generalized other" and the more limited and specific set of others for the purpose of identifying those others with whom the person is interacting in a particular context of experience and be- havior and who he must take into account in order to organize his be— havior relevant to that context. I have presented some "evidence" to support this notion of the consequence of a more restricted set of others for particular sets of activities in the previous footnote discussing the sequence of activities which novice marihuana users pass through in the process of becoming consistent users. Their definitions of their activities, and thus their self identifications, are not produced by a "generalized other" but by a specific set of others relevant to the particular category of activities in which they engage . Mead's notion of the "generalized other" has been criticized by other students on conceptual grounds. Harry Stack Sullivan has sug- gested that Mead's concept of "generalized other" is too related to and thus limited by the assumption of an extremely homogenous community where all others "count" for all of the person's activities and everyone knows about and shares expectations and definitions for all activities. In substitution for this, based on what Sullivan believed to be an 53 It is consistent with Mead's orientation to contend that a person's "others" need not be present in order to have consequence for his behavior. It is for this reason that I have further delimited the activities directed toward objects to include: Class 2: Religious activities taken by the person in the presence of others who are not included among his "religious others. " These persons did not provide the person's definition of the activities as religious but they may share and thus confirm that definition, or, they may not share the person's definition of the activities. Mead spoke to this issue in terms of the activities which the person directs toward himself as an object. . . the social individual is already in a perspective which be- longs to the community within which his self has arisen. He has become a self by responding to himself in the attitudes of other selves. This involves the assumption of the community attitudes where all speak with one voice in the organization of social con— duct. The whole process of thinking is the inner conversation going on between this generalized other and the individual. The perspective of the individual is, therefore, that of the social-a—c-t --an act which is inclusive of the act of the individual but extends beyond it. 37 (emphasis supplied) extremely heterogenous comtemporary society, he proposed the notion of "clusters of significant others"; i.e. , the person, in contemporary society, doesn't have a "generalized other" but rather has clusters of significant others for the various different activities in which he partici- pates with others. See Patrick Mullahy, The Contribution of Harry Stack Sullivan, N. Y.: Hermitage House, 1952, p. 40. A similar point is made by Cottrell and Foote in their essary "Sullivan's Contribution to Social Psychology" in the same volume. See p. 191. Working within a somewhat different theoretical orientation, C. Wright Mills also takes issue with the generality and homogeneity implied in Mead's concept of the "generalized other. " See his "Language, Logic and Culture, " ASE, Vol. 4, No. 5 (October, 1939), pp. 670-680, footnote 12. ' While I would concur with the criticism which Sullivan, Foote and Cottrell, and Mills level against the generality and homogeneity im- plied in Mead's concept and would propose the necessity of attending to specific sets of others relevant to particular categories of activities, I would not completely rule out the notion that for some activities all others are relevant. Thus for some activities all people are in a position to validate or to disconfirm. It is in this sense that Mead' s notion of "generalized other" is useful. 37Mead, Philosophy of the Act, p. 152. 54 Thus, the responses of others which produce the person's conception of self at time1 are retained as relevant elements of his coterminous and ongoing experience at time where that self is presented. The act taken 2 by the person is a social act involving relevant others whether or not those others are physically present; the act is taken by the person but in doing this those others may be taken into account, present or not. It is in this way, Mead argues, that the person can behave consistently from situation to situation independent of the physical presence of the others whose activities are implicated in the person's own conduct. In this fashion the individual attains the universality of the com- munity responses [e. g. , religious others 1 , which may involve the responses of an indefinite number of individuals, and yet it is a universalized attitude which is specifically related to his individual conduct--it lies within his perspective. It is in this sense, I would argue, that the person who takes religious acts in the presence of others who are not included in his "religious context of experience and behavior" will continue to de- fine those activities as religious in terms of responses shared with his "religious others." The responses of the situational others will certainly be taken into account, but they are not likely to disconfirm his definition of those activities as religious because the situational others are not considered relevant others for the definition of religious activities. Therefore, the activities will continue to be defined by the person with the responses of those others who make up, with him, the community of relevant religious others; those persons with whom he engages in reli- gious activities constituting a religious context of experience and beha- vior. It is with these others that he shares definitions of activities and objects as religious. The remaining class of activities to be discussed is: Class 3: Religious activities taken by the person outside the ‘ presence of others. 38Ibid., p. 153. 55 It is obviously impossible to directly observe this third class of activi- ties. For relevant observations we are dependent upon the person's report of such behaviors. However, consistent with the above discus- sion is the assertion that the person who engages in such activities, according to his report, will define them as religious based on his on- going experience of the responses of his religious others. These are the relevant definitional responses until some disconfirming response is directed toward the activity by his religious others. In the preceding pages I have attempted to specify and discuss three classes of activities which the person can address toward a category of objects. In this case the category is composed of religious objects. I have further suggested that the objects, as well as the activi- ties directed toward the objects, are defined by and with those relevant others who make up with the person a particular "context of experience and behavior."39 Class 1 activities are those which are most frequently considered. Nevertheless, Class 2 and Class 3 activities are also 39 In considering these three classes of activities which the person can address toward a category of objects, I have limited the dis - cussion, by example, to religious objects. In the development of my comments on the notion of the "generalized other" and the relevance of such a set of persons for "universal" activities, and, "particular other" and .the relevance of such a set of persons for more "particular" or non-universal sets of activities, it should become apparent that it is possible to re-state the three classes of activities so that they can apply to all universal and particular categories of activities. Expanding the initial statement, then, the classes of activities are restated as follows: Class 1: Activities taken by the person in the presence of or with those persons who provide, share, and sustain the ongoing definition of the activities through their response to the activities, making up a context of experience and behavior. This would hold for both universal activities (and the notion of "generalized other") as well as par- ticular activities (and the notion of "particular other" in relation to those activities) . Class 2: Activities taken by the person in the presence of persons who did not provide the person's definition of the activi— ties. If the activities are "universal, " these persons 56 crucial in terms of the problems that a researcher has in making obser— vations of the person's activities and reports about activities which take place outside the presence of the person's relevant others. An indi— vidual (e. g. , an interviewer) who is engaged in the observation of these reports is engaged in a social transaction with the person in question and yet, in terms of my above discussion, cannot be considered as a constituent of the person's "religious others" whose activities with the person make up his "religious context of experience and behavior. " Further attention will be directed to the nature of the research situation and the interviewer-respondent relationship in Chapter III. Specification of Others, Circumstances and Activities Making up the Religious Context of Experience and Behavior I have discussed some different classes of activities which a person may direct toward any category of objects and the implication of relevant sets of others in those activities. The next task is that of specifying those "others" who are in a position to direct identifying may share and thus be in a position to confirm or dis- confirm the person's definition of the activities. In such a case these persons would constitute the "general- ized other" of which Mead speaks. If the activities are "non-universal, " i.e., particular, these persons may share and thus be in a position to confirm the person's definition of the activities but not in a position to discon- firm the activities. This is so by virtue of their not being included in the set of "particular others" with whom the person is implicated in the particular context of ex- perience and behavior which the activities make up. Class 3: Activities taken by the person outside the immediate presence of other persons. "Universal" activities would still be defined in terms of the response of a "generalized other" and non-universal activities would be defined in terms of the response of the particular set of others rele— vant to the context of experience and behavior which the activities make up. 57 behaviors toward the person and his activities and/or the circumstances in which those "others" are in a position to direct identifying behaviors toward the person and his activities. It is necessary to consider the circumstances in which the person is likely to be engaged in "religious activities" where "religious others" can direct identifying behaviors toward the person and his activities; circumstances within which the person is likely to be engaged in "religious activities" in the presence of "non-religious others"; and, circumstances in which the person is likely to engage in religious activities outside the presence of other persons. The resolution of these issues should provide the basis for specifying the circumstances in which the activities of the person with relevant others take place, making up what is referred to by the con- struct "religious context of experience and behavior. " The issues, unfortunately, cannot be resolved by recourse to the theoretical orientation. Mead offers no specific indications con- cerning the circumstances within which such activities should occur nor does he provide directives for the identification of persons who should, or in given circumstances could, direct identifying behaviors toward the person in question. It is necessary, therefore, to suggest certain categories of persons who are in the position to direct identi- fying behaviors toward the person in question. A group of persons which immediately presents itself for consideration is the family. It is with the family members that the person's first awareness of self as an object (for which there are defi- nitions, expectations, and evaluations) comes about. The family mem- bers play this initial role and may continue to do so for the person. The behaviors which the family members direct toward the person may spread across a wide range of "contexts of experience and behavior. " On the other hand, this group of persons may, and in our society usually does, come to be supplemented if not supplanted by other groups or clusters of persons relevant to different contexts of experience and behavior. These additional groups of persons also make demands upon 58 the person, direct identifying and evaluative behaviors toward him, etc. One such cluster of persons which may be in a position to direct such behaviors toward the person is that of his close friends. By close friends I mean those persons with whom the person associates frequently, trusts, likes to be with, and whom he conceives as relating to him in a similar manner. Such persons are obviously in a position to direct identifying activities toward the person in question. An additional set of persons is in a position to direct identi- fying behaviors toward the person. These others are those with whom the person interacts in the group which he designates as the most im- portant group among all those to which he belongs. All three groups of persons--family members, close friends, and associates in one's most important group--should be relevant to and included within the first two propositions stated above and derived from Mead's theoretical orientationfjtO To repeat, 1. The self (defined as acts of naming one's activities) is produced by the behaviors which others direct toward the person. 11. The self is produced in the various "contexts of experience and behavior" by the corresponding activities others direct toward the person. Ihave indicated that "context of experience and behavior" refers to ac- tivities in which the person is involved, directly or indirectly, with other persons. "Religious context of experience and behavior" refers to activities in which the person is involved with others where the objects of the activities are religious. Therefore, if the objects of activities are, n__ 40 The reader will recognize that I have selected only three obvious classes of others as examples of those who may direct identify- ing behaviors toward the person. Of the three classes of others only two were systematically examined in the present investigation of a reli- gious context of experience and behavior. Undoubtedly there are addi- tional classes of others who are in a position to direct identifying behaviors toward the person. The specification of these additional classes of others would be contingent upon the context of experience and behavior under consideration by an investigator. 59 or could be, religious for the three groups of persons referred to above, we would expect them to be in a relevant position to direct religious identifying behaviors toward the person in question. Many different categories of objects could be the attention of activities for one's close friends and one's "most important group" associates. However, any object or activity which the person might share in common with his close friends or the associates in his most important group would seem to be a basis for activities taken in terms of that activity or object. A common activity of "religious preference" would be no exception. To the extent that the members of such groups are similar to the person with respect to some common object or activity we should expect it to be more likely that they would direct activities toward the person in question concomitant with that similarity. If one's close friends are homogenous with respect to religious preference, we could infer that they would be more likely to address religious acts toward the person 41 than if they are heterogenous with respect to religious preference. The argument could be made that such an attribute would be so common within this group of persons as to be "taken for granted" in the conduct of those persons' activities regarding the attribute. Thus the "taken for granted" attribute would not be attended to. This is much more likely with one's close friends, who share in common many similar attributes and activities, than with associates in one's most important group. In the latter the very object and purpose for the interaction, in the case under discussion, is the common attribute: religion and religious activities. Moreover, this should not suggest that "taken for granted" attributes are not of consequence. Just the opposite is more likely to be the case. This is a major theme in the work of the phonomenologist Alfred Schutz and has been carried out in rather systematic observations by Harold Garfinkel. Garfinkel shows the consequence of "taken for granted" activity or meanings by assuming them to be operating and then violating them. It is at this point of violation that their importance can be seen. When they become focal points, are challenged, questioned and disconfirmed, we can observe the importance and consequence of "taken for granted" meanings or activities in social interaction. See Garfinkel's article, "Studies of the Routine Grounds of Everyday Activi- ties," Social Problems, Vol. 11 (Winter, 1964), pp. 225-250. 60 The same set of relationships should hold with respect to one's most important group associates. If that group of persons has as its focus political, recreational, educational, or occupational activities, religious activities as objects for attention should not be frequent. If, however, the group is a religious group, such activities should be fre- quent. In designating his most important group as a religious group, the person is identifying a group of persons with whom he interacts, and he is indicating a particular class of activities in which he engages with those persons, a class of activities related to religious objects. These persons and the activities taken with these persons are logically con— sistent referents for Mead's construct "context of experience and beha— vior. " Therefore, a person who designates his most important group as religious is expected to identify himself as religious due to the fact that he is interacting within a religious context of experience and beha- vior and thus is in a position where others can direct religious identi- fying behaviors toward him. A fourth set of persons who are obviously relevant are those others with whom the person is implicated in formal religious activities; fellow church members, co-workers and co-participants in religious clubs and organizations. In the above comments I have indicated specific groups of persons who are in a position to direct "religious identifying behaviors" toward the person in question. It is to these persons and the activities they direct toward the person that our attention should be turned in relation to the first two propositions stated above. The first two propositions, however, did not exhaust the set of logically consistent propositions which I attempted to derive above from Mead's implicit theoretical orientation regarding the self. The remaining two propositions, III and IV are restated at this time. III. The self corresponds with the behaviors directed toward' the person by others in the various contexts of experience and behavior in which the person and the others are im- plicated. 61 If those "others" direct religious identifying behaviors toward the per- son in that religious context of experience and behavior in which both he and they are involved, we may, in line with the first three proposi- tions, state the following proposition which is the major hypothesis to be dealt with in the present study. 42 IV. The person who is implicated in a religious context of experience and behavior will identify himself as religious. The reader will note that beginning with Proposition III an attempt is made to develop a continuing set of propositions which are logically consistent with the demands of the theoretical framework but which deviate from that framework in the strictest sense. It is at this point that I shift my argument to account for an expected relation- ship between self identification and the context of experience and beha- vior in which the person is involved; i.e. , Propositions III and IV do not deal directly with self identification and the behaviors directed toward the person by the others within the context of experience and behavior in which they are involved. This shift is made in order to attempt to account for empiri- cal data at my disposal prior to the development of the propositions and hypotheses although not prior to my concern with the theoretical orien- tation and its relevance to the phenomena of social relationships and ' self identification. The attempt is made, nevertheless, to account for those data in a manner which is consistent with the theoretical orienta- tion and with the propositions derived fr om that orientation. In short, my data do not include observations of the activities which others direct 42 . . . . This assumption is a rather cruCial one to make. My justification is that if the person continues to interact with others in the various religious activities making up the religious context of experience and behavior, and if the others have in common with the person the attri- butes of religious preference, religious experience and beliefs, and re- ligious activities, they will have to engage in the direction of religious identifying behaviors toward one another. At the point when they cease to do this the person, theoretically would no longer be able to interact with them in terms of religion, religious beliefs, activities, etc. 62 toward the person. Neither do they include observations of the reports by the person of activities directed toward him by others in this specific context of experience and behavior. The data for which I will attempt to account include: the person's report of circumstances in which he engages in interaction with others who are in a position to direct religious identifying behaviors toward him (Class 1 Religious Activities); observations of the person's religious activities and his report of religious activities in the presence of non-religious others (Class 2 Religious Activities); and, the person's report of religious activities in the absence of other persons (Class 3 Religious Activities). These observations I take to be logically consistent with Mead's construct of "religious context of experience and behavior. " By interacting with others in such a context of experience and behavior the person is in a position where others should direct identifying beha- viors of a religious nature toward him resulting in a religious self identification as suggested in Propositions III and IV. I do not hold, nor do I wish to imply, that the "religious context of experience and behavior" produces self identification. The self is produced by the behaviors which others direct toward the person. I have attempted to suggest above, however, that a person who is interacting with others in a religious con- text of experience and behavior is in a position where others should direct religious identifying behaviors toward him. Therefore, the greater the extent of the person's participation in a religious context of experience and behavior the greater the likelihood that he will be in a position where others can direct religious identifying behaviors toward him. I have outlined above the circumstances in which the person is implicated with others who could direct religious identifying behaviors toward him and his activities; i.e. , close freinds and associates within the person's most important group. In addition to this, however, one must attend to the religious activities in which the person engages in their presence, in the presence of non-religious others and in the presence of no other persons. These are additional and relevant aspects of the 63 person's "religious context of experience and behavior" and thus corres- pond to the hypothesized relationships stated in propositions three and four. One set of religious activities in which the person can engage is verbal activities or behaviors such as statements of religious belief. I have already referred to the fact that verbal activities are behaviors and that they are treated as such in the present study. Support for this conceptualization is found in the comments of the vehement critic of metaphysics, John Watson, when he stated: Let me make this fundamental point at once: thay saying is doing --that is behaving. Speaking overtly or to ourselves (thinking) is just as objective a type of behavior as baseb all/l:3 Where such behaviors have as objects religion and religious beliefs, it is appropriate to speak of those behaviors as religious behaviors. Such behaviors may be taken in the presence of religious others, non-religious others, or no others. If the "meaning" for such behaviors, however, is acquired and sustained in the manner suggested in our discussion above, one's "religious others" are still taken into account regardless of who is present when the activities are taken by the person. Thus because the person is engaging in religious activities in relation to his "religious others, " we can say that this class of activities is also a logically con- sistent referent for Mead's construct "context of experience and behavior. " The last class of religious activities to be considered are those verbal and non-verbal activities in which the person engages as a participant in the formal and informal patterns of activities of religious groups and organizations; e. g., attendance at religious activities, prayer, meditation, and proselytization, etc. Because these are religious acts, again taken directly or indirectly in the presence of religious others, they are logically consistent referents for Mead's construct "context of experi- ence and behavior. " I have discussed above four sets of activities or categories of interaction as referents for the construct "religious context of experience and behavior": interaction with religiously homogenous close friends; 64 interaction with persons in one's most important group when that group is defined as a religious group; verbal activities directed toward reli- gious objects; and, verbal and non-verbal religious activities as a par- ticipant in the formal and informal patterns of activities of religious groups and organizations. All four sets of activities, or categories of interaction, are logically consistent referents for the construct "reli- gious context of experience and behavior. " I have defined that construct as: the activities in which the person is implicated, directly or indirectly, with others where the object of the activities is religious. Referents for these sets of activities and categories of interaction and the operational procedures for their observation will be discussed in Chapter III. I will nOW address some final remarks to the major hypothesis of the study and the relationship of that hypothesis to Mead's central proposition regard- ing the production of the social self. Discussion of the Major Hypothesis The major hypothesis of Mead concerning self identification was stated above as Proposition 1: The self, defined as the acts of naming one's activities, is pro- duced by the behaviors directed toward the person by others. Propositions II and III above were derived from this major proposition to indicate that the behaviors directed toward the person will vary, and in turn the self which is produced will vary, with the specific context of experience and behavior in which the person and others are implicated. When that context of experience and behavior is a religious one, an association between self identification and the context of experience and behavior should take the form of the relationship asserted in Proposition IV which is the major hypothesis investigated in the present study: The person who is implicated in a religious context of experience and behavior will identify himself as religious. I have pointed out above, however, that the religious context of experience and behavior does not produce religious self identification. 65 Religious context of experience and behavior is a referent for the activi- ties in which the person is implicated, directly or indirectly, with others where the object of the activities is religious. The self is produced by the activities which others, within that context of experience and beha- vior, direct toward the person as stated in Proposition I. Because ob- servations of such activities directed toward the person by others were not available to me, I attempted to specify four sets of activities and categories of interaction which are logically consistent referents for the construct of "religious context of experience and behavior. " My argu- ment, as presented above, was that persons interacting in such a context of experience and behavior with religious others and engaging in religious activities with those others (directly or indirectly) would be in a position where the others could direct religious identifying behaviors toward him. Thus the greater the extent of the person's implication in a religious context of experience and behavior the greater the likelihood that others will direct religious identifying behaviors toward him. This proposition is not to be confused with Mead's proposition that the self is produced by behaviors which others direct toward the person; i.e. Mead did not say that "the greater the behaviors directed by others toward the person the greater the probability that the person will identify him- self in such and such a manner. " Mead made the simple assertion, of a nominal nature, that the behaviors directed by others toward the per- son would produce the self. Thus I do not intend to imply or impose an ordinal statement of relationship between the behaviors of others and the likelihood of the self being produced. I am examining, the reader will recall, the relationship between the extent of implication or partici- pation in a religious context of experience and behavior and the person's self identification. Thus the greater the extent of the person's implica- tion in a religious context of experience and behavior, the greater the likelihood that others will direct religious identifying behaviors toward him. If they address religious identifying behaviors toward him (nomi- nal assertion) he will identify himself as religious. If they do not 66 address religious identifying behaviors toward him (nominal assertion), he will not identify himself as religious. Because I do not have obser- vations of their behaviors, however, I must restate the major hypothesis in terms of the person identifying himself as religious in relation to the extent of (ordinal assertion) his implication in a religious context of ex- perience and behavior where the others involved could in fact direct religious identifying behaviors toward him. Summary The purpose of this chapter was to present the theoretical orientation which gave rise to the question: "What is the relationship between religious activities in which a person is engaged and that per- son's self identification?" The question of self and self identification in relation to the activities of the person with others was discussed within G. H. Mead's implicit theoretical orientation. Mead considers the self as the act of naming ones own activities and views that self as the product of the behaviors which others direct toward the person corresponding to a particular context of experience and behavior in which the person and those others are implicated. Mead's implicit theoretical orientation was stated and then four propositions were de- rived from that orientation arriving at the major hypothesis considered in the present investigation: The person who is implicated in a religious context of experience and behavior will identify himself as religious. The construct--context of experience and behaviors--was defined as the activities in which a person is implicated, directly or indirectly, with others. I specified several classes of activities which can be taken by the person toward an object in which other persons are directly or indirectly implicated. I then proposed a reconceptualization of Mead's "generalized other" in order to specify that set of others relevant to the expectations, definition, and evaluation of the person's activities in a particular context of experience and behavior. I then 67 specified others who should be in a position to direct identifying beha- viors toward the person and his activities and/or the circumstances in which others would be in a position to direct identifying behaviors toward the person's activities. This specification suggested four cate- gories of religious interaction and activities which are logically con- sistent referents for the construct-—religious context of experience and behavior. In conclusion, and based on a discussion of the limitations of the data brought to bear on the major hypothesis, I discussed the relation- ship between the primary Meadian hypothesis regarding the production of the social self and the major hypothesis tested in the present inves- tigation. I restated the Meadian proposition--which holds that the self is produced by the behaviors which others direct toward the person. I then attempted to show that while the religious context of experience and behavior obviously does not produce the self, a person who is im- plicated in such a context is in a position where others can direct religious identifying behaviors toward him. The relationship between his implication in such a context of experience and behavior and self identification should take the form stated in the major hypothesis. CHAPTER III METHODS OF INQUIRY: A DISCUSSION AND SPECIFICATION OF THE OBSERVATIONAL AND ANALYTICAL PROCEDURES Introduction The purpose of this chapter will be to discuss the method of inquiry followed in submitting the major hypothesis to test. This will include a statement, discussion, and definition of the major concepts in the hypo- thesis; a discussion of the specification of operational procedures for the observation of referents versus operational definitions; a preliminary description of the operational procedures for making observations of referents for concepts in the major hypothesis; a restatement of the major hypothesis and statement of sub -hypotheses; a discussion of the validity and reliability of procedures of inquiry; the specification and discussion of observational procedures for referents of the major con- cepts; a discussion of the sample of observations on which the present investigation was based and the issue of sample -universe relationships; a discussion of the analytical procedures; and an Operational restatement of the major hypothesis and the sub -hypotheses. Statement of Hypothe sis The person who is implicated in a religious context of experience and behavior will identify himself as religious . 68 69 Discussion and Definition of Major Concepts in the Hypothesis There are two major concepts in the hypothesis with which I will be concerned in the present section: religious context of experi- ence and behavior (hereafter referred to as RCEB) and Self Identifica- tion (hereafter referred to as SI). I will define and discuss each of the se in turn . RCEB In Chapter II, I defined RCEB as a construct referring to the activities in which the person is implicated, directly or indirectly, with others where the object of the activities is religious. As discussed in Chapter II, RCEB is a construct with several constituent components. Each component was discussed in the previous chapter and the logical relationship of that component to the construct RCEB and to the theoretical orientation was discussed in de- tail in that chapter. The components discussed included; the religious preference homogenity of the person and his close friends (hereafter referred to as RHOMOG); the person's definition of the group which he designates as most important among all the groups to which he belongs (hereafter referred to as MIG); the person's statement of religious belief (hereafter referred to as RBEL); and the person's participation in the formal and informal activities of religious groups (hereafter referred to as RPART) . I will state and define each of these RCEB components in turn: RHOMOG: A context of interaction within which identifica- tion behaviors can take place. It refers to the similarity of the person's religious preference to the religious preference of those persons desig- nated as close friends. MIG: A context of interaction within which identification behaviors can take place. It refers to the group, among all the groups to which he belongs, that the person designates as most important and the person's definition of that group as religious or non-religious. 7O RBEL: A category of verbal religious activities. It refers to the person's statements of religious belief. RPART: A category of verbal and non verbal religious activities. It refers to the person's participation in the formal and in- formal activities of religious groups and organizations. The second major concept in the hypothesis is Self Identifi- cation. By Self Identification I mean the class of identifying or desig- nating behaviors which the person directs toward himself as an object. Those behaviors to be considered in the present study may take the form of either religious self identifications (RSI) or non-religious self identifications (RSI) . The Specification of Operational Procedures Versus Qperational Definitions The operational "definition" of terms occupies a significant place in the methodology discussion of any study. I would like to indicate at the outset of my methodology discussion the distinction between the conventional procedures of "operationally defining" a concept and the procedures which I will employ here. I will not "operationally define" concepts but rather will specify the operational procedures necessary to make observations of the referents for each of the concepts as discussed and defined above and in Chapter II. This distinction is one which has a rather lengthy history in the debate over operationalism, 1 but one 2 which was initially set in sharp relief by the discussion of Orbach and 1The article I am most familiar with here is Herbert Feigl, "Operationism and Scientific Method" in the "Symposium on Operationism, " The Psychological Review, Vol. 52, No. 5 (September, 1945), pp. 250-258. A summary statement of positions, including those by Bergmann, Hempel, and Bridgman, is found in Phillip G. Frank (ed.) The Validation of Scientific Theories, New York: Collier Books, 1961. This is a collection of papers first read at the AAAS meetings in Boston, 1953. 2Harold Orbach, "Operational Definitions and the Natural Science Trend: A Methodological Note, " Midwest Sociologist, Vol 19, No. 2 (May, 1957), pp. 101-103. 71 then extended by Francis. 3 Simply stated, the argument of both Orbach and Francis is that one cannot proceed to make observations of phonomena until one knows what the phenomena are that one is going to observe. To "define" a concept by means of a set of operations pre- supposes that one already knows what the phenomena are to which the concept refers. It implies that one is making observations of that "known" phenomena with the set of operations. This presupposition is not tenable; it has seldom if ever been justified in the social sciences.4 The alternative set of procedures, moving from one's theory and the concepts involved to the processes of observation, should be carried out as follows. First one must specify and define the concept in a manner which is logically consistent with the theoretical orienta- tion within which one is operating. Next it is necessary to specify observable referents for the concept which are logically and theoreti- cally consistent with the specification and definition of the concept for which the referent will stand. Third, the investigator must specify the procedures for making observations of those referents. In all of the above the investigator must attend to the logical consistency of each step with the theoretical orientation within which he is considering the phenomena. That is, one must ask: What observations are required by the theory?; What observations have I made?; and, if there is any discrepancy between the answers to the first two questions, the 3Roy G. Francis, The Rhetoric of Science, Minneapolis: U. of Minnesota Press, 1961. For example, the classic case of "crass operationalism, " discussed by Orbach, is that presented by Lundberg--"intelligence is what intelligence tests measure. " See George A. Lundberg, "The National Science Trend in Sociology, " A_J§ , Vol 61 (November, 1955) , pp. 191-202. This argument, as Orbach points out, presupposes that one knows one is measuring intelligence and not temperature, or weight, or volume, or writing speed, etc. 72 investigator must offer some explanation for failing to take the theory into account. Preliminary Description of Operational Procedures for Making Observations of Referents for Concepts in the Major Hypothesis RHOMOG: The operational procedures specified for making observations of the referent for this concept will be the recording of the respondent's answers to questions concerning his religious preference and the religious preference of most, some, and/or none of his close friends. Five types of "religious homogeneity" are constructed from these answers allowing us to place the person in one of five classes of RHOMOG ranging from complete religious preference heterogeneity, 5Careful attention to my discussion of the logical relation- ship of the various concepts to the theory and of the classes of activity and categories of interaction to the theory as discussed in Chapter II should give preliminary evidence that I have attempted to employ obser- vations of referents that are logically consistent with the theoretical orientation. The various sections in the remainder of the present chapter should further substantiate my attempt to follow the procedures set forth in my discussion of operational specification versus operational definition. I do want to indicate again, as I carefully pointed out in Chapter II, that I did not directly attend to the primary hypothesis of the theory concerning self identification nor the observations required by the theory to test that hypothesis. That is, the theory requires that the in- vestigator observe the behaviors of others directed toward the person in relation to the behaviors which the person directs toward himself. This would be necessary to directly test the primary hypothesis. The present investigation was based upon data which provided observations of certain religious activities, reports of contexts of interaction within which reli- gious activities take place, and reports of religious activities. These observations are logically consistent with the demands of the theory in relation to the hypothesis which was derived from the theory and which is consistent with and related to Mead's central proposition concerning self identification. I will attempt to spell out, in the section on proposals for future investigation in Chapter V, a proposed set of procedures for making observations of the behaviors which others do direct toward the person and which he should, in turn, direct toward himself within any given context of experience and behavior. 73 between the person and his close friends, to complete religious prefer- ence homogeneity, between the person and his close friends. MIG: The operational procedures specified for making observations of the referent for this concept will be a simple classifi- cation of the person's definition of that group which he designates as the most important group among all those to which he belongs. Having designated the most important group, persons are asked to define or categorize that group by placing it within one of a provided set of seven alternatives. These alternatives are then subsequently dichotomized into the categories of "religious" or "non-religious" most important group. RBEL: The operational procedures specified for making observations of the referent for this concept will be the recording of the extent of the respondent's agreement or disagreement with five statements of religious belief. These statements form an ordinal scale approximating Guttman's formal model and have a total cummulative score ranging from zero to twenty. Quartile values are established and assigned on the basis of the person's score within this range. The statements are: "I believe there is a real hell where men are punished for their sins"; "I believe God has a plan for every person's life"; "I believe in a life after death"; "I believe there is a devil who tries to make men sin"; "To me the most important work of the church is saving people's souls. " RPART: The operational procedures specified for making observations of the referent for this concept will be the recording of responses to a five item index concerning the extent of the person's participation in religious services, in the activities of religious clubs and organizations, the frequency of his consideration of a divine being These five items are originally discussed by Snell Putney and Russell Middleton and reported in their "Dimensions and Correlates of Religious Ideologies, " Social Forces, Vol. 39, No. 4 (May, 1961), pp. 285-290. See footnote 25, Chapter I. (D 74 in everyday decisions, and the extent to which he would proselyte or ex- tend his religious beliefs to other persons. These five items also form on ordinal scale approximating Guttman's formal model and have a total cummulative score ranging from zero to fifteen. Quartile values are established and assigned on the basis of the person's score within this range. Self Identification: the operational procedures specified for making observations of the referent for this concept will be the recording of the respondent's statements in response to the question, "Who Am I?" (This will hereafter be referred to as the Self Identification Problem or SIP. )‘7 These statements are then content analyzed, by means of a pre- established set of coding categories, for the presence or abscence of statements designated as "religious.” The self identification index, or SIP, thus becomes a dichotomous classification of "religious self identi- fication statements" and "no religious self identification statements." 7I have chosen to refer to the "Who Am I? " technique as the Self Identification Problem instead of using the conventional designation of TST or Twenty Statements Test. The reasons are fairly obvious. First, as the technique was used in this investigation it did not ask for nor did it elicit twenty statements. That number is an arbitrary one and not intrinsic to the validity of the technique. Second, I do not view the technique as a test. The respondent doesn't pass or fail, he simply makes statements or does not make statements. Third, the technique is a problem which is presented to the respondent. McPartland has pointed this out quite clearly" ". . . respondents are confronted with the problem of identifying themselves and are left to decide for themselves how this identification will be made. " Thomas S. McPartland, Manual for the Twenty Statements Problem, Kansas City: Kansas City Mental Health Foundation, 1959, dittoed. Based on these considerations the designation of Self Identification Problem will be used to refer to the technique using the question "Who Am I ?" 75 Restatement of Major Hypothesis and Statement of Sub-Hypotheses This preliminary description of the operational procedures for observations of the C omponent referents for the construct RCEB makes it apparent that the referents, separately or in combination, can take on a range of values from low to high contingent upon the person's response to questions specified to determine the extent of his implication in the various activities and categories of interaction. The consequence of this variation in the extent of the person's implication in RCEB requires a re-specification of the major hypothesis and a statement of sub -hypo- theses regarding the relationship between each of the RCEB components and the dependent variable of Self Identification. Major Hypothesis: Relationship between RCEB and Self Identification The greater the extent of the person's implication in a religious context of experience and behavior the greater the likelihood that person will make religious self identification statements. Statement of Sub -Hypotheses: Relationship between RCEB Com- ponents and Self Identification l. The greater the extent of RHOMOG the more likely the person will be to make religious self identification statements. 2. The person who defines this MIG as a religious group will make religious self identification statements. 3. The greater the extent of RBEL the more likely the person will be to make religious self identification statements. 4. The greater the extent of RPART the more likely the person will be to make religious self identification statements. 76 Discussion of the Issues of the Validity and Reliability of Procedures of Inquiry Introduction This section will discuss the problems of validity and reli- ability in general as they relate to the procedures of inquiry in the present investigation. In the subsequent section on the Specification and Discussion of Observational Procedures the problems of validity and reliability of each specific set of observational procedures will be reintroduced only where the problems of the specific procedure require special attention. Otherwise, I will make references in that section to my discussion in the present section on the validity and reliability of procedures of inquiry in general. The reader will note that the present discussion is directly related to my earlier consideration of the issues involved in the speci— fication of observational procedures for referents of concepts as opposed to the operational definition of concepts. Therefore, the reader should carefully attend to the presence or absence of a logically consistent relationship between the comments contained in these two sections. The Validity of Observational Procedures There are several procedures which can be employed to determine the validity of observation techniques: construct validity, criterion validity, and what I will refer to as quasi-validity which is simply an inappropriate combination of the first two procedures. I will introduce this discussion by quoting from Abraham Kaplan whose com- ments are particularly relevant to the validity of observation techniques . The validity of a measurement consists in what it is able to ac- complish, or more accurately, in what we are able to do with it. Plainly, this "what" depends on the context of the measurement's use. Validity is not determined just bythe instrument and scale of measurement, nor even also by the "intrinsic" nature of the 77 magnitude [the measurable attribute] being measured. We must take into account as well the functions in inquiry which the measurement is intended to perform, or with respect to which-- whether by intention or not--its validity is being assessed. The basic question is always whether the measures have been so arrived at that they can serve effectively as means to the given end. The usual characterization of a valid measurement is that it is one which "measures what it purports to measure. " Whether it does so is in turn established in two fundamentally different ways, though which is relevant may vary from case to case, and the distinction is seldom a sharp one even for a single context. Briefly, one is a matter of definition, the other of empirical con- nections . Kaplan is speaking here of the procedures referred to above as construct validity and criterion validity. I will discuss these in reverse order and then proceed to a discussion of the bastard form, quasi -validity . C rite rion Validity The procedure of criterion validity established the validity of a new technique of observation by checking the empirical connections between the observations it produces and the observations produced by an older and well confirmed, i.e., valid, technique. 9 Where such established and validated techniques are available for the purposes of demonstrating this empirical connection they mbe used. But, they may be used if and only if they are techniques which are constructed 8Abraham Kaplan, The Conduct of Inquiry, San Francisco: Chandler Publishing Company, 1964, p. 198, emphasis supplied. 9Kuhn and McPartland have pointed out the dubious virtue of accepting the validity of new tests just because they correlate with "established tests. " They state, "There has been a considerable ten- dency to validate each new personality test by correlating its results with those obtained by the already existent ones, without inquiring into their validity. See Leonard W. Ferguson, Personality Measurement, New York: McGraw-Hill, 1952. Ferguson points out (p. 178) that the Bernreuter Personality Inventory was validated by correlating its Scales with scores on the Allport Ascendance -Submission scale, the 78 and conceptualized in such a manner that the observations produced, as well as the interpretations of those observations, are logically consistent with the theoretical framework within which the new technique has been developed. Furthermore, if one already has a set of observational techniques which have been determined to be valid, the construction and test by criterion of the validity of a new technique has little merit unless it can result in a more efficient or economic manner of data collection. Given the theoretical orientation of the present study and the phenomena specified as important by the theory, no criterion measures are avail- able which meet the demands set forth in the above discussion. Construct Validity Kaplan refers to this as a matter of definition. More ap- propriately stated, construct validity is essentially a process of starting with the theoretical orientation within which the investigator is working, and then logically deriving the steps necessary to make observations of the phenomena. Construct validity, then, is essentially a logical process; one simply goes through the procedures of making explicit the logical steps which are involved in relating a particular technique of observation Bernreuter Self-Sufficiency Scale, the Laird Introversion -Extroversion Schedule, and the Thurstone Personality Inventory. The correlations were high. But the Laird and Thurstone tests had been through no vali- dation process whatsoever, and the other two were unsatisfactorily validated: He points out, later, that the Bell Adjustment Inventory was validated against the Allport, Thurstone and Bernreuter tests (p. 232), thus pyramiding still another validation on the original shaky base. And so it goes until people have completely forgotten all details of the con- struction of the earliest tests on whose validity the whole series rests as far as this variety of validation is concerned. " Footnote 9 in Manford Kuhn and Thomas McPartland, "An Empirical Investigation of Self Atti- tudes, "_AS_R, Vol. 19, No. 1 (February, 1954), pp. 68-76. 79 of some phenomena to the theoretical framework within which the phe- nomena are being considered. Therefore construct validity involves what I have discussed above under the rubric of "operational specifications" vrs. "operational definitions. " I have suggested that one must operationally specify pro- cedures for making observations of some phenomena, following the steps set forth in the preceding paragraphs, but only after he has first defined that phenomena in a manner which is logically consistent with the theor- etical orientation within which the phenomena is being considered. Then the task is simply one of specifying procedures of observation which are logically consistent with the concept as defined and with the theoretical orientation within which one is working. Quasi -Validity This approach attempts to take both the definitional and em- pirical avenues of test to determine the validity of observational techniques . Given the manner in which I have discussed construct validity and cri- terion validity, either approach can be employed if the investigator fol- lows carefully the specifications which are involved. Kaplan points out that the two procedures are fundamentally different, but he does not explicitly state that both procedures cannot be used in combination. I would like to point out two cases where the use of both procedures would be in error. The first would be that instance where the investigator attends to the definitional issues concerning the new technique, as noted by Kaplan, but then fails to pay attention to the problems of theoretical and logical relevance of the criterion technique to the new technique when he employs empirical procedures to determine the validity of the latter by the former. An additional misuse of the quasi-validity approach can occur if the investigator begins by showing the logical connections between his new technique and his theory and then steps outside the 8O construct validity approach to employ the empirical approach in the fol- lowing manner: demonstrating that the new technique produces phenomena which are related to other phenomena in a manner specified by the theory. Such an approach is exemplified in Kuhn and McPartland's efforts to demonstrate the validity of the Twenty Statements Test. 10 They made clear the logical steps involved in the construction of this technique in accordance with the demands of the theoretical orientation within which they were working. However, they then attempted to demonstrate this validity by checking the observations produced by the technique in re - l l lation to the kinds of behavioral phenomena predicted by the theory. 0Ibid. They argued that there should be a relationship between the number of consensual statements which respondents made, their "locus scores, " and the respondent's differential social anchorage in majority vs. minority or differentistic religious groups. Their hypo? thesis was confirmed and the association values which they present can- not be questioned. However, the reader will recall that this very argument was one basis upon which the present investigation was launched . I argued in Chapter I that the hypothesized relationship was not consis- tent with the Meadian orientation and consequently the relationship would not hold across a number of investigations. I then proceeded to demon— strate this by presenting the data from two additional investigations, showing that the relationship was not a "universal" or invariant one. I suggested that we should not expect such a relationship to hold if we care- fully examine the postulates of the Meadian orientation; viz., that mem- bership in organizations per se will not produce particular kinds of self identification but rather that the person must behave toward himself as others behave toward him if concomitant self identification is to take place. Thus while I cannot question the data which they present, the argument, which they use the data to support, does not hold up empiri- cally. Furthermore, I would suggest that they are amiss both theoreti- cally, in terms attending only to group affiliation, and logically, in terms of the means by which they attempted to establish the validity of their technique. 1The authors designate the theory with which they are working as symbolic interaction and argue that their conceptualization of self is consistent with Mead's view of the self as an object. Careful examination of the writings of the senior author, Manford Kuhn, would reveal that this conceptualization is more consistent with his own theore - tical orientation, which he designates as "self theory, " than it is with “=1 J. 81 The problem with such an approach is that if the hypothe- sized relationship between the observations produced by the technique and behaviors predicted by the theory is not confirmed, the investi- gator can always blame this lack of relationship on the procedures instead of on the theory. Thus he is always left with an out and he has no way of disconfirming the theory. It may turn out that the tech- nique is quite valid and that the theory is incorrect. When construct validity is employed the investigator must logically connect the procedures to the theory. By following such a tack the disconfirmation of any hyothesis in which the procedure is employed must lead back to the theory. If one has logically established the connections between his procedures of observation and the demands of the theory, and thus has demonstrated the construct validity of those procedures, the disconfirmation of a hyothesis leads to the rejection or modification of the theory. This is what the scientific procedure of inquiry is concerned with. Therefore it turns out that construct validity is the most appropriate way of establishing validity of an observational technique or procedure. It is certainly the most relevant means of establishing the validity of observational procedures when there are no criterion measures and /or when criterion measures may exist but can- not be considered as theoretically relevant standards by which to demon- strate the validity of new procedures of observation. Mead's own position although there are a great number of similarities. For Kuhn's position see C. Addison Hickman and Manford Kuhn, @ividuals, Groups and Economic Behavior, New York: Dryden Press, 1956, "Self Theory" in Chapter III, p. 21-46. For a comparative exami- nation of "Self Theory" with Mead's social behaviorism, see Charles W. Tucker, "Some Methodological Problems of Social Self Theory, " un- published paper, Department of Sociology, Michigan State University, 1 964. 82 The Reliability of Observation Procedures There are several aspects of the reliability of observation procedures. The reliability of the initial observation, the reliability of the manipulations which are performed on those observations and the reliability of observations of phenomena across time are the crucial ones with which the present discussion will be concerned. The last mentioned aspect is most frequently considered in the social sciences and ordinarily is discussed under the rubric of re- peatibility, or some connotation thereof. Collapsing all aspects of reliability under this one guise may be misleading in at least two respects: ( 1) it focuses attention on only one of what turns out to be a multi -faceted question of reliability if one is concerned with the entire research process; and (2) it implies a certain dual reality, independent of the relationship between the observer and respondent, as well as in- dependent of the relationships between the respondent and his relevant others in numerous contexts of experience and behavior. The implication is that the "real" phenomenon and its "real" characteristics are lodged "out there" and that the investigator must grope around until he finds the observation procedure which will uncover the reality of the phenomenon and its characteristics. A view more consistent with the theoretical orientation of the present investi- gation would hold that the phenomena obtained are dependent upon the questions which the researcher asks of the respondent and that the kinds of questions we ask literally determine the kinds of phenomena we ob- serve in terms of the answers to those questions. By this tack the observer determines the reality he observes by virtue of the transaction in which he is engaged with the respondent. Furthermore, as will be discussed in subsequent pages, the respondent's activities, which we observe with our procedures and which we know by virtue of our pro- cedures, are products of his implication in the activities of other per- sons in his various contexts of experience and behavior. Therefore, his activities and the reports of these activities which he makes in re- sponse to the questions we ask remain constant across time only insofar as his relations to others remain constant. Should those relationships change, so his activities and their characterics will change. Mead was quite concerned with the assumptions which the scientist must make in beginning, continuing and summarizing his 83 The emphasis on repeatibility and the possible erroneous implications of such an emphasis have been pointed out by Kaplan. He notes: Repeatibility is often spoken of here as a requirement for scien- tific acceptability. I believe that this is a mistaken specification, or, at best, a misleading one. Many important scientific obser- vations take place on special occasions whose recurrence is inquiry. His position on the relationship between the scientist and the world he sought to investigate is relevant to the issue of dual realities. Andrew Reck, in his introduction to the collection of Mead's articles, points out that Mead argued for the assumption of some reality as a preface to any scientific investigation. He presents Mead's position by extracting quotations from Mead's various works: "First, at its outset science as reflective behavior accepts the reality of perceptual things. Secondly, the scientist returns to the perceptual world of percep- tual things for verification of his hypotheses, aworld he never questions . Thirdly, the scientific method of measurement, although its results are not stateable in terms of physical things, nonetheless makes use of physi— cal things, that is, the instruments of measurement, and so presup- poses their reality. Fourthly, the "exception" that instigates the scientific investigation depends upon the acceptance of reality of per- ceptual things in the manipulatory area of the act. " Andrew Reck (ed. ) Selected Writings of G. H. Mead, Indianapolis: The Bobbs-Merrill Company, Inc., 1964, p. liv. Thus, Mead argues that the scientist must assume the reality of perceptual things. Speaking of the scien- tist Mead states: "His goal in the pursuit of knowledge is not a final world but the solution of his problem in the world that is there" And, ”‘we can conceive of a nature that would not be uniform . . . but . a world that is not there we cannot conceive of. The individuals and their worlds are the presuppositions of all thinking. " Philosophy of the Act, Chicago: U. of Chicago Press, 1938, pp. 60 and 275 respec- tively. Therefore Mead is saying that the presence of persons and "their worlds" must be a given. Otherwise we could not think of them at all. Now, presupposing their existence is one thing. On the other hand, assertions about the regularity or universality of relationships regarding that which we presuppose is another thing. Mead states: "The uniformity appears in experience. The character of universality arises out of the social attitude of the individual [investigator] toward the world. " Ibid. Thus, Mead contends that we do not discover regu- larity of relationship "out there. " We do assume the presence of per- sons and their worlds of experience but we impose uniformity and the character of relationships onphenomena in terms of our response to those phenomena. Thus Mead avoids the trap of dual realities in his conceptuali- zation of the relationship between the investigator and the world which he examines . 84 incidental to their scientific significance . . . Of course when such events happen again we can observe them again, but we cannot repeat the observations at will. And the recurrences can be expected to differ in ways relevant to the purposes of the observation--the core of soundness in the misplaced empha- sis on the "uniqueness" of the subject matter of behavioral science . . . For the scientist, repetition is a device to im- prove the quality of observations, but not the only device, and not necessarily the best. ' The methodological importance of what is called repeatability is, I think, made more plain by its restatement as intersub- jectivity. A scientific observation could have been made by any other observer so situated: nature plays no favorites, but ex- poses herself promiscuously. The intersubjectivity becomes the mark of objectivity, for it testifies that the observation is uncontaminated by any factors save those common to all ob- servers . . . . The question is always limited to whether what is reported as an observation can be used in subsequent inquiry even if the particular observer is no longer a part of the con- text. I ask "Do you see what I see?" to help decide whether what I see is to be explained by self-knowledge or by knowledge of the presumed object. 13 Therefore my discussion of reliability will focus upon the intersubjectivity of observations; i. e . , I will be concerned with whether or not any other observers, similarly situated, could have made the same observations and could have manipulated those observations as l have in this study. The first issue that must be dealt with is the intersubjec- tivity of the initial observation in any study employing interview methods as a means of making observations. Basically this issue has two facets: the question and the response, and, the manner in which both are treated by the observer. In the present study, as in most studies of this type, both open-ended and closed-ended questions were used. When either type of question is employed it is crucial that the wording of those questions be such that they call out in the respondent the same or similar response (i.e. , meaning) that is called out in the investi- gator; i.e. , we must be concerned with the use of what Mead calls 3 Kaplan, op. cit., p. 128. 85 "significant symbols. " Thus, care must be exercised in the con- struction of the questions so that they are intelligible across the language community within which they are employed. This can be accomplished by paying attention to the vocabulary level of the popu- lation, wherein the study is being conducted, when the questions are designed; by carefully observing the ease with which the respondent awswers the questions in the pretest explorations which must precede any study of this kind; and, by instructing the initial observer, i.e., the interviewer, to read the questions to the respondent verbatim from the interview schedule. Assuming that each of these criteria are met; that questions are intelligible; that respondents do not say "huh" or in other manners indicate the lack of clarity of the questions; and, that well trained interviewers carry out the instructions of the in- vestigator, we can assume that our questions are meaningful for the respondents across observation situations. In like manner the re- sponse to the question must be treated in terms of recording the be- havior of the respondent which actually occurs. Particularly in open- ended question sequences the interviewer must record the behavior which does occur and not his interpretation of that behavior; i.e. he must not record what he thinks the behavior means or implies. That the observation of the interviewer is selective is a given. All percep- tion, or conception, is selective. In the process of scientific inquiry, however, that selectivity should be governed by the instructions of the investigator to the interviewer. When competent interviewers are employed the investigator can hope that such is the case. The second issue which must be discussed involves the in- tersubjectivity of dealing with the observations once they are recorded. This is particularly problematic with open-ended questions but also is involved in the manner in which the responses to closed-ended ques- tions are manipulated for the purposes of further analytic procedures. I will discuss each category in respective order. 86 The utilization of responses to open-ended questions re- quires some form of content analytic procedure. These responses are simply another form of verbal behavior whose objectivity is con- tingent upon the manner devised by the investigator for responding to them. The conventional procedure is to establish some desired num- ber of categories which are of theoretical relevance to the investigator; to specify the criteria by which the verbal behaviors are to be coded or distributed within these categories; and finally, to investigate the extent to which independent judges can, following the established criteria, reliably, or intersubjectively, place the verbal behaviors within these categories . By dealing with verbal behaviors in this fashion the manipulation of these observations can be approached as with any other behavior. Closed-ended questions yield responses which can be treated separately or in combination, as in the construction of an index, de- pending upon the analytic procedure chosen by the investigator. A frequent combining procedure employed by social scientists is some type of scaling model which permits the investigator to combine a series of observations of a similar order into a more convenient sum- mary device for the purpose of referring to multiple behaviors of the respondent. One such device is Guttman's formal model for ordinal scales, often referred to as Scale Analysis or Scalogram Analysis. This model, and the manipulation procedure involved, was employed in the present investigation for examining the relationship between the behavioral referents for both RBEL and RPART. A more compre- hensive discussion of the specific procedures relevant to those two indicies will be set forth in a subsequent section of this chapter. At this point I simply wish to present a discussion of the Guttman model and procedures as they apply to my consideration of the intersubjecti- vity of manipulating recorded observations. 87 One ordinarily associates the Guttman model with attitude analysis in contemporary social science. The model, and related procedure, was not put to this conventional use in the present investi- gation. The reasons for this are conceptual and logical ones and will be spelled out at a later point. It turns out, however, that the use to which the model was put in the present study (i.e. , as a combining and analytic procedure) is more appropriate to and consistent with the initial formulation and discussion of Scalogram Analysis as set forth by Guttman. In Guttman's initial presentation he sets forth the position that attitudes may be considered simply as behaviors toward a parti- cular object or group of objects. Assuming that it is extremely in- convenient to make direct observations of all possible behaviors, verbal and non -verbal, toward a particular object, parsimonious at- tention can be directed toward a sample or subarea of behaviors toward a particular object. The verbal behaviors obtained by means of inter- view questions are one such sample or subarea. Guttman states: A subarea of behavior is itself of interest if the larger area of which it is a part is of interest. Furthermore the relationships of a subarea to outside variables can be studied and be useful regardless of the role of the subarea in the total area. . The behavior performed by a respondent on a written question- naire is either to place a check mark opposite a category of the printed answers to a printed question, or to write out a response by himself. In an interview situation, this writing is done by the interviewer himself. If the context of the question is, for example, expression of opinion about the respondent's officers [ or statement of religious belief or frequency of religious par- ticipation] then the responses are classifiable as a subuniverse of the men's attitude [behavior] toward their officers, [or toward religious objects] 88 Such a subuniverse of behavior of responding to a questionnaire is of course, not the same as other behavior with respect to officers [or other behavior with respect to religious objects.] But, as Guttman clearly points out, it is one class of relevant beha- vior which may be considered representative of a larger possible class of behaviors toward the particular object in question. The questionnaire technique is simply one means of making observations of the class of behaviors in which the investigator is interested. Scale analysis was developed by Guttman to deal with the problem of establishing the unidimensionality of any set or sample of observations of behaviors from some given universe of behaviors of interest to the investigator. Guttman's scale analysis provides a test of the representativeness of a sample of behaviors from a given uni- verse of behaviors. The Guttman technique is applicable to any tech- nique of observations or any combination of techniques of observations of behaviors from any universe of behaviors which are of interest to the investigator. 15 Guttman remarks that "the important thing is not how the observations were obtained but that the observations be of central interest to the investigator. " Scale analysis does not determine if the particular content is from the given universe. "An attribute [i.e., behavior] belongs to the universe by virtue of its content. The investigator indicates the content of interest by the title he choses for the universe, and all at- . . . . . l7 tributes Within that content in the universe . " 4Louis Guttman, "The Problem of Attitude and Opinion Measurement" in Samuel Stouffer (ed. ), Measurement and Prediction, Vol. IV, "The American Soldier: Studies in the Social Psychology of World War II," Princeton, N. J.: Princeton University Press, 1951, pp. 52-53. 1 5Guttman, op. cit., "The Basis for Scalogram Analysis, " p. 88. "Scale analysis is a formal analysis, and hence applies to any universe of qualitative data of any science, obtained by any manner of observation. " l6 . Ibid., p. 86. 7Ibid., p. 85. 89 Scale analysis "tests the hypothesis that a group of people can be arranged in an internallymeaningful rank order with respect to an area of qualitative data. A rank order of people is meaningful if, from the person's rank order, one knows precisely his responses . . . 18 to each of the questions or acts included in the scale." The scalogram analysis procedure determines the presence or absence of a scale by ranking behaviors from most extreme or dif- ficult to least extreme or difficult and by ranking persons from greatest extent of favorableness, or greatest extent of behaviors, to least extent of favorableness, or least extent of behaviors . The resulting pattern of internal order among items, among persons, and between items and persons will take the form of a parallelogram (or half-parallelogram) if a scale obtains. A perfect distribution of behaviors or responses in rela- tion to the half-parallelogram scale model would allow for perfect pre - dictability of behavior response as a function of scale type order. "Perfect scales are not found in practice. " "The degree of approximation to perfection is measured by a coefficient of reproducibility [hereafter referred to as C. R] which is the empirical relative frequency with which the values of the attributes [i.e. , ' extent of acts] do correspond to intervals of a scale variable [i.e. , scale scores]. "20 It tells you how accurately or reliably you may predict the behaviors of the person on the basis of his overall scale type score. Because C.R.is a measure of this approximation it is a measure of the internal rank order between items, between pe0ple, and between items and people. Ninety per cent scales or better, i.e. . 18 Ibid., p. 88. (emphasis supplied) 19 . Ibid., p. 89. 20 , Ibid., p. 89. 9O C.R. = . 90, have been accepted as efficient. approximations to a per- fect scale.21 When one obtains a C.R. of this value he can be reason- ably assured that the sample of behaviors, from the universe of behaviors in which he is interested, does interrelate to the dimension; i.e. , the investigator has a representative sampling of behaviors from the universe of behaviors in which he is interested. He knows that 21 The theoretical maximum C.R. = l. 00 and Guttman states that there is a theoretical minimum C.R. of . 50. However, Guttman has been criticized for not providing a measure of a unique minimum reproducibility which could be calculated for each scale. Two scales might have C.R. = . 90, but one scale might have a minimum reprodu- cibility of . 60 while the other might have a minimum reproducibility of . 80. It would be misleading to say that both C.R. values provided the same information regarding the prediction of acts based on know- ledge of scale score or to say that both provided the same information about the internal rank order between persons and behaviors. C.R. is severely affected by the per cent of persons answer- ing an item correctly or we could say that it will be affected by the per cent of persons taking a particular act versus not taking that act. This is often referred to as the difficulty of an item. White and Saltz have pointed out that "the reproducibility figure C .R. can approach its ab- solute lower limit of 50 per cent only when all items have a difficulty level of 50 per cent, a trivial case in which 100 per cent reproducibility could be obtained only if one -half the subjects passed all items while the other half failed all the items . With even slight departures from this strict condition, the lower limit of the reproducibility index rises sharply . . . . This fact makes it exceedingly difficult to evaluate an obtained index of reproducibility. With short scales and wide spread in item difficulties, Guttman's figure of 90 per cent may on occasion be very little higher than the minimum reproducibility of the scale. " Benjamin W. White and Eli Saltz, "Measurement of Reproducibility, " The Psychological Bulletin, Vol. 54, No. 2 (March, 1957), pp. 85-86. White and Saltz review and present the various techniques for checking the unique minimum C.R. for each scale as well as measures of the ratio of the minimum C.R. to the obtained C.R. The present writer used techniques developed by Jackson and reported by White and Saltz. (See J. M. Jackson, "A Simple and More Rigorous Technique for Scale Analysis, " in A Manual for Scale Analysis, Part 11. Montreal: McGill U., 1949, mimeographed.) Jackson's measure is called Plus Percentage Ratio (PPR) and PPR is so calculated that an absolute minimum C.R. can be established for each scale. This ratio involves the relationship between MR (Minimum Reproducibility) and 91 these behaviors are representative because they are reproducible; i.e. , they form an internal rank order among themselves with respect to the patterned extent to which people carry out these behaviors. The last and most important aspect of scale analysis, as discussed by Guttman, is the relationship of the internal rank order pattern of behaviors to the issue of repeatability. Guttman argues that "scales are relative to time and to population. "22 This is, in the present writer's opinion, an extremely sound theoretical position given the man- ner in which Guttman defends it. The position is not congruent with that taken by most social scientists who use scales, particularly those who use, and discuss the use of, scales to "measure" attitudes. While C.R. I have been introduced to these measures under different labels. I will refer to these as MMR (Minimum Marginal Reproducibility) and IOC (Improvement Over Chance), the former being the same as Jackson's MR and the latter the same as his PPR. MMR is obtained by computing the modal response to each item for all items and dividing by the number of subjects times the num- ber of items. Remembering that C.R. is an indication of the extent to which we can predict the person's response to all items by virtue of knowing his total rank score on the interrelated items, MMR indicates how well one can predict the responses made by a sample of subjects to all items if the most frequent response made to each item were pre- dicted as the response made by every person. IOC is obtained by dividing the difference between C.R. and MMR by 1. 00 minus MMR. # Errors . ‘ C.R. -MMR C.R. =1.00- IOC: # Subjects X # Items 1.00 - MMR The IOC measure is thus an improvement over the C.R. value in that it provides the opportunity of determining a C .R. of unity or a C.R. of zero. It is not a substitute for C. .R . but simply a means of providing additional information. It indicates how great the extent of obtained reproducibility is over the above minimum marginal repro- ducibility. While no standard level of IOC has been established, Jack- son has suggested an arbitrary, and tentative, level of IOC = . 70 as desirable. Thus, the investigator would want to have a C.R. = . 90 and an IOC of around . 70. 22 Guttman, op . cit . 92 these persons consistently employ Guttman's techniques, it is ap- parent that they have payed little, if any, careful attention to the theoretical framework within which Guttman discusses the scale model. Thus, the statement of "scale relativism" may sound strange coming from one of the high priests of contemporary socio- logical technicians. The contemporary attitude technicians hold, as I have pointed out earlier, that there is a real set of latent tenden- cies toward behavior which can be uncovered if one asks the right question. Careful attention to Guttman's discussion of attitudes as behaviors will reveal that he conceives of attitudes as behaviors and not as tendencies toward behavior nor as unobservables nor as per- manent entities lodged somewhere in the cranium.23 Guttman takes a position that is consistent with the assertion that behaviors are contingent upon the relationships in which the person is engaged with others and thus states that any ordered pattern of relationship among those behaviors, i.e., scales, must be relative.24 Guttman states: For a given population of objects [persons], a universe [of behaviors] may be scalable at one time but not at another. A universe may be scalable for one population but not for another, or it may not be scalable for an entire population, but scalable for a subpopulation. However, 'if a universe is scalable for an entire population, it will be scalable for all major subpopulations . 2 5 I have suggested that this last aspect of the discussion of manipulating observations is crucial because it includes an c,.__._ Guttman makes it very clear in his discussion of "The Problem of Attitude and Opinion Measurement" that he considers "attitudes -as -behaviors" and that sociology must deal with observables rather than with imputed or inferred "predispositions" to behave . Op. cit., pp. 49-52. 24 Ibid., p. 82. 25 , Ibid., p. 89. 93 introduction, based on Guttman's discussion of the ordered pattern of relationship between behaviors, to the issue of repeatibility which was earlier set in sharp relief by the comments of Kaplan. As Kaplan points out, any connotation of replication in the use of repeatibility leads to an untenable position. The tenuous position is that the be- haviors or traits or characteristics of persons will exist across time independent of the person's relations to those others with whom he is implicated in various activities. Kaplan's point is that no observation is ever repeated although we can make additional observations of similar phenomena should they occur again. Therefore, he urges that the entire reliability issue be rephrased in terms of the inter- subjectivity of observations: to repeat, could other observers make the same observations if they were so situated? This is the major and crucial question with which the investigator should be concerned in considering the "reliability" of his observational procedures. For it turns out that whether or not similar phenomena occur again, and thus result in additional observations of the similar phenomena by the same or different observers, is a matter of the net- work of social relations in which the person is implicated and not of intrinsic reliability of the observational technique. Regardless of the class of behaviors with which sociologists are concerned they cannot be considered, at least by this investigator, as permanent or fixed entities or things. The self, for example, or some class of religious activity, to take another relevant example, is a product of the person's relationships with others. Consequently as those relationships change, i.e. , either the others change or the nature of the relationships changes, we should expect concomitant changes in self identification or in the particular class of religious activities. To my knowledge no investigator has, of this date, followed a group of respondents across time for the purpose of investigating the nature of this relationship and concomitant changes in self identifi- cation or any other class of activities. Such an investigation would 94 require an extensive and intensive effort, and it is an investigation which is sorely needed to verify some of the basic assertions of the Meadian orientation. And, I must add, this problem represents one of the most wide ranging limitations of the majority of theoretical orientations and observation techniques employed in social science research today. Where time dimension studies or panel studies have been employed little if any attention has been given to this crucial variable which I mention above. The relationship of the person with others should, if it remains constant, result in similar observations across time and if not constant should result in different observations across time whether the phenomena under consideration are self identification statements, voting behaviors, or pre -marital sexual behaviors . Summary In this section I have discussed the issues of validity and reliability of procedures of inquiry. I have discussed the various means of establishing validity, the criteria which must be met for each, and, the limitations and consequences of employing those pro- cedures when the criteria are met as well as when they are not. Based the discussion I indicated that the validity of techniques employed in the present investigation could only be established by means of demon- strating their construct validity. I discussed the importance of the question of reliability at all points during the research process and pointed out the value of emphasizing the intersubjectivity of both observations and the manipu- lating of observations rather than the often emphasized repeatibility of observations . Following Kaplan I emphasized the intersubjectivity of any observation procedure or of any procedure of manipulating Observations. This simply requires an affirmative answer to the question of whether any other observer, similarly situated, could 95 have made the same observation or performed the same manipulation of observations. I further pointed out the erroneous implications of pursuing the emphasis on the repeatibility of observations by suggesting that the observations obtained are a function of the transaction between persons. They are a function of the transaction between the observer and observed. Furthermore the activities reported by the observed are a function of his relationship to the others with whom he is implicated in various contexts of activity. Any change in this relationship, be it a change in the others with whom the observed is implicated in activities about which he reports or a change in the relationship with the orserver by virtue of the activities which he directs toward the observed, will alter the observations that are made. Therefore, the reliability or intersubjectivity of observations is not intrinsic to the observation technique but rather to the transaction between observer and observed as well as between the observed and those persons to whom he is related in other contexts of activity. I concluded by suggesting that this is a serious problem but that it was not systematically investi- gated in the present study nor in any other study of which I am aware . Nevertheless, it is an issue that is crucial to many assumptions of the present theoretical orientation as well as to the observational pro- cedures employed by every other theoretical orientation. As such, it should receive systematic attention in the future. In summary, then, I have attempted to indicate the crucial issues of validity and reliability as they relate to the present investigation and to procedures of inquiry in general. I have intro- duced the criteria which should be met for validity and reliability. In Chapter II, as well as in the present chapter, I have attempted to indicate the logical connections between the theory and the proce- dures of inquiry proposed for this study, thereby attesting to their con- struct validity. I have specified the criteria for intersubjectivity of the 96 observation procedures and will, in the next section, describe the operations performed in the present study as a means of demon- strating the intersubjectivity of the procedures against the criteria discussed above . Specification and Discussion of Operational Procedures for Observing Referents of the Major Concepts I will specify and discuss the observational procedures in the following manner. For each concept I will repeat the definition stated earlier in this chapter; I will indicate the class of activity or category of interaction within which the particular class of activity could take place;26 and I will describe the observational procedures for the referent for the class of activity. RHOMOG RHOMOG refers to the similarity of the person's reli- gious preference to the religious preference of those persons disig- nated as close friends. It is based on a report by the respondent of I will repeat, for the reader's convenience, a state- ment of the three classes of religious activities discussed in Chapter II. Class 1: Religious activities taken by the person in the presence of or with other persons who provide, share and sus- tain the ongoing definition of the activities as religious. Class 2: Religious activities taken by the person in the presence of others who are not included among his "religious others." These persons did not provide the person's definition of the activities as religious but they may share and thus confirm that definition, or they may not share the person's definition of the activities. Class 3: Religious activities taken by the person outside the presence of others. For a discussion of these classes of activities in general, the reader may wish to re -read the comments in footnote 40, Chapter II. 97 a category of interaction within which Class 1 religious activities can take place. The observations, fromwhich this index was composed, were based on the person's answers to a series of questions asking for his religious preference, and the religious preference of close friends. The interviewer provided the person with a definition of what the investigator had in mind by close friends. That defini- tion is repeated here. "By close friends, we mean people you like to be with, people you trust, people who would help you, or that you would help. " The respondent was then asked if all of his close friends had the same religious preference; if he (the respon- dent) had the same religious preference as all of his close friends, most of his close friends, or none of his close friends. The res- pondent's answers to these questions allowed the construction of a five -type index ranging from total homogeneity of religious prefer- ence (between respondent and his close friends) to total religious preference heterogeneity. Assuming that the respondent's religious preference was Baptist, the following examples illustrate the five types of re- ligious homogeneity with close friends that were established for the purposes of the present study. Type I: The respondent is a Baptist, E} of his close friends (as defined) are Baptists, and wof his close friends are of any other religious preference; i.e., they are homogenous with respect to religious preference. Type II: The respondent is Baptist, M of his close friends are Baptist, but some of his close friends are of a different religious preference, e.g., Methodist. Type III: The respondent is Baptist, m of his close friends are not Baptist, but are of some other religious preference, e.g., Presbyterian, but some of his close friends are Baptist. Type IV: The respondent is Baptist, but all of his close friends are some other one religious preference, e.g. , Lutheran (i.e., he is the only Baptist preferent among his group of close 98 friends). Type V: The respondent is Baptist, no one other of his close friends is Baptist, and each of his other close friends is of a different religious preference; i.e. , no two people in the group of close friends, respondent included, are of the same religious preference; e. g., respondent is Baptist, one is Methodist, one is Presbyterian, and one is Lutheran. Due to the small number of persons falling in Types IV and V, those two categories were com- bined resulting in four types of RHOMOG ranging from most homo- 2 genous, Type I, to least homogenous, Type IV. 7 MIG MIG is the person's definition of that group, among all the groups to which he belongs, that he designates as most impor- tant. It is a report of a category of interaction within which Class 1 religious activities can take place. In the present study respondents were asked a number of questions about the groups to which they belonged, the character- istics of those groups, the extent of their participation in those groups, and the characteristics of fellow participants or members in those groups. This information was obtained, where applicable, on the respondent's church and other religious activity groups, labor unions, farm, business or professional organizations, societies, fraternal organizations, educational groups and recre- ational organizations. Following this series of questions the re- spondent was asked: "Taking all of these groups and organizations 2 7162 persons could not be classified with this index due to lack of response to one or more of the questions used in con- structing RHOMOG. 99 into consideration, which One is the most important to you?”2 Upon naming a group, the respondent was asked, "What kind of group is that?" followed by the suggested categories of; "political?", "religious ?", "occupational or professional?". Hsocial'?", "educational?", "civic?", "service?", or "recreational?" The respondent was asked to use these categories to define his group which he had designated as most important. Based on this information it was possible to classify the respondent's "most important group" into eight different categories. For the purposes of the present investigation the responses were then dichotomized into either "religious most important group" or "non-religious most important group. " RBEL RBEL refers to the person's statements of religious belief. As such it is a Class 2 religious activity taken in the presence of the interviewer, a non-religious other. 28This question alone reduced the original sample size of 1, 528 to 1, 199, or by about 22 per cent. Respondents were asked this question only if they had previously indicated that they had a religious preference and attended church services, or belonged to a labor union, farm organization, business or professional organization, fraternal, social service, civic, recreational, or educational organization. If they had indicated no such membership, and 21 per cent fell into this category, they were not asked the question to determine their most important group. Thus only 78 per cent of the original sample of 1, 528 were asked this particular question. Of these remaining 1, 199 persons, 175 mentioned no organization in response to the question. This left a sample of l, 024 persons, or 67 per cent of the original sample of 1, 528, who indicated some organization as most important among all the groups to which they belonged. This figure may seem extremely high. Several studies indicate that only about 36 per cent of the U. S. adult population belong to some voluntary organization. But this figure of 36 per cent ex- cludes church and often union membership from consideration. If we add the persons who belong to a church or a union to this fugure of 36 per cent our obtained figure of 67 per cent does not then seem so high considering the fact that our question included all associations or organizations . 100 The respondent was handed a card by the interviewer on which was printed five categories of response. The respondent was then asked to indicate to the interviewer the one response, among the five, which best indicated his agreement or disagreement with the following statements: "I believe there is a real hell where men are punished for their sins"; "I believe God has a plan for every person's life"; "I believe in a life after death"; "I believe there is a devil who tries to make men sin"; and "To me the most important work of the church is saving people's souls." The five answer categories and their assigned values were: strongly agree - 4; slightly agree - 3; don't know - 2; slightly disagree - l; and strongly disagree - 0. Summing across the item value for all five statements of religious belief, the total score of the respondent could range from a maximum of 20 to a minimum of zero. These five items were subjected to a Guttman scalogram analysis using Waisanen's typewriter notation technique. 30 A mini- mum Coefficient of Reproducibility of . 94 was obtained on three separate samples prior to the present study. 31 For the present investigation the five statements of religious belief were again sub- mitted to scalogram analysis on a 10 per cent random sample of the 2 9Taken from Putney and Middlteon, op. cit. See foot- note 6, Chapter III. Fred B Waisanen, "A Notation Technique for Scalo- gram Analysis, " Sociological Quarterly, Vol. 1, No. 4 (November, 1960), pp. 245-252. 31 , The three samples conSisted of two "available samples of college students in Introductory Social Psychology classes at Michigan State University during 1962-63, and one non-random sample of adults, taken on an area-proportion cluster sample basis, from Lansing, Michigan, during the pre-test phases of the present study in the Spring and Summer of 1963. 101 respondents. On this subsample the C.R. was .94, the Minimum Marginal Reproducibility was . 70 and the Improvement Over Chance Score was . 80. Based on this information from four'different samples of persons we can say that the five items approximate the formal Guttman model for an ordinal scale and that they may be considered as a unidimensional set of verbal behaviors with respect to the object of religious belief. Based on the total score range of zero to twenty, quartile values were determined for the entire group of respondents in the present study. The quartile ranges were as follows: Q1 = 0 to 12; Q2 =13 to 17; Q3 = 18 to 19; and Q4 = 20. Based on these quartiles each respondent was assigned his appropriate quartile score desig- nation. It should be apparent by reading the quartile ranges for the total scores on these five statements of religious belief that they constitute a scale of behaviors which load toward the upper end of the scale; i.e., the items were easy to agree with for the respon- dents in the sample for the present investigation. This is apparent not only in the assignment of quartile scores, but also apparent from sight inspection of the half-parallelogram pattern produced by the scalogram analysis . While a nearly perfect scalogram pattern ob- tained, the items were pushed toward the upper end of the pattern. RPART RPART is the person's participation in the formal and informal activities of religious groups and organizations. It is made up of reports of Class 1, Class 2, and Class 3 religious ac- tivities as discussed in Chapter II. The index was composed of ob- servations of the person's response to the following five questions or series of questions. (1) "When you have a decision to make in your everyday life, do you ask yourself what God would have you to do?" If Yes, "Always, most of the time, or only sometimes?" 102 (2) "Other than church membership, do you belong to any religious groups or church clubs of any kind?" If yes, "Would you give me the name of ONE which is most important to you? And, How often do you participate in the activities of this group?" (3) "How often do you attend religious services? " And, the respondent was asked to indicate the extent of his agreement or disagreement with two statements which are viewed as statements of intent, or lack of same, to proselytize others in terms of the person's religious beliefs. The respondent was asked to respond to these two statements in the same manner as to the statements of religious belief discussed in RBEL. (4) "It really doesn't matter what an individual believes about reli- gion as long as he is happy with it. " (5) "I believe the world would really be a better place if more people had the religious beliefs which I have . "32 Each observation was assigned a score with the possible range from zero to three making possible a total score ranging from zero to fifteen, or from low to high religious participation. These five items were also submitted to Guttman Scalogram analysis on a 10 per cent subsample of the persons observed in the present inves- tigation. A Coefficient of Reproducibility of . 90 was obtained, with a minimum Marginal Reproducibility of . 75 and an Improvement Over Chance Score of .60 32 The five items were scored in the following manner. (1) Frequency of consideration of "divine will" in decisions: never - O; sometimes - 1; most of the time - 2; always - 3. (2) Participation in activities of religious groups other than church membership: never- O; less than once a year - 1; once a year - l; a few times a year - 1; once a month - 2; a few times a month - 3; and at least once a week - 3. (3) Frequency of attendance at religious services: never - O; less than once a year - O; once a year - 1; few times a year - 1; once a month - 2; a few times a month - 2; once a week - 3. (4) It doesn't matter what a person believes about religion: strongly agree - O; slightly agree - 1; don't know - 2; slightly disagree - 3; strongly disa- gree - 3. (5) More people should have my religious beliefs: strongly disagree - O; slightly disagree - 1; don't know - 2; slightly agree - 3; strongly agree - 3. 103 Quartiles were established for the total range of scores of all respondents. The quartile ranges, and concomitant quartile values assigned to each person on the basis of his score within the ranges, were as follows: Q1 = 0 - 3; Q2 = 4 - 6; Q3 = 7-9; and Q4 =10 - 15. Self Identification Self Identification is defined as the class of identifying or designating behaviors which the person directs toward himself as an ob- ject. The operational procedures specified for making observations of the referents for this concept were the recordings of the person's re- sponses to the question "Who Am I?" which is referred to as the Self Identification Problem (SIP). The following instructions were read verbatim by the inter- viewer to the respondents in the present study: Now we have quite a. different thing for you to do. Although probably new to you, it is easy and I think you will find it quite enjoyable. Everyone we have asked to do this has found it to be interesting. Now let me tell you what we have in mind. Ask this question of yourself, "Who Am I?" Think of as many answers as you can in answer to the question, "Who Am I?" In a moment I would like you to give me the answers as if you were giving them to yourself, not to me or anyone else Take a little time to think about it. (INTERVIEWER PAUSES HERE MOMENTARILY) Now please make what you consider to be the most important state- ment about yourself first. If the respondent began to make statements in response to the question and continued spontaneously, the interviewer simply recorded those responses verbatim. If, however, the respondent made a response or responses and then stopped the interviewer utilized the following probe procedure. Now make what you consider to be the next most important state- ment about yourself. 104 If the respondent continued to make statements the interviewer simply recorded them. If not, the interviewer continued to probe in the above manner up to a limit of three minutes. The final probe question which the interviewer could employ was, Are there any other statements you could make about yourself in answer to the question, "Who Am I?” With such "open-ended" questions the interviewer must, as discussed in the section on validity and reliability, read all ques- tions verbatim from the schedule and record all responses verbatim. The possible loss or misconstruing of information which can occur at this juncture is minimized only by the assumption that one's interviewers are following the rules for observation procedure as layed down by the investigator. Such an assumption is made in this study given the reason- ably high level of competence and experience of the trained interviewers employed in the present investigation. The respondent's total set of statements (hereafter referred to as the respondent's protocol) was then subjected to a procedure of Content analysis. Coders utilized a pre -arranged set of 44 categories, developed over a lengthy and extensive period of prete sting this par- ticular format of the "Who Am I? " observational technique. Of the 44 COding categories only three are of relevance to the present investiga- ti-On. These three categories pertain to statements of religious self identification. I will describe and refer to the three categories here by the same designations which were employed in the code book. 48 - Religiosity (Qualified or Unqualified) Excluding references to church, formal religion or religious organization. Examples: I am a religious person. I am a child of God. I lead a Christian life. 52 - Reference to Religion and Religious Organizations ( Unqualified) Examples: I am a Catholic. I am a church member. I go to Church. 105 33 53 - Qualified Reference to Religion and Religious Organizations Examplesz I am a good church worker. I am loyal to my church. I go to church every Sunday. Statements which could be placed in any of the above three categories were coded as statements of religious self identification. All other statements were coded as non-religious self identifications regardless of the remaining 41 categories in which they could be placed. 1,528 respondents were asked this question. 1, 506. or 98 per cent, of the persons made at least one statement in response to the question "Who Am I?" Five persons made no statements, and one per cent of the responses could not be coded. All respondents did not make the same number of responses and some respondents made more than one response which could be coded within the same category. Of the 1, 506 respondents who made at least one self identi- fication statement, 28 per cent, or 420 persons, made statements that were classified as religious self identification statements; i.e.,. state- ments that could be placed in either category 48, 52, or 53. The distri- bution into the three categories was as follows: f '70 48 -Religiosity . . . . . . . . . . . . . 248 .59 52 -Religious Organization . . . . . . . . 113 .27 53 - Qualified Religious Organization . . . . 59 .14 For the purposes of the present analysis, however, all three of these categories were "lumped" together and designated as statements of religious self identification. This was justified on the basis that all of the statements had as objects religious activities, events, organizations and/or beliefs. Thus the Self Identification Index becomes, for the present investigation, a simple dichotomy based on the presence or absence of religious self identification statements in response to the question "Who Am I?" If the respondent made any 33 The instructions to the interviewer for presenting the Self Identification Problem and for recording self identification statements, and the coding instructions for the Self Identification Problem are appended to this thesis as Appendix A. 106 statement, classificable as 48, 52, or 53, irrespective of the order of his statement in the series of responses, and irrespective of the num- ber of such statements which he made in one or more than one of the three categories, he was classified as having made a statement of religious self identification. The validity and reliability of this observation technique were discussed in a preceding section. The observation technique meets all of the criteria of construct validity as set forth in that sec- tion. The technique also meets all of the criteria of intersubjectivity of observations as set forth. However due to the factithat this techni- que falls in the category of open-ended questions where the intersub- jectivity of manipulation of observations becomes a crucial issue, some additional comments are in order at this point. The initial coding of respondent protocols was carried out by the Gallup Organization following the instructions of the research team of which the present investigator was a member. 34 This initial group of coders (the Gallup Organization staff) was instructed to carefully follow the specified procedures for manipulation of these observations. This consisted of instructions and examples provided in the code book for the Self Identification Problem (SIP) . Their coding decisions, then, were taken as the base with which all subse- quent inter-coder reliability comparisons were made. 3 4The Gallup Organization, Inc. , was responsible for drawing and interviewing the sample of persons used in this study and also for all coding of all raw data from interview schedules. The cod- ing procedures were developed by members of the Five Nations Study research team. This research operation was officially headed by Charles P. Loomis, Research Professor, Department of Sociology, Michigan, State University. It is initiated under the direction of Fred B Waisanen and carried through under the direction of Robert L. Stewart, then both of the Department of Sociology, Michigan State University. The Coding Instructions for the Self Identification Problem were developed by Robert L. Stewart, Charles W. Tucker, Cay Bettinghaus, and the present writer. 107 Using the Gallop coding staff decisions as the reference point, a series of intercoder reliability checks were initiated. These inter-coder reliability checks were carried out with five judges and were based on a random probability sample of 150 protocols (or roughly 10 per cent) of the original 1, 528 schedules. Comparisons were made between the judgments of the original coding team and each of the five additional independent sets of judgments. Comparisons were made in two ways. One set of compari- sons was made of the separate categories of religious self identification statements to determine the extent to which all judges placed the same statement in the exact category as the Gallup Coding team, i.e., 48, 52, or 53. I will refer to this comparison as the "exact category" inter-coder reliability assessment. A second set of comparisons was carried out to determine the extent to which all judges agreed if a statement was simply a "religious self identification" irrespective of which of the three categories in which it was placed (i.e., it could have been coded as 48, 52, or 53) . I will refer to this comparison as the "combined categories" inter-coder reliability assessment. While both sets of comparison merit attention here, only the "combined categories" is of crucial importance for the present investigation. This is due to the fact that I was dealing with religious self identifi- cation statements in general and not with any one of the three different categories of religious statements in particular. Nevertheless. both sets of comparisons will be discussed as a demonstration of the ex- tent of intersubjectivity of observation manipulations which can be achieved with observations of this kind. Of the 150 protocols in the subsample 39, or 26 per cent, contained statements designated as religious self identification by the Gallup judges. Four protocols had two religious statements each and 35 . . . . . This series of intercoder reliability checks was initiated and supervised by Cay Bettinghaus, Department of Sociology, Michigan State University, 1964. ll 108 one protocol had three religious statements providing a total of 45 religious statements on which the inter-coder reliability comparisons were made. With 45 statements and five judges the comparisons were thus carried out on 225 judgments. In the "exact category" comparison there were 225 judg- ments and 29 errors; i.e., there were 29 discrepancies in judgment concerning the "exact category" into which the statement could have been placed when the Gallup team judgment was used as a base. This resulted in an error of 12. 9 per cent or a total inter-coder reliability assessment of 87.1 per cent for the "exact category" assessment. In the "combined categories" comparison there were 225 judgments and 11 errors among the five judges, using the Gallup team judgment as a base. This resulted in an error of 4. 9 per cent or a total inter-coder reliability assessment of 95.1 per cent for the "com- bined categories" assessment. The merits of the inter-coder reliability assessment must stand with this last category of "combined" judgments. These are the judgements which are directly relevant for the analysis carried out in the present investigation. The results of this assessment clearly indicate that statements made by persons in response to the SIP ob- servation technique can be dealt with across a number of observers with a high degree of intersubjectivity. Additional Comments on the Validity of the Self Identification Problem I have followed the procedures set forth in the earlier dis- cussion regarding the establishment of construct validity for this observation technique. It is important, however, to point out the as - sumptions that are made in the use of the "Who Am I?" technique. 36 I am indebted for many of the ideas here concerning the assumptions of the Self Identification Problem to my numerous and lengthy discussions with both Robert L. Stewart and Charles W. Tucker. Tucker has set forth many of these same ideas in an unpublished paper 109 The self is defined as the acts of naming one's own activities. Self Identification is defined as the class of designating or identifying behaviors that the person directs toward himself as an object. The observational procedures specified for observing this activity is the recording of the statements which the person makes in response to the question "Who Am I? " The respondent is asked to answer the question "Who Am I? " but he is asked to answer that question under special condi- tions. He is asked to make statements in response to the question "as if you were giving the answers to yourself, not to anyone else." Thus the person is asked to respond to the question "Who Am 1?", to make statements about himself, as though he were making those statements to himself. Thus, the initial assumption which is made about this technique is that the respondent, if he attends to the directions given him by the interviewer, is responding to or identi- fying himself as an object. This is the first and perhaps most crucial assumption which is made in using this observation tech- nique to obtain information about the self. The second assumption which is made is that the person "knows" who he is and that he can put this knowledge into words. McPartland states, in his Manual for the Twenty Statements Test, that " . . . respondents are confronted with the problem of identifying evaluating the methodological relationship between Manford Kuhn's "Self Theory" and the Twenty Statements Test (Self Identification Problem) . See his "Methodological Problems of Social Self Theory, " Department of Sociology, Michigan State University, 1964. 110 themselves and are left to decide for themselves how this identification will be made. "37 This assumption has direct relevance for the kind of pro- cedures which one can employ in making observations of the self and thus to the question of the validity of this procedure. This assumption explicitly precludes the use of any set of fixed responses to obtain information about self that is theoretically relevant given the orienta- tion within which the present investigation is outlined. The use of any group of pre-set response categories assumes that the person doesn't know who he is and thus the investigator must suggest some possibili- ties to the person. Or, it assumes that the investigator has knowledge regarding the respondent's self identification. 37Thomas S. McPartland, op. cit. Frank Hartung has done an excellent job of pointing out the limitations of the assumption that man doesn't know who he is or what his experiences are. Hartung illustrates this with the humorous but appropriate example of the projective technique which makes such an assumption. He uses as an example the Thematic Apperception Test of Henry Murray. The assumption with this technique is that the per- son does not know who he is or what the meaning of his experiences is. The person's responses to the TAT stimuli are called alpha protocol by Murray and these responses can only be interpreted and understood by an "informed expert." Said interpretation constitutes what Murray calls the beta protocol which is the "meaningful and real definition" of who the person is and what his experiences mean. Hartung points out that his own interpretation of the beta protocol produced by the clini- cian makes the latter protocol an alpha protocol and his own interpre- tation ( i.e. , Hartung's) the beta protocol. Then as the reader surveys Hartung's interpretation it becomes an alpha protocol in light of the reader's now established beta protocol, and on the interpretation can go to infinite absurdity. This is the kind of open-ended testing absurdity that precludes the establishment of any sort of intersubjectivity of ob- servation or manipulation of observations. See Hartung's "Manhattan Madness: The Social Movement of Mental Illness, " The Sociological Quarterly, Vol. 4, No. 3 (Summer, 1963), pp. 261-272. Several classes of observation techniques have been em- ployed to obtain information relevant to the self. Most are of the fixed- response variety. Perhaps the most frequently used technique is that 111 The notion that the person does not know who he is is clearly contradicts Mead's conceptualization of the self. Consequently, pre-set response categories such as adjective check lists are out of the question as legitimate techniques of observation for any investigation of the self which is outlined, carried out, and interpreted within the Meadian framework. In addition, a legitimate question can be raised about the necessity of doing research in the case of the investigator who presumes knowledge about the re spondent's self identification be- fore the fact. If such knowledge is in the possession of the investigator the research procedure is repetitious and wasteful. The above considerations lead to the statement of the third assumption made when one employs the question "Who Am I? " That assum tion is that the uestion "Who Am I?" doesn't su est answers P q 88 of having persons characterize themselves by checking words on anad- jective checklist of self-descriptive terms. These range from the simple adjective checklist to the use of the semantic differential. See, respectively, Richard Videbeck, "Self Conceptions and the Re- actions of Other," Sociometry, Vol. 23, No. 4 (December, 1960), pp. 351-339, and M. Helper, "Learning Theory and the Self Concept, " J. of Abnormal and Social Psychology, Vol. 51, 1955, pp. 184-194. Other means of making observations include the use of responses to some standard personality inventory such as the MMPI, or, having persons make a Q-sort of "common" responses to TAT cards. Here see, respectively, Richard M. Lundy, "Self Perceptions and Descriptions of Opposite Sex Sociometric Choices, " Sociometry, Vol. 19, No. 4 (December, 1956), pp. 273-277, and I. Friedman, "Phenomenal, Ideal and Projected Conceptions of Self, " Journal of Abnormal and Social PsychologL Vol. 51, (1955), pp. 611-615. Finally, several investigators have employed the adjective check list but have set this up so that the subject makes judgments of himself in conparison with others. Most notable here is Frank Miya- moto and Sanford Dornbusch, "A Test of Interactionist Hypotheses of Self Conception, " g, Vol. 61 (March, 1956). PP. 399-403, and John J. Sherwood, "Self Identity and Referent Others, " Sociometry, Vol. 28, No. 1 (March, 1965). PP. 66-81. 112 to the respondent, and all the answers that are given are the respon- dent's own statements of self identification. The forth and last assumption that is made in using this technique is that the question "Who Am I?" is general enough, and the phenomena of self and self identification are general enough, that 39Admittedly, the question "Who Am I'" may suggest certain forms of response with respect to subject -predicate patterns. There is no evidence to indicate that it suggests responses with respect to the object of the self identification statement. With respect to the sugject- predicate pattern, not all responses employ the verb "to be. " The majority of responses do employ that infinitive. That the form of the question affects the form, but not the context, of the response can be illustrated by the different subject -predicate patterns which are obtained when the investigator asks the respondent to give twenty statements about himself beginning with the work "I. " Here, the responses more frequently take the predicate form of possession or location verbs, doing verbs, at- traction and aversion verbs rather than simple "being" verbs. For a report of one such investigation which demonstrates the relationship between question form and response form see Carl J. Couch, "Aliena- tion and Self Identificatiion, " a paper read at the American Sociological Association Annual Meetings, Los Angeles, California, 1963. The Self Idenfication Problem, or Twenty Statements Test as Kuhn labelled it, was initially developed for the explicit purpose of avoiding suggestions to the person as to the appropriate statements he might make. As I have pointed out, above in the text, the "Who Am I?" technique simply presents the person with the task of identifying himself. How he does it is up to him. That other techniques persent suggested self descriptions of self identification, e. g., adjective check lists, is their most obvious handicap. Kuhn and McPartland pointed this out in their initial discussion of the "Who Am I'" technique. In a footnote they cite Theodore Newcomb's comments on this issue. Newcomb is quoted as saying: "The weakness of direct questions is that they provide no wayof measuring the salience of an attitude; we never know whether the attitude would have been expressed at all, or in the same way, apart from the direct question. " Newcomb, Social Psychology, N. Y.: Dryden, 1950, p. 151. Kuhn and McPartland were concerned with self attitudes rather than self identification behaviors but the arguement which Nucomb presents is relevant to any form of behavior. (I have simply attempted to avoid the use of the word attitude because of the contemporary conno- tation of that word as opposed to a more behavioral connotation. ) Kuhn and McPartland note: "Thus, when a respondent in reply to the "Who Am I? " question . . . writes "I am a man, " "I am a student, " "I am a football player, " it is reasonable to believe that we have a far more solid know- .Pl. \tv. ~ A)» ~ .-~ 113 it does not elicit responses which are particular to a limited situation; e. g. , the testing situation. Careful consideration of the question makes it clearly apparent that it is possible for the person to include in his statements past activities and not just situational activities. The above assumptions are made in line with the assertion that the observation procedures involved in the use of the Self Identifi- cation Problem are logically derived from and consistent with the con- cept of self identification as defined and with the theoretical orientation within which self identification is being considered. Attention is now directed to a brief discussion of the modification in procedure for em- ploying the technique in this study as compared with previous procedures. Modifications in the Conventional Use of the Self Identification Problem The manner in which the present investigator has used the Self Identification Problem deviates from the way in which it has been conventionally employed. The technique was first used by Kuhn and his students as a paper and pencil test.40 Respondents were handed a piece of paper with a set of written instructions at the top, followed by 20 numbered lines. The instructions have conventionally been as follows: There are twenty numbered blanks on the page below. Please write twenty answers to the simple question, "Who Am I?" in the blanks. Just give twenty different answers to this question. Answer as if you were giving the answers to yourself, not to somebody else. Write the answers in the order that they occur to you. Don't worry about logic or importance. Go along fairly fast, for time is limited.41 1eEdge of the attitudes [behaviors] . . . than if, on a checklist and among questions we had asked "Do you think of yourself as a man?", "Do you 1ihink of yourself as a student? ", "Do you think of yourself as an athlete? " KUhn and McPartland, op. cit., pp. 72-72, footnote 10. 40 . . . Ibid. For a list of studies employing the technique in this rmanner see footnote 10, Chapter I of this dissertation. 41 Mc Partland, op . cit . 114 The discrepancy between this set of instructions and those given at the beginning of the previous section are fairly obvious. The manner in which the instructions were presented and the setting in which the question was asked also differ. The problems involved in this deviation away from the conventional use of the Self Identification Problem should be spelled out. Perhaps the most obvious deviation is that the question was asked orally with a set of instructions provided by an interviewer (rather, by several interviewers) in a variety of different interview situations. The problems involved in this deviation away from the conventional use of the Self Identification Problem should be spelled out. Perhaps the most obvious deviation is that the question was asked orally with a set of instructions provided by an interviewer (rather, by several interviewers) in a variety of different interview The present study is not the first time the SIP has been Mul- situations . employed in an interview or survey research type of situation. ford and Salisbury report the use of the technique in a quota sample of the adult population of the state of Iowa. 42 This simply provided the respondent with a brief set of oral instructions and then handed him a pencil and piece of paper with the question "Who Am I?" written at the top of the page. "The sheet was otherwise blank, and the respondent was asked to record his own answers. The instructions Were designed to encourage the subject to respond as spontaneously and unreflectively as possible and they were repeated verbally by the interviewer. Interviewers were instructed to give the respondent as much time as he needed, but to proceed with the remainder of the 42 Harold A. Mulford and Winfield W. Salisbury II, "Self COnceptions in a General Population, " Sociological Quarterly, Vol. 5, No. 1 (Winter, 1964) pp. 35-46. 115 interview if the respondent seemed unable to give further answers to the que stion . ” Thus, Mulford and Salisbury also used the Self Identification Problem in an interview situation but modified the conventional pro- cedures of presentation. The reasons which they present are relevant to the modifications employed in the present study. They state: These procedures differ from those followed in past studies in that previous investigators numbered the lines and gave the respondent six minutes to answer. They also urged the re- spondents to write twenty statements, or at least as many as as possible in the allotted time. While the effects of these procedural differences are unknown, we were of the opinion that omitting the line numbers and not insisting upon the com- pletion of a specific number of statements would reduce the danger of antagonizing the respondent. However, the new procedure may also account for the relatively small number of statements obtained in the present study."‘4 The procedures employed by Mulford and Salisbury had the effect, at least, of reducing the number of statements given by the respondents. The mean number of statements was 4.4 and the median number was 4. Eighteen respondents gave no response and fifty-nine respondents gave ten or more responses. The procedures employed by the present investigator also had the effect of reducing the number of statements although the average number in the present study was higher, where the respon- dent gave verbal statements, than in the paper and pencil method used by Mulford and Salisbury. In the present study the mean number 0f statements was 5.23. Only 5 of the 1, 528 respondents refused to answer the question. 43 , Ibid., p. 36. 44 Ibid. 45 . . . N This modified set of procedures for asking the question Who Am I?" was described in detail in the section on Specification of Operational Procedures in the present chapter. The modification Was originated by Professors Fred Waisanen and Hideya Kumata and Was further revised to the present form by Professor Robert Stewart, all of the Department of Sociology, Michigan State University, during 1963-64. l 16 Perhaps a more serious modification of the conventional usage of the Self Identification Problem was the introduction of the words "most important statement about yourself" in the instructions to the respondent. I do not know what the effect of such a modification is. This particular instruction followed the major portion of the instructions which are, as the reader may plainly see, quite com- parable to the conventional instructions. The procedures were also modified in terms of including the instruction to the respondent to "make the most important state- ment first, " and in terms of probing if the person did not make a spontaneous series of statements (vis. , "Now make what you con- sidered to be the next most important statement about yourself. ") Both modifications, but particularly the latter, were introduced in order to get as many statements as possible from the respondent in the interview situation. The Self Identification Problem as placed near the front of the interview schedule to avoid contamination of the responses. That preceding question-answer exchanges between the interviewer and respondent do contaminate the number and kind of statements made by the person is an assumption. Conventionally the Self Identification Problem has been presented prior to other questions involved in the investigation in an attempt to preclude non-spontaneous statements; i.e., statements suggested by the content of the prior questions. This is an assumption which has not been verified. The only kind of prior question which could contaminate Statements of self identification relegant to the present study, if the as Sumption of contamination is valid, would be statements having to do with religion. Unfortunately,such a question was asked prior to the Self Identification Problem in the present study. The respon- dent was asked to place himself along a continum regarding the im- P°rtance of religion in his everyday life. This question about religion WaS preceded by several questions, in order to establish rapport with 117 the respondent, and then was followed by seven additional questions prior to the introduction of the Self Identification Problem. (These additional questions did not, however, deal with religion.) The re- ligion question should have, according to the contamination assump- tion have had considerable effect on the percentage of persons making religious self identification statements. Mulford and Salisbury's study, which had no religious question or any other content question prior to the Self Identification Problem report that 25 per cent of their respondent's made religious self identification statements. In the present study, where a religious question was asked prior to the Self Identification Problem, 28 per cent of the respondents made religious self identification statements. The difference between the per cent of persons making this category of self identification state- ments is not significant and this modification from the conventional procedure does not seem to have had effect.46 The consequence of the other modifications, discussed above, is not known. The statements made in response to the Self Identification Problem do not differ in kind, Lag, from the types of statements which are made in response to the conventional pro- cedures of the Self Identification Problem format. The limitations introduced by the modification of procedure with respect to question wording, the number of statements sought, and the request for "most important statement" are simply not known to this investi- gator. The limitations in the number of statements acquired are ‘ 6I'examined the relationship between per son' indica- tion of the importance of religion in their everyday life and the pre- Sence or absence of religious self identification statements. The Chi square test for independence between quartile values for the importance of religion continum and the presence or absence of reli- gious self identification statements yielded a value of x = 112. 07, 3 degrees of freedom, p. .001. (The chi square value necessary to reject the null hypothesis for a one -tailed test consists of all values of x2 equal to or greater than 6.25.) The extent of association between importance of religion and religious Self Identification 118 relatively obvious, at least in one sense; you simply get fewer pieces of information concerning your involved hypotheses and therefore this modification may work against the investigator rather than his statements, measured by Cramer's V, is V = .32. Despite the fact that Importance of Religion in Everyday Life and Religious Self Identification are highly related it remains the case that the presence of this religious question prior to the SIP had a negligible effect in terms of increasing the per cent of religious Self Identification statements made in this study. As I point out in the text above, Mulford and Salisbury report that 25 per cent of their re- spondents made religious Self Identification statements and in their study the Self Identification Problem was the first question employed. In comparison, 28 per cent of the respondents in the present study made religious Self Identification statements even though a question on religion preceded the SIP. As I mentioned in the text, seven additional questions not dealing with religion were asked of the respondent after the reli- gion question and prior to the SIP in the present study. All of the questions asked the person to place himself along a continum repre- sented by the picture of a ladder with eleven rungs. The first question asked the person to indicate the extent to which he had the opportunity and chances to do anything he wanted to do. The second asked the person to indicate the extent to which he could do very much about making his present life happier. The third question asked the person to place himself, figuratively, on the middle rung of the ladder and to indicate if his community would rank higher or lower than himself on the ladder in terms of importance. The fourth .through seventh questions asked for similar comparisons of importance between the person and his family, his country, his occupation, and his political party, respectively. Thus a variety of different questions intervened between the Importance of Religion question and the SIP problem. Finally, and most important, it should not be surprising that this moderate association of V = . 32 should exist between Impor— ta.nce of Religion and religious Self Identification statements. To say that the person who identifies himself as religious also regards reli- gion of importance in his everyday life does not contradict any of the Considerations of the theoretical orientation within which the present Study was couched. However, to say that the person who indicates that religion is important to him will consequently identify himself as a. religious person, is a meaningless statement of relationship as far as explaining or pointing to the basic process which produces and maintains person's conceptions and identifications of themselves. ThOse who would take such a tack must defend their "explanation" by 119 favor. However, this may be counterbalanced by asking the respondent to give the statements about himself which he considers most important. This may work for or against the investigator. I do not know. All of the modifications taken into consideration, given the knowledge that we do and do not have about their effects, the responses made are quite relevant statements of self identification. These observations are those which should be produced with the Self Identification Problem technique. Therefore, I will assume that the modifications have not produced undue bias on the observations acquired with this technique and will put them to the appropriate use in the present investigation. Discussion of the Sample and the Issue of Sample -Universe Relationships The sample of persons on which the present study was based was constructed and drawn by the Gallup Organization accord- ing to a modified area probability sampling design. Such a design is calculated to represent the U. 5. adult population living outside of institutions, by household units . systematically accounting for the basic processes which causally con- nect an indication of the importance of some aspect of persons' lives to the production and maintenance of self identification statements in terms of that aspect. 47 . . . . - The observations were based on personal interViews With an original sample size of l, 528 persons, 21 years or older. The de- Sign of the sample was that of a probability sample down to the block 1evel in the case of urban areas, and, to segments of townships in the Case of rural areas. After stratifying the nation geographically and by size of community in order to insure the conformity of the sample With the latest available estimate of the Census Bureau of the distri- bution of the adult population, 143 different sampling points or areas We re selected on a strictly random basis, with probability of selection Preportional to population size . Approximately 10 interviews were conducted in each such 1'andomly selected sampling point. In each of these points, approxi- Inately five interviews were conducted. The interviewers had no choice 120 The use to which I will put the data gathered in this sample does not correspond to the universe represented by the characteristics of the non-institutional adult population nor does it correspond to the kinds of restrictions upon which an area probability sample is drawn. It is inappropriate to employ such a sample to generalize to such a universe based on the arbitrary criteria by which such samples are customarily drawn unless the phenomena of concern to the investigator are related to those criteria. This is to say that the criteria by which such sampling designs are constructed do not resemble the phenomena specified as important by the theoretical orientation Within which this investigation was developed. The criteria by which modified area probability samples are constructed are primarily socio -economic ones. An attempt is made, following such a procedure, to guarantee that certain educational, occupational, income and dwelling area groupings are represented in correspondence with those attributes among the U. S. adult pOpulation living outside institutions. These are relevant criteria for studies of phenomena for which such socio- economic attributes are of consequence. They are not of consequence, whatsoever concerning the part of the city or county in which they con— ducted their interviews. Interviewers were given maps of the area to which they were assigned, with a starting point indicated, and required to follow a specified direction. At each occupied dwelling unit, inter- viewers were instructed to select respondents by following a prescribed Systematic method and by a male -female assignment. This procedure Was followed until the assigned number of interviews was completed. AS mentioned in the text, this sampling procedure was designed to pro- duce a sample which approximated the adult civilian population (age 21 ancl older) living in private households. in the United States (that is, ex(:luding those in prisons, hospitals, hotels, religious and educational institutions, and military reservations. ) The description of the sampling procedures is taken from the report of the Gallup Organization. See their "Report to the Depart- Inerit of Sociology and Anthropology, Michigan State University, " March, 1964, GO/6368 GP, (30/6369 RM, and GO/637O sw, The Gallup Organi- za«tion, Inc., Princeton, New Jersey. This is available in the Depart- ment of Sociology, Michigan State University. 121 however, and therefore are irrelevant to any consideration of non- socio-economic self-other relationships except by the most remote sociological juxtaposition. The nature of the theoretical system I am employing would hold that irrespective of the collection or the number of persons examined, the proposed relationship between the specified religious context of experience and behavior, and, self identification would take the same form. I do not have any basis for testing such an assertion and I do not wish to imply any such basis. However, I do have a large number of persons which were drawn and observed without any planned bias. This last qualification--or more appropriately, lack of qualifi- cation--is quite relevant and should be given full consideration. 8One argument which might be raised against this would be as follows. Consistent findings indicate that the number of volun- tary membership associations a person belongs to is related to his socio-economic standing. Assuming that the greater the number of associations belonged to is an indication of the frequency of partici- pation in voluntary membership associations the argument would be: (1) the higher the socio -economic standing the greater the participation in voluntary membership associations; (2) the greater the participation in voluntary membership associations the greater the likelihood that others can direct identifying behaviors toward the person correspondent to the activities of the particular association or associations; therefore. ( 3) the higher the socioeconomic standing the greater the likelihood the person will identify himself in terms of the voluntary membership a s sociations . The fallacy of the argument should be obvious on a number 0f grounds. First, research on voluntary membership associations usually does not include all such associations; e. g., union and religious group membership. Second, it is probably not legitimate to argue that belonging to a number of groups is the same thing as participation in 13heir activities. (In fact, this may be one limitation of the data on group association in the present investigation) . Third, it would be necessary to specify the frequency of association in particular group group and then examine the relationship between the presence or ab- SeIlce of identification statements in terms of the activities of those Particular groups. Fourth, given the specification which must take Place regarding each of the terms in the above syllogism, the conclu- sion is misleading. There is no theoretical justification for saying that 122 Therefore, rather than considering this sample of persons as "a random sample from the hypothetical universe of all possible samples which could have been produced under similar conditions of time, place, culture and other relevant factors", 49 I will consider the sample as a random sample from the hypothetical universe of all such possible rela- tionships between the specified context of experience and behavior and self identification behaviors. Thus, I will be considering the sample of persons who did and who did not make religious self identification state- ments as two independent random samples from the two hypothetical universes of all persons who do and who do not make religious self identification statements in relation to their implication in religious contexts of experience and behavior. That the persons themselves would be different in various samples drawn from this hypothetical universe is irrelevant. Any sample of persons simply provides the basis for determining the presence or absence of the relationship speci- fied by the theory and the extent of association regarding those rela— tionships . Mead's discussion of the ontological assumptions in science has direct relevance to the problem of samples and universes. He states: . . . the logic of scientific method applies to the solution of some immediate problem in which the test of experience as given in observation and experiment refers not to such an ultimate uni- verse i.e., a universe transcending scientific theory and inde- pendent of observation but to the world of experience in so far as this is not involved in the actual problem with which the scien- tist is occupied. The experimental test can never be in terms of m 5.0cio -economic standing should be systematically related to self identi- f1-<:-.‘—3.tion in terms of, e.g., religious groups, unless all of the above mentioned conditions were specified and present. 49Margaret Hagood and Daniel Price discuss the issue of hYIDOthetical universes in a similar manner. See their Statistics for Sociologists, N. Y.: Holt, Rinehart and Winston, 1952, rev. ed.. PP- 419-423, and, p. 293. 123 a universe which is and which will be subject to no reconstruction. What is approached in observation and experiment is as such an abstraction from the perceptual object and its world that it will be translatable into any hypothetical universe . 50 Thus, the world of immediate experience and observation with which the scientist must deal is viewed by Mead as an abstraction from, or one can say a sample of, the hypothetical world of all such possible observations of the phenomena. Certainly, this turns on Mead's notions of the universe of social processes which must be assumed by the scientist before he can proceed in the formulation and investigation of a problem. Mead points to two aspects of the scientific assumption. One is that the hypothesis, when properly formulated, must be universal in its application--the assumption of the uniformity of nature. It is the assumption upon which rationalism has always built. The other is the assumption upon which empiricism has rested: that it is possible to so state a hypothesis that it can be tested for its applicability by experiment or observation. There is an obvious relation between the two. It is only upon the assumption of the uniformity of nature that one can advance to the reliability of a hypothesis from the results of observation and experiment. In his discussion of the "universality of relationships, " the limitations of directly observing only the data of immediate experience, and, reliability, Mead explicitly rejects the notion of a real and trans - Cending universe which provides a permanent set of data. He suggests that "these data of immediate experience . . . are certain abstracted agreements and disagreements of certain characters, certain succes- Sions of such characters. " 50 Mead, Philosophy of the Act, p. 278, 51 , - Ibid., p. 275. 52 Ibid., p. 277. 124 Thus, Mead is arguing that the investigator imposes certain uniformities, or contingent statements of relationships based on the specifications of his theory, onto an assumed universe of data which is independent of any particular observer. 53 One cannot make all possible observations and thus makes only a limited set of observations. Insofar as the investigator procedes in the selection of that set of observations to allow for the equal opportunity of the rejection or confirmation of his statements of relationship, the so-called representativess of the set of observations is not generic to that particular class of observations. Rather, based on a theory which specifies a universal state of such relationship, the investigator may abstract from the particular set of observations to the hypothetical universe of all such possible observa- tions which, his theory indicates, should manifest the universal rela- tionship. Mead states: The experimental datum, its scientific form, on the one hand, and its factual location, its imbeddedness, in the different perceptual worlds in which it may be realized, on the other, constitute the real problem disguised in the assumption of a per- ceptual universe which is ultimate and independent of the per- ceiving individual. Its ultimate character is expressed in the assured reference to a world of experience that is not questioned in the immediate problem with which the scientist is occupied. Its independent character is expressed in its universality which implies its applicability to any hypothetical situation and abstraction from any specified perceptual world. 5 Analysis Procedures The major hypothesis in this study states a relationship between the referents for a multivariate construct, and, the presence 01‘ absence of a specified class of activity. The proposed relationship takes the form such that an increase in the referents, either separately g 53Ibid. 54 Ibid., pp. 278-279. 125 or in combination, should result in an increase in the proportion of persons taking the specified class of activity. The analytical pro- cedures necessary to deal with this hypothesis will require examina- tion of each of the separate relationships as well as the various forms of the combined relationship between the referents for the multivariate construct and the presence or absence of the specified class of activity. I will refer to the component referents for the construct RCEB as independent variables and to the referent for self identification as the dependent variable. Due to the level of measurement established for the inde- pendent variables (nominal and ordinal) and for the dependent variable (nominal) it is obvious that the analysis of variance technique cannot be employed. Some analogue to analysis of variance would be desir- able which would provide information concerning: the main effects, (i.e. , the overall relationship between all independent variables and the dependent variable); the simple effects, (i.e., the separate rela- tionships between each of the independent variables and the dependent variable); and, the interaction effects, (i.e., a specification of the extent of the contribution of each of the separate independent variables to the overall relationship with the dependent variable) . Such an analogue to analysis of variance is not available given the levels of measurement established for the variables in the present study. Therefore, the next most appropriate procedure which can yield Similar information will be employed. The analytical procedures will consist of bivariate and multivariate contingency tables examining the association between the RCEB components, separately and in combination, and the presence or absence of religious self identification statements. First, I will examine the relationships between each of the RCEB component independent variables and the dependent variable of Self Identification. I will pro- ceed in two ways: (1) I will determine the presence or absence of ass ociation by means of chi square and if association exists, will 126 determine the extent of that association by means of Cramer's V; (2) I will examine the percent of cases which fall in predicted cells in each of the bivariate tables. Here I will be concerned with the per cent of cases in the extreme diagonal cells of each table. This will be the per cent of persons making religious Self Identification statements in the maximum RCEB cell and the per cent of persons making no religious Self Identifi- cation statements in the minimum RCEB cell. (Hereafter, I shall refer to these respectively as MAXRCEB and MINRCEB.) By example see Figure 1 for the association between RHOMOG and Self Identification. Figure l. --RHOMOG and Self Identification Self Identification RHOMOG RSI RSI 4 (70 R51 MAXRCEB 3 Z 1 % RSI MINRCEB Each of the specified cell percentages, MAXCREB and MINRCEB, should approach 100 consistent with the specifications of the hypotheses. A convenient index for examining the per cent of all cases accounted for in these two predicted cells can be established in the following manner. First, add the marginal totals for the two rows involving the MAXRCEB and MINRCEB corner cells. This will give the total number of cases which should appear in the cells if each rea.ches 100 per cent. Second, add the observed frequencies in the two cells. The latter total divided by the former total gives the total 127 per cent of cases accounted for in all predicted cells. I will refer to this as the SI % Index. Second, having examined the bivariate relationships between each of the RCEB components and Self Identification, I will examine suc- cessive multivariate relationships between the various combinations 0f RCEB components, and, Self Identification. I will do this by introducing additional RCEB components into the contingency table and examine those components together against the dependent variable of Self Identi- fication. Conceptually, this is simply a matter of increasing the number of RCEB observations considered simultaneously in relation to the presence or absence of Self Identification. Statistically, this is simply a matter of examining the independence of the two groups of persons who do and who do not make religious Self Identification statements with respect to their proportionate distribution on the increased multiple values of the RCEB polytomy. 55 If association exists then the appropriate measure of extent of association may be applied; in this case the measure will again be Cramer's V. 55Ordinarily, chi square is employed as a test of independence between populations as their independent and randomly selected samples of observations are distributed on the values of one polytomy with respect to their distribution on the values of a second polytomy; i.e., it is ordi- narily used in the bivariate case. This statistic may also be used in the multivariate case; i.e. , to examine the independence of populations on one polytomy with respect to their distribution on the combined values of multiple polytomie s . The chi square value for any contingency table is determined by taking the square of the difference between observed and expected frequen- Cies in each cell and dividing by the expected number of cases in each cell. As chi square has the property of additivity the total chi square value for any table is obtained by summing across all cell chi square values. This Obtained value is then examined in relation to the degrees of freedom for the contingency table in order to determine the independence of the popula- tions with respect to the variables under consideration. The degrees of fr'Eiedom simply depend on the number of cells in the contingency table. F03? the general case of the r x c contingency table the degrees of freedom may always be determined as (r - l) (c - 1). This very brief discussion should make it quite clear that the 128 For example, I may examine the relationship between RHOMOG-together-with-MIG, and, Self Identification. If there is asso- ciation between RHOMOG and Self Identification, and association between MIG and Self Identification, there should also be as association between criteria and assumptions of chi square do not preclude its use as a test of independence for any r x c contingency table, the multivariate case included. The degrees of freedom are determined by the total number of cells in the table and the total chi square value is simply the sum of the cell chi square contributions for the table. There are no specifications or criteria which determine the manner in which the cells on two or more polytomies under consideration must be arranged or ordered. Chi square is, to repeat, a statistic which determines the extent of discrepancy between the observed and expected frequencies for the total number of cells in the contingency table. Should the hypothesis of independence be rejected the investi- gator may then wish to determine the extent of association between the variables under consideration. Measures of association should meet at least two criteria when a large number of contingency tables are being examined. They should allow for comparison of the extent of association across contingency tables and they should be measures which can reach unity as an upper limit and either zero or -1. 00 as a lower limit. Two conventional measures, the phi coefficient and Pearson's Coefficient of Contingency - C, can not be employed in the present study for one or both of the above reasons. The phi coefficient can only be employed and com- pared with the 2 x 2 table and C can only be compared with r .= c, or square, contingency tables. Furthermore C cannot reach unity and it is therefore always necessary for the investigator to compute a Cmax and then compare that value with the obtained C value. Such a ratio can only be determined with r = c tables. Cramer's V, a measure of association based upon chi square, avoids the limitations of both C and phi. Goodman and Kruskall point out: "Cramer suggests the . . . [variant of C,] V . . . which gives better norming than does C . . . since it lies between 0 and l and actually at- tains both end points appropriately. Cramer's suggestion does not seem to be well known by workers using this general kind of index. " Leo A. Goodman and William H. Kruskal, "Measures of Association for Cross Classifications, " Journal of the American Statistical Association, Vol. 49 (December, 1954), p. 740. Cramer's V is originally reported in Harald Cramer, Mathematical Methods of Statistics, Princeton University Press, 1946, as cited in Goodman and Kruskal, op. cit. V may be compared across similar size contingency tables due to the manner in which the divisor figures into the V value quotient. The formula is reproduced below: 129 RHOMOG -together-with-MIG, and, Self Identification consistent with the major hypothesis. Furthermore an increase in the percent of cases ac- counted for in MAXRCEB and MINRCEB should also be reflected in this 2 where Min (r - 1, c -1) is equal to 2 the number of rows minus one or the number of columns minus one which- ever is smaller. N-Min(r=l,c-l) Thus the V values can be compared in the case of any number of 2 x k contingency tables and of course in the case of all r = c tables. The majority of bivariate tables and all multivariate tables in the present study are 2 x k. They involve two columns, the dichotomous values of RSI and RSI for the dependent variable, and k number of rows for the in- dependent variables( 5) in both the bivariate and multivariate case. I have taken the computational formula for Cramer's V, above, from Hubert M. Blalock, Jr. Social Statistics, N. Y.: McGraw-Hill Book Co.. Inc., 1960, p. 230. Briefly stated, and using the RCEB components illustrating Figure 2, the chi square statistic used in the multivariate case tests for the independence of the two populations from which the samples of per- sons are drawn who do and who do not make religious Self Identification statements with respect to their proportionate distribution on the values of the combined polytomies of RHOMOG-together-with-MIG. Obviously, the measure of association, V, based on the obtained chi square value says nothing of the RHOMOG x MIG association nor their respective contributions to the overall extent of association. The V value simply reflects the extent of association between RHOMOG-together-with-MIG, and, Self Identification. The relative contribution or consequence of RHOMOG and MIG must be determined by other analytical procedures which will be discussed in a subsequent portion of the text. That V may be used in such a manner is made clear by Good- man and Kruskall's comments on measures of association for the multi- variate contingency table. "When there are more than two polytomies one may well be interested in the multiple association between one of them and all the others. One simple way of handling this in the unordered Case will be described here for three polytomies A, B, and C . We suppose that the multiple association between A e. g., Self Identifi- cation and B-together-with-C RHOMOG-together-with-MIG is of interest. Simply form a two -way table whose rows represent the A polytomy and Whose columns represent all combinations of Bb, Cc and then apply the a~Ppropriate two -polytomy measure . . . Note that this procedure does not take the B x C association into account. There is a rough analogy he re with the motivation for the standard multiple correlation coefficient 130 multivariate relationship of RHOMOG-together-with-MIG, and, Self Identification. The MAXRCEB and MINRCEB cells, in this multivariate relationship are illustrated in the hypothetical table presented in Figure 2. Figure 2. --Hypothetical 2 x k Contingency Table Examining Relationship between RHOMOG-tegether-with-MIG, and, Self Identification Self Identification RHOMOG MIG RSI E51 TOTAL (70 R51 R MAXRCEB ”Wu 4 E R 3 E R 2 i R. 1 E 070 RSI rown MINRCEB TOTAL col col N n n ‘ 0f normal theory. The standard multiple correlation coefficient may be (and often is) motivated by defining it as the maximum correlation co- efficient obtainable between a given variate and linear combinations of the other variates. That is, it is a measure of association between a given variate based upon all other variates . " Goodman and Kruskall, 02. cit., pp. 761-762. 131 Similarly, the extent of association and per cent of cases ac- counted for in MAXRCEB and MINRCEB should increase as additional RCEB components are added into the multivariate contingency tables. This should be reflected in the SI % Index. Having examined the maximum multivariate relationship between RCEB and Self Identification, (i.e., all RCEB components in combination, and, Self Identification) I will then attempt to determine which of the RCEB components, considered simultaneously, is of greatest consequence in the overall relationship between RCEB and Self Identification. The usual manner of dealing with such a problem is by a partialling procedure wherein the investigator examines each set of possible relationships between independent variables and the dependent variable, controlling on each independent variable while examining the relationship between the remaining independent vari- ables and the dependent variable. This is a legitimate procedure and is appropriate to the present investigation. However, the yield of information with such an approach, both in clarity and in parsimony, leaves a great deal to be desired in relation to the effort in the con- struction and analysis of the tables as well as the difficulty of clear and parsimonious presentation and interpretation. More important, such an analysis only allows the investigator to indicate, upon com- pletion, that the various independent variables being considered in relation to the dependent variable stand in some sort of ordered or rank relationship; i.e.. variable A is of more consequence than vari- able B which is of more consequence than variable C, etc. This par- tialling procedure, like any multivariate analysis procedure excepting analysis of variance, does not allow the investigator to determine the extent of each component variable's contribution to the overall rela- tionship between the independent variables and the dependent variable. Thus, any procedure employed on the data considered in this investi- gation can only indicate the relative contribution of the various inde- pendent variables to the overall relationship. 132 An alternate set of procedures, in place of the multi- variate partial technique described above, would be desirable. Those procedures would have to allow the investigator to make the same determination regarding the relative contribution of component vari- ables but would have to facilitate that determination by allowing greater parsimony and clarity. An alternative procedure is available which meets all of these criteria and that procedure will be employed in this study. The reader will recall the discussion of Guttman's formal model of scale analysis in an earlier section of the present chapter. Ordinarily this analytical procedure is used by the investigator to determine the unidimensionality of a set of behaviors drawn from some universe of behaviors of interest to him. Such a procedure was fol- lowed in the construction of both RBEL and RPART. In establishing the unidimensionality of a set of behaviors, moreover, this analytical procedure provides the investigator with information concerning the ordered relationship between items, between persons, and between items and persons. One of the most crucial pieces of information provided by this technique is the designation of that behavior, from among all those being examined, which is of greatest consequence. To use Guttman's terms, interspersed with by own to indicate the relevance of the technique to the present problem, Two basic steps are involved in this procedure . . . First, the questions [behaviors] are ranked in order of "difficulty", with the "hardest" question [most consequential behavior] i.e. , the one that fewest persons got right, placed first and with the other questions [behaviors] following in decreasing order of "difficulty." [consequence] Second, the people are ranked in order of "knowledge" [extent of implication in in behaviors] with the "most informed persons", i.e.. those who got all questions right, place first, the other individuals following in decreasing order of "knowledge. " [implication in the behaviors] These two steps are the basic procedure for the scalogram . . . technique of scale analysis.56 5 6Guttman, op. cit., p. 70. 133 If a scale is present, i.e., if the items or behaviors can be arranged in an internally meaningful rank order, the pattern repre- sented in Figure 3 will occur. Figure 3. --Perfect Half-Parallelogram Pattern for Guttman Formal Scale Model Total Behavior Behavior Behavior Scale Score A B C Type Errors 3 X X X 4 O 2 O X X 3 O l 0 0 X 2 O o o o T l 0 Here we see the perfect half-parallelogram pattern for a Guttman Scale with three behaviors ranked from left to right in terms of their importance or consequence, and with persons ranked from top to bottom in terms of the extent of their behavior as indicated by the sclres ranging from zero to three. 57 An X indicates that the person took the behavior and a 0 indicates that he did not take the behavior. In this case, with a perfect scale, the investigator could say that if the person engaged in Behavior A he will also engage in Behavior B and Behavior C 100 times out of 100, and thus the C"R" - l. 00. Therefore, if the person engaged in the most consequential behavior he will engage in the next most consequential behavior as well as the least consequential behavior. However, if the person engages in the next most 57I will refer to the arrangement of behaviors from left to right as the order arrangement. In the case of Figure 3 the order arrangement is ABC. The reader can quickly determine that seven other order arrangements are possible with the three hypothetical Behaviors of A, B, and C. 58 I Will refer to the person's pattern of response to the par- ticular order arrangement of behaviors under consideration as his {gsponsepatterm In the case of Figure 3, the response pattern for the person with a total score of 3 is XXX. 134 consequential behavior he will engage in the least consequential behavior but not in the most consequential behavior. In this case Behavior A is most important, or of most consequence, Behavior B is the next most important, etc. Thus, the use of this analytical technique will provide the investigator with a clear and parsimonious means of determining and of demonstrating the relative importance of any series of behaviors. The technique allows the determination of the behavior, among all those under consideration, which is of greatest consequence relative to the other behaviors. In the present investigation it is possible to examine the relationship between each of the RCEB components to determine which component variable is of greatest consequence by fitting the various order arrangements to Guttman's formal model. The order arrangement between behaviors which yields the highest C.R. value will indicate which of the behaviors being con- sidered is of greatest consequence or importance. Figure 3, above, shows the perfect pattern of response for three behaviors, with a total score range of zero to three, which would yield a C.R. = l. 00. Examination of the corresponding scale types for the various scores indicates that only one response pattern will yield a Scale Type 4, and only one response pattern will yield a Scale Type 3, etc. In the perfect pattern of response there are no errors. In practice, the perfect correspondence to the formal model is never obtained; i.e., there are errors in the response patterns. Considering again the hypothetical order arrangement of three behaviors, (A, B, and C), the formal model most closely approximating the perfect formal response pattern is presented in Figure 4. 135 Figure 4. --Half-Parallelogram Pattern for Guttman Model with Various Item Responses Patterns and Error Scores for Total Item Scores of 0 - 3 and Scale Types of l - 4 Col. 1 Col. 2 Col. 3 Col. 4 Col. 5 Col. 6 Total Behavior Behavior Behavior Scale Score A B C Type Errors 3 X X X 4 0 2 X X 0 4 1 2 X 0 X 4 l 2 O X X 3 0 l O X 0 3 l 1 0 0 X 2 0 1 X 0 O l 1 0 O O 0 l O The reader will note that both requirements for scalogram analysis are met in the order and response patterns represented in Figure 4; i.e., respondents are ranked from high to low in terms of their total score (column 1) and the three behaviors are ranked from left to right in terms of their difficulty (or consequence). Note that with three behaviors and a total score ranging from zero to three there are eight possible response patterns for this order arrangement of Behaviors A, B, and C. There is only one response pattern that will yield a total score of three. There is only one response pattern that will yield a total score of zero. There are three response patterns that will yield a total score of two, and three response patterns that will yield a total score of one. Given the analytical scheme discussed above for determin- ing the relative consequence of Behaviors A, B, and C, there are six possible order arrangements which would have to be examined. The formal pattern, presented above in Figure 4, would not change for any of these arrangements. All that would change would be the headings for Columns 2, 3, and 4 above. The six possible order arrangements 136 would change those headings in the following manner: ABC (as in Figure 4); BAC; ACB; CAB; BCA; CBA. Using the formal pattern and hypothetical order combination of Behaviors A, B, and C in Figure 4, the C.R. can be determined in the following manner. All that is needed is to add a seventh and an eighth column to Figure 4. Column 7 will be the frequency of responses to each of the eight response patterns. Column 8 will be the product of the frequency of responses to the response patterns times the number of errors for that response pattern. Summing down Column 8 will give the total number of errors for the particular order combina- tion. C.R. is then computed in the conventional manner: divide the total number of errors by the total number of items (in this case, three) times the number of respondents and subtract this quotient from 1.00. In the same manner the C.R. could be determined for all of the hypothetical order combinations of ABC, BAC, ACB, CAB, BCA, and CBA. The formal pattern will remain the same for each order combination. The person's total item score will remain the same. The scale types, the number of errors based on the response patterns in the formal model, will also remain the same for any order combin- ation. However, the respondent's scale type can change from one order combination to another, unless the respondent answered all items "correctly" or all items "incorrectly. " Thus, the frequency distribution for scale types would change for the six hypothetical order combinations considered above. The change of this frequency distribution of scale types will be demonstrated with the scalogram 5 analysis for the RCEB components in Chapter IV. 5 9The order combinations for RCEB components will be discussed and the C.R. values will be presented for each order com- bination in Chapter IV. The procedure by which the present investi- gator conducted the scalogram analysis is discussed in Appendix B to this thesis. 137 What is important at this juncture is the fact that the Guttman Scalogram Analysis of the various item order combinations can provide the investigator with only one, but certainly the crucial, piece of infor- mation regarding the pattern of relationship between the behaviors under consideration. The analysis can determine which order arrangement produces the highest C.R. and thus which behavior is of most conse- quence. It is legitimage to then examine the relationship between Scale Types for that order combination and the dependent variable of Self Identification. It would not be appropriate, on the other hand, to com- pare the relationship between Scale. Types for each of the remaining order combinations and the dependent variable of Self Identification for the reasons spelled out above; viz., the various order combinations would not yield the same frequency distribution for each scale type nor would the same persons have the same scale type on each order com- bination . Summary In this section I have discussed the procedures by which the analysis of data for the present investigation will be carried out. The procedures are three: (1) I will examine the association between the RCEB components, separately and in various combinations, and the dependent variable of Self Identification. The presence or absence of association will be determined by the use of chi square analysis. The extent of that association will be determined by the use of Cramer's V statistic. (2) An examination will be made of the per cent of cases accounted for, in line with the major hypothesis and each of the sub- hypotheses, in predicted cells of MAXRCEB and MINRCEB. Consistent With those hypotheses, the per cent of cases in MAXRCEB and MINRCEB should approach 100. The SI % Index will be computed to indicate the over -all per cent of cases accounted for in both cells. (3) An examina- tion of the relative consequence or importance of the RCEB component var iables will be made by means of Guttman's Formal Model for Scale Ana~lysis . 138 Operational Restatement of Hypotheses Five hypotheses have been derived from the theoretical orien- tation within which the present investigation was conducted. In the pre- ceding sections of this chapter I have gone through the procedures of specifying the major concepts in the hypotheses, defining those concepts, specifying referents for those concepts, and prescribing operational procedures which are logically consistent with and necessary in order to make observations of the referents for those concepts. I will present the five hypotheses again and then restate those hypotheses in operational form. Major Hypothe sis: The greater the eigtent of the person's implication in a religipus context of experience and behavior the greater the likelihood that person will make religious self identification statements. Sub -Hypothe se 5: 1. The greater the extent of RHOMOG the more likely the per- son will make religious self identification statements. 2. The person who defines his MIG as a religious group will make religious self identification statements. 3. The greater the extent of RBEL the more likelj' the person will make religious self identification statements. 4. The greater the extent of RPART the more likely the person will make reliflaus self identification statements. I will restate the hypotheses in operational form, beginning With the four sub -hypotheses in order and then the major hypothesis. Operational Statement of Sub-Hypotheses: l. H : There is no difference between the two groups of per- 0 . . . . sons who do and who do not make religious self identi- fication statements in response to the SIP with respect to the magnitude of their scores on the RHOMOG index. 139 H : The greater the magnitude of persons' scores on the RHOMOG index the more likely those persons will make religious self identification statements in re- sponse to the SIP. 60 2. H : There is no difference between the two groups of persons who do and who do not make religious self identification statements in response to the SIP with respect to theirdefi nition of MIG as religious. H : Those persons who define their MIG as religious are more likely to make religious self identification state- ments in response to the SIP than are those persons who do not define their MIG as religious. 601n the case of every null hypothesis stated, the test for independence is between two populations from which the samples of persons are drawn who do and who do not make statements of religious self identification. Chi square does not test for independence between the two sample groups but rather for independence between the two popu- lations from which the sample groups are considered independent and random selections. In the case of the present study, as discussed in the section of Sample and Universe Relationships, I am considering the observations employed for the study as independent and random samples from the hypothetical populations of all persons who do and who do not make statements of religious Self Identification in relation to their im- plication in religious contexts of experience and behavior. The level of significance to be used in all tests of the null hypothesis will be . 05. The region of rejection for the null hypothesis will in every case consist of x2 values large enough that the probability of their occurrence is equal to or less than the stated level of significance, .05. The alternative hypothesis predicts the direction of association between the two groups of persons who do and who do not identify them- selves as religious. Therefore, the region of rejection will be one -tailed and the use of chi square will be refered to as a one -tailed test. In the case of the 2 x 4 contingency table for the test of independence between RHOMOG and Self Identification there are 3 degrees of freedom. The table of Critical Values of Chi Square indicates that for a one -tailed chi square test, with 3 df, a value of x2 = 6.25 or larger has the probability 0f occurence under Ho of p = 1/2 (.10) = .05. Thus the region of re- JECtion, for 3 degrees of freedom, consists of all x2 values of 6.25 or greater if the direction of the results is that predicted by H1 contingent upon the rejection of Ho’ The region of rejection for each one -tailed chi square test Will be specified given the degrees of freedom appropriate to the con- tingency table under consideration. In the case of each rejected null hypothesis, and the conse- quent acceptance of the alternative hypothesis regarding the direction 140 3. H : There is no difference between the two groups of persons who do and who do not make religious self identification statements in response to the SIP with respect to the magnitide of their scores on the RBEL index. H : The greater the magnitide of persons’ scores on the RBEL index the more likely those persons will make religious self identification statements in response to the SIP. 4. H : There is no difference between the two groups of persons who do and who do not make religious self identification statements in response to the SIP with respect to the magnitude of their scores on the RPART index. H : The greater the magnitude of persons' scores on the RPART index the more likely those persons will make religious self identification statements in response to the SIP. ' Operational Statement of Major Hypothesis: H0: There is no difference between the two groups of per- sons who do and who do not make religious self identi- fication statements in response to the SIP with respect to the magnitude of their ordered placement on the RCEB indices separately or in combination. H : The greater the magnitude of persons' ordered place— ment on the RCEB indices separately or in combina- tion, the more likely those persons will make religious self identification statements in response to the SIP. of association, Cramer's V will be computed to determine the extent of association. The results of the chi square test determine whether or not the association is present and the extent to which the investigator can be confident of association. That confidence is expressed by the p. value for the obtained chi square value. Cramer's V, then, provides the index of extent of the association given that level of confidence. For the basis of my discussion of the region of rejection for one -tailed chi square tests see Sidney Siegel, Nonparametric Statistics for the Behavioral Sciences, New York: McGraw-Hill Book Co.. Inc., 1956, pp. 13-14. 141 Summary of Chapter III In the present chapter I have stated and discussed the major hypothesis and sub -hypotheses and have defined the major con- cepts in those hypotheses. I discussed the issue of operationalism and indicated the necessity of specifying the operational procedures for making observations of the referents for each of the concepts versus the more presumptious procedure of operationally defining concepts. Following a preliminary description of the operational procedures employed in this investigation I then restated the major hypothesis and sub -hypotheses in order to take the assertion of di- rection of relationship into account. I next discussed the issues of validity and reliability of procedures of inquiry. I presented the various procedures which may be used for determinining the validity of observational techniques; criterion validity, construct validity, and quasi -validity I discussed the appropriate and inappropriate application of each and the consequences of such application. I then demonstrated the legiti— mate and appropriate use of construct validity to determine the validity of observational procedures in the present study. I pointed out the various aspects of reliability in the research process and indicated the shortcomings of focusing on repeatibility alone as the major cri— terion. I discussed the importance of the intersubjectivity of the ini- tial observation, of the manipulation of those initial observation, and, the problems involved in establishing the reliability of observations across time. In my discussion of the reliability of manipulation pro- cedures I discussed the treatemnt of responses to both open- and closed-ended questions. The former involved a discussion of the intersubjectivity which can be obtained by means of content analysis. The latter involved a discussion of Guttman Scale Analysis as a reli- able means of determining the relationship between any set of obser- vations of interest to an investigator. In my discussion of the problems 142 of establishing the reliability of observations across time I suggested that the result of reliable observations across time was more a func- tion of the relationship between the subject and others than of some function intrinsic to the observation technique. By the relationship between the subject and others, I referred to his relationship to those persons in various contexts of activities of interest to the investigator as well as his relationship to the observer who querries him about his activities. I then discussed the operational procedures for observing referents for the concepts in the hypotheses investigated in the present study. I followed the procedure for each index of defining the concept. specifying the class of activity to which it refers, and then describing the operational procedures utilized to make observations of the referent activity or class of interaction. This discussion was developed for the RHOMOG, MIG, RBEL, and RPART indices. I followed the same pro- cedure in discussing the observation technique for the dependent variable of Self Identification: SIP. I discussed the procedures as they were followed in the present study and the assumptions which were made in the use of SIP. I also discussed the manner in which the present usage deviated from conventional procedures and offered some speculation on the effects of those modifications. The sample on which the present study was conducted was discussed in terms of the issue of sample ~universe relationships. I presented my argument against the use of the sample to generalize to an irrelevant population of U. S. adults living outside of institutions. I presented my case for claiming such irrelevance on the basis of comparing the criteria by which such samples are drawn with the criteria of importance to the theory within which the present study was couched. I suggested instead that the sample could be legitimately considered as a random sample from the hypothetical universe of those persons who do and who do not make religious self identification statements. Support for this argument was drawn from the ontological 143 assumptions of G. H. Mead concerning the relationship between the scientist, the observations of his immediate experience, and the universality of relationships required by his theory. The analysis procedures were presented and discussed. I indicated that I would follow three procedural lines of analysis. First, I proposed an examination of the bivariate and multivariate relationship of RCEB components to the dependent variable of Self Identification. I specified the use of chi square to determine presence or absence of association and Cramer's V to determine the extent of association. Second, I discussed the proposed examination of the cases of Self Identification accounted for in MAXRCEB and MINRCEB in accordance with the hypotheses and the use of the SI % Index to reflect the over-all per cent of cases accounted for. Finally, I dis- cussed the use of Guttman's Formal Model for Scale Analysis to determine the relative importance or consequence of the RCEB com- ponents under consideration. The last section of the present chapter included a statement of the hypotheses in conceptual form and a restatement of those hypo- theses in operational form based on the earlier discussion of the in- dices constructed from the observations acquired through the speci- fied operational procedures. The bearing of the observations of those hypotheses will be presented in Chapter IV. CHAPTER IV TEST OF HYPOTHESES Introduction The purpose of this chapter will be to present the obser- vations obtained inthe present study as they bear on the hypotheses developed and discussed in the preceding chapters. I will restate the hypotheses in operational form and then present the data in relation to those hypotheses . Statement of Hypotheses in Operational Form Sub -Hypothe se 3 l. H : 0 H1: 2 H : 0 H1: There is no difference between the two groups of persons who do and who do not make religious Self Identification statements in response to the SIP with respect to the magnitude of their scores on the RHOMOG index. The greater the magnitude of persons' scores on the RHOMOG index the more likely those persons will make religious Self Identification statements in re- sponse to the SIP. There is no difference between the two groups of persons who do and who do not make religious Self Identification statements in response to the SIP with respect to their definition of MIG as religious. Those persons who define their MIG as religious are more likely to make religious Self Identification statements in response to the SIP than are those per- sons who do not define their MIG as religious. 144 3 H H 4 H H Major Hypothe sis H : O 145 There is no difference between the two groups of per- sons who do and who do not make religious Self Identi- fication statements in response to the SIP with respect to the magnitude of their scores on the RBEL index. The greater the magnitude of persons' scores on the RBEL index the more likely those persons will make religious Self Identification statements in response to the SIP. There is no difference between the two groups of per- sons who do and who do not make religious Self Identi— fication statements in response to the SIP with respect to the magnitude of their scores on the RPART index. The greater the magnitude of persons' scores on the RPART index the more likely those persons will make religious Self Identification statements in response to the SIP. There is no difference between the two groups of per- sons who do and who do not make religious Self Identi- fication statements in response to the SIP with respect to the magnitude of their ordered placement on the RCEB e indicies separately or in combination. The greater the magnitude of persons' ordered placement on the RCEB indicies, separately or in combination, the Test of Hypotheses on the Original Samile The initial tests of hypotheses for the bivariate relationships between RCEB components and the dependent variable of Self Identification was carried out on the original sample, 1:1 = l, 528. Deviations from that total N in the total number of cases reported in each contingency table are attributable to the lack of observations on respondents for one or the other of the variables being considered. In discussing the hypothesis under test for each contingency table, I will not repeat the operationalized hypotheses but will simply refer to them by number, as above, and will indicate whether the null or alternative hypothesis is under discussion. 146 Hypothe sis l The null hypothesis specified the absence of relationship between religious Self Identification and the extent of persons' religious homogeneity with close friends as indicated by their score on the RHOMOG index. The data bearing on this hypothesis are presented in Table 2. Table 2. --RHOIVIOG and Self Identification Self Identification RHOMOG RSI is: TOTAL 4 152 377 530 ( 28) (.72) (100 0) 3 82 182 264 2 108 346 454 1 32 73 105 (.30) (.70) (100.0) TOTAL 375 978 1, 353 x2 = 5,82, n.5, MAXRCEB = .28 MINRCEB = .70 v = .00 SI %Index = .34 The chi square test for independence yields a value of ' 1 x2 = 5.82, 3 df, which is not significant at the . 05 level and therefore does not permit the rejection of the null hypothesis of no association. Inspection reveals the percent of cases accounted for in MAXRCEB 1The Table of critical Values of Chi Square indicates that for a one -tailed chi square test with 3 degrees of freedom, a value of X = 6.25 or larger has the probability of occurrence under Ho of p = 1/2 (.10) = .05. Thus, the region of rejection consists of all x2 values 0f 6.25 or greater. 147 and MINRCEB and the total per cent of cases accounted for with the SI % Index are likewise inconsistent with the alternative hypothesis of association between RHOMOG and Self Identification. A discussion of this absence of relationship between RHOMOG and Self Identification will be withhled until the remaining bivariate relationships have been examined. Hypothe sis 2 The null hypothesis specified the absence of association between persons making religious Self Identification statements and their definitions of MIG as religious or non-religious. The data bearing on this hypothesis are presented in Table 3. Table 3. --Mig and Self Identification Self Identification MIG RSI ESI TOTAL R 207 334 541 (.38) (.62) (100 0) E 107 363 470 (.22) (.78) (100.0) TOTAL 314 697 1, 011 x2 = 27.11. p<.001 MAXRCEB = .38 MINRCEB = .78 V = .16 SI %Index = .56 The chi square test for independence yields a value of x‘2 = 27.11, p<.001, 1 df. for the association between MIG and Self Identification and thus permits the rejection of the null hypothesis of 148 no association. Computation of Cramer's V yields a value of V = .16, indicating a slight association between MIG and Self Identification in accordance with the specification of the alternative hypothesis. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, computation of the SI % Index indicates that 56 per cent of all cases were accounted for in the predicted cells consistent with the alternative hypothe sis . Table 4. -- RBEL and Self Identification Self Identification RBEL RSI E51 TOTAL 4 228 321 549 2 (.41) (.59) (100.0) x = 92.01, p.<.001 3 50 124 174 V = '25 MAXRCEB = .41 2 79 285 364 MINRCEB = .85 1 62 351 413 51 %Index = .60 (.15) (.85) (100.0) TOTAL 419 1, 081 1,500 The chi square test for independence yields a value of 2 x = 92.01, p< .001, 3 df, and thus permits the rejection of the null hypothesis of no association. Computation of Cramer's V yields a 2The Table of Critical Values of Chi Square indicates that f r a one -tai1ed chi square test with 1 degree of freedom, a value of x = 2. 71 or larger has the probability of occurence under Ho of p = 1/2 (.10) = . 05. Thus, the region of rejection consists of all x‘2 Values of 2.71 or greater. The region of rejection for a one -tailed chi square test With 3 degrees of freedom, . 05 level, consists of all x2 values of 6 .25 or greater. 149 value of V = .25, indicating the extent of association between RBEL and Self Identification in accordance with the alternative hypothesis . Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, computation of the SI % Index indicates that 60 per cent of all cases were accounted for in the predicted cells consistent with the specification of the hypothesis. Hypothe sis 4 The null hypothesis specified the absence of relationship between religious Self Identification and the extent of persons' parti- cipation in religious activities as indicated by their quartile score on the RPART index. The data bearing on this hypothesis are presented in Table 5. Table 5. --RPART and Self Identification Self Identification RPART RSI 851 TOTAL 4 158 128 286 2 (.55) (.45) (100.0) x = 186.32, p<.001 V = . 35 3 148 302 450 J, MAX‘SRiEB = . 55 2 82 362 444 MINRCEB = .92 l 26 273 299 SI % Index = .74 (.08) (.92) (100 0) TOTAL 414 1,065 1, 479 The chi square test for independence yields a value of 2 x = 186. 32, p<.001, 3 df, for the association between RPART and Self Identification, and thus permits the rejection of the null hypothesis 150 of no association.4 Computation of Cramer's V yields a value of V = . 35, indicating the extent of association between RPART and Self Identification in accordance with the specification of the alter- native hypothesis. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, computation of the SI % Index indicates that 74 per cent of the cases were accounted for in the predicted cells consistent with the alternative hypothesis. Table 6 presents a summary of the statistical analysis for the bivariate relationships between the RCEB components and Self Identification for the general sample. Table 6 also includes the V values for the relationships between each of the RCEB com- ponents. A discussion of these initial examinations of relationships will follow the presentation of the summary table. Examination of Summary Table 6 clearly reveals some basis for questioning the relationship of RCEB component RHOMOG to Self Identification and the the remaining RCEB components. First, the null hypothesis of no association between RHOMOG and Self Identification could not be rejected at the . 05 level. Second, and consistent with this lack of association, there was a low per cent of cases accounted for in the predicted MAXCREB and MINRCEB cells . This is also reflected in the total per cent of cases accounted for in all predicted cells by the SI % Index. Third, the reader will note that the V values for association between RHOMOG and the remaining RCEB components are considerably lower than among any of the other RCEB components. There is, however, a significant limitation to such com— parison. All of the associations examined above were based on unequal numbers of observations in each contingency table. Also the degrees The region of rejection for the one -tailed test with 3 degrees of freedom, .05 level, consists of all x values of 6. 25 or greater. 151 -- -- -- 3. 62.; a So.va $53 em z we flew: x 6633.2; ex. 5 Ems Em .e we .1; N mucmcomEoU mmOm on» MOM mcoEmMUOmmMHOuGH pad mgmamm Hangs—U on“ now comumoflficog “How was mucocomEoD mHUm Govern—on mnofimfioonwmaa Oumwudkfim mo ewumgadmuu .o 3an 152 of freedom are not comparable for every comparison, particularly in the case of associations between RCEB components. On this basis conclusive determinations cannot be made from this preliminary examination. Therefore, a subsequent examination was made of the bivariate association between RCEB components and Self Identification, and, among RCEB components on a comparable sample of observations. The selection of this sample was determined by whether or not com- parable observations were available for each respondent on all five variables under consideration; RHOMOG, MIG, RBEL, RPART, and Self Identification. The results of this analysis are presented in the next section. Test of Hypotheses on Reduced RCEB Sample Elimination of those persons for whom there were no ob- servations on all five variables reduced the sample size from 1, 528 to 915 cases. I will follow the same procedures for presenting the observations on these cases as in the previous section and will again simply refer to each hypothesis by number, indicating whether the null or alternative hypothesis is under discussion. Hypothe sis l The null hypothesis specified the absence of relationship between religious Self Identification and the extent of persons' reli— gious homogeneity with close friends as indicated by their scores on the RHOMOG index. The data bearing on this hypothesis are presented in Table 7. The chi square test for independence yields a value of 2 x = 5. 74, 3 df, which is not significant at the . 05 level and therefore 153 Table 7. - -RHOMOG and Self Identification Self Identification RHOMOG RSI 851 TOTAL 4 107 237 344 (.31) (.69) (100.1) 3 65 122 187 2 90 234 324 1 25 35 60 (.41) (.59) (100. ) TOTAL 287 628 915 x2 = 5.74, n.5, MAXRCEB = .31 MINRCEB = .59 V = .00 51 % Index = .35 does not "permit the rejection of the null hypothesis of no association. Cramer's V yields a value of V = .00. Inspection reveals the per cent of cases accounted for in MAXRCEB and MINRCEB and the total per cent of cases accounted for with the SI % Index are likewise in- consistent with the alternative hypothesis of association between RHOMOG and Self Identification. 6 A discussion of this absence of relationship between RHOMOG and Self Identification will be withheld until the remaining bivariate relationships have been examined. 5The region of rejection for the one -tailed test with 3 degrees of freedom, . 05 level, consists of all x2 values of 6. 25 or greater. 100 per cent of all cases should be accounted for in the MAXRCEB and MINRCEB cells. According to the alternative hypo- thesis the SI % Index should then approach 1. 00 if all cases are ac- counted for in all predicted cells. In the case of RHOMOG and Self Identification only 35 per cent of the cases in all predicted cells are accounted for consistent with the specification of the hypothesis. 154 Hypothe sis 2 The null hypothesis specified the absence of association between persons making religious Self Identification statements and their definitions of MIG as religious or non-religious. The data bearing on this hypothesis are presented in Table 8. Table 8. --MIG and Self Identification Self Identification MIG RSI RSI TOTAL R 187 300 487 (.38) (.62) (100.0) E 100 328 428 (.23) (.77) (100.0) TOTAL 287 628 915 x2 = 23.32, p<.001 MAXRCEB = .38 MINRCEB = .78 v=.16 51 %Index=.57 The chi square test for independence yields a value of x2 = 23. 32, p< . 001, 1 df, for the association between MIG and Self Identification and thus permits the rejection of the null hypothesis of no association. 7 Computation of Cramer's V yields a value of V = . 16. indicating a slight association between MIG and Self Identification in accordance with the specification of the alternative hypothesis. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, computation of the SI % Index indicates that 57 per cent of all cases were accounted for in predicted cells consistent with the specification of the hypothesis . 7The region of rejection for the one -tailed test with 1 degree of freedom, .05 level, consists of all x2 values of 2 . 71 or greater. 155 Hypothe sis 3 The null hypothesis specified the absence of relationship between religious Self Identification and the extent of persons' state- ments of RBEL as indicated by their quartile score on the RBEL index. The data bearing on this hypothesis are presented in Table 9. Table 9. --RBEL and Self Identification Self Identification RBEL R51 851 TOTAL 4 165 203 368 (.44) (.56) (100.0) 3 35 81 116 2 43 165 208 1 44 179 223 (.19) (.81) (100.0) TOTAL 287 628 915 2 x = 28.06, p< .001 MAXRCEB = .44 NHNRCEB =.81 V = .18 SI % Index = .58 The chi square test for independence yields a value of x2 = 28.06, p< .001, 3 (if, and thus permits the rejection of the null hypothesis of no association between RBEL and Self Identification. Computation of Cramer's V yields a value of V = . 18, indicating the extent of association between RBEL and Self Identification in accord- ance with the alternative hypothesis. Inspection reveals the per cent The region of rejection for the one -tailed test with 3 degrees of freedom, . 05 level, consists of all x2 values of 6.25 or greater. 156 of cases accounted for in MAXRCEB and MINRCEB and the overall per cent of cases accounted for by the SI % Index are also consistent with the alternative hypothesis. Fifty-eight per cent of the cases in the predicted cells are accounted for in terms of the specification of the hypothesis. Hypothe sis 4 The null hypothesis specified the absence of relationship between religious Self Identification and the extent of persons' partici- pation in religious activities as indicated by their quartile score on the RPART index. The data bearing on this hypothesis are presented in TABLE 10. Table 10. -—RPART and Self Identification Self Identification RPART RSI £51 TOTAL 4 128 94 222 (.57) (.43) (100.0) 3 91 202 293 2 53 195 248 1 15 137 152 (.09) g (.91) (100.0) TOTAL 287 628 915 2 x = 115.02, p< .001 MAXRCEB = .57 MINRCEB = .91 V==.35 51 %Index=.71 The chi square test for independence yields a value of 2 X = 115. 02, p< . 001, 3 df, for the association between RPART and Self Identification and thus permits the rejection of the null hypothesis 157 of no association. 9 Computation of Cramer's V yields a value of V = . 35, indicating the extent of association between RPART and Self Identification, consistent with the specification of the alternative hypothesis. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, compu- tation of the SI % Index indicates that 71 per cent of all cases were accounted for in the predicted cells consistent with the alternative hypo- thesis . Of the four null hypotheses tested, three were rejected, permitting acceptance of the alternative hypotheses for association between MIG and Self Identification, RBEL and Self Identification, and, RPART and Self Identification. Computation of V values indicated that the extent of association is .16, .18, and . 35, respectively, for those three bivariate relationships. The total per cent of cases accounted for in predicted cells, as indicated by the SI % Index, reflected a similar increase for the same three bivariate relationships. The null hypothesis of no association between RHOMOG and Self Identification was not rejected and the SI % Index for RHOMOG and Self Identification accounted for the smallest per cent of cases among the four relation- ships examined. Table 11 presents a summary of the statistical analysis for bivariate relationships between the RCEB components and Self Identification. The table also includes the association values for the relationships between each of the RCEB components. A dis- cussion of these relationships will follow presentation of Table 11. A number of conclusions can be drawn based upon exami- nation of the summary information presented in Table 11. The first and most obvious conclusion is that RHOMOG is not related to Self Identification. The null hypothesis of no association could not be rejected and the V value indicates no relationship between these two The region of rejection for the one -tailed test with 3 degrees of freedom, . 05 level, consists of all x2 values of 6. 25 or greater. 158 -- E. -- mm. m; a :5 .v a mm .93 affix a gamma -- 2. -- em. 2a m 89v a an .02 0:: sea/Em -- -- -- mm. m; m Sova 8.: oazuqmmm I -- -- E. m; a 89v a gem 52mm a 0020mm -- -- -- S. m; a 89v a em .am .553 a 0020mm -- -- -- 3. mg m 89v a mm .2 one a ooEomm Z. :3. 3. mm. m; m 25¢ a No.9: Hma Siam mm 3. it 2. m; m Sov a 8.3 E» nmmm S: 2. rm. f. m; a Sova 3.3 Ea 0:2 mm. am. 3. co. m; m .63: is Hm“ 0020mm 68?: mmong mmomxfiz > z :0 seems... x $33.23, ea 8 Ems Em as e6 .3 N mucmcomEoO mHOm 9.3 Mom mcofimHUOmmduoucH paw .Cofimodficog :om paw musmcomEoO mHOm £09,303 mGOSMUOmm/w oumwnmzm mo >umeadmuu JH 3an 159 variables. Furthermore, of all the contingency tables for which the SI % Index was determined, the RHOMOG and Self Identification rela- tionship produced the smallest per cent of total cases accounted for in the predicted cells of MAXRCEB and MINRCEB. All of the V values for the association between RHOMOG and the remaining RCEB com- ponents are not comparable. However, two of the three values in question can be compared; the value of V = .10 for the association between RHOMOG and RBEL, and, the value of V = .14 for the asso- ciation between RHOMOG and RPART. These may be compared with the obtained V value for the association between RBEL and RPART which is V = .28. 11 The reader will note that the highest of the association values involving RHOMOG, i.e., V = .14 for RHOMOG and RPART, is only half as great as the V value for the association between RBEL and RPART which is V = .28. The association between RHOMOG and RBEL of V = .10 is less than half as great as the extent of association between RBEL and RPART. The interassociation values for the RCEB components is an additional empirical basis for dropping 12 RHOMOG from consideration as a component variable of RCEB. Thus, the alternative hypothesis of association between RHOMOG and Self Identification was not accepted based on an exami- nation of observations in the general sample as well as in the sample presently under discussion. 11 As I have pointed out in Chapter III, Footnote 55 , values of association obtained by means of Cramer's V may be com- pared when the Min (r-l, c-l) is comparable for contingency tables under examination. In the cases just discussed in the text 4 x 4 con- tingency tables were being compared. Thus, in each case the Min (r-l, c-l) was 3. The same holds for comparison of 2 x 4 contin- gency tables which will be discussed in the text following. In the latter instance Min (r-l, c-l) is 1. In this and in all subsequent discussion of the extent of association among RCEB components I wish to make it quite clear that a high extent of association among the RCEB components is inter- preted to mean that they are highly related aspects of a religious con- text of experience and behavior as operationalized. Consistent with 160 On the other hand, the reader will note that the association values among the remaining RCEB components are relatively high. Two of them may be compared; the value V = .28 for the association between RBEL and MIG, and, the value of V = . 36 for the association RPART and MIG. This indicates a relatively strong association among RPART, RBEL, and MIG. The comparative extent of these associations becomes clear when the values for the association between RBEL and MIG and RPART and MIG are compared with the value for the as socia— tion between RHOMOG and MIG. The V value for the latter as socia- tion is V = . 14 which is only half the magnitude of the V value for the association between RBEL and MIG and less than half the magnitude of the association between RPART and MIG. All of the above considerations lead to the conclusion that the first sub -hypothesis specifying a relationship between RHOMOG, as an RCEB component, and Self Identification must be rejected. Such rejection requires a modification in the theoretical orientation from which the hypothesis was derived and also requires a reorganization of the analytical procedures. A discussion of both these consequences will followmy continued discussion of Table 11. At the same time, the preceding examination of the as soci- ation among RCEB components indicates that MIG, RBEL, and RPART are interrelated. It is then appropriate to inquire about the conclusions which can be drawn about the relative strength of association between each of these RCEB component variables and the dependent variable of Self Identification. my earlier comments on attitudes and my explicit disagreement with those who speak of underlying and latent factors, I do not interpret the association among RCEB components to mean that all are "really manifestations of some underlying, hidden or latent continuum, etc. " Momentary inspection of the observations used to make up the com- ponent indicies will indicate that they are conceptually and theoretically different and are combined into the various indicies for that reason. To turn around and say that high associations among the indicies re- veals the presence of some underlying continuum or latent factor is, to me, patent nonsense. 161 Based on the bivariate associations between the remaining RCEB components and Self Identification, as summarized in Table 11, the first set of comparisons which I wish to make is based on the SI 0/0 Ind ex. Here we see that MIG, RBEL, and RPART account for 57, 58, and 71 per cent respectively, Of all cases in the predicted cells for Self Identification. The second set of comparisons is based on the V values for the extent of association between the remaining RCEB components and Self Identification. All of the V values involved, for the relation -— ships between RCEB components and Self Identification, can be com- pared:13 the value of V = .16 for the association between MIG and Self Identification; V = 18 for RBEL and Self Identification, and, V = . 35 for RPART and Self Identification. Examining these V values we see that RPART is more strongly associated with Self Identification than is RBEL and that RBEL is more strongly associated with the de- pendent variable than is MIG. Based on this comparison of the bi— variate relationships we may posit a priority of relationship between these remaining RCEB components and Self Identification based on the observations from this sample. However, the comparison is based on the examination of the bivariate relationships for 915 cases, the number of cases in the RCEB sample based on available observations for all respondents on the RHOMOG, MIG, RBEL, RPART, and Self Identification indices . Now that RHOMOG will be dropped from con- sideration it is necessary to re -examine the relationships discussed above on a sample of all persons for whom observations were avail— able on the MIG, RBEL, RPART, and Self Identification indices; i.e. , those persons who were dropped from the immediately pre- ceding analysis for lack of observations on the RHOMOG index alone must now be included in the total sample of cases on which the 13 . . . In every instance the contingency tables involved are either 2 x 2 or 2 x 4. Thus Min (r-l, c-l) is l in each case. 162 hypotheses are tested. This procedure was carried out and a subse- quent examination of bivariate and multivariate relationships between the remaining RCEB components and Self Identification was under- taken as specified earlier in Chapter III. The results of that exami- nation will be presented following a discussion of the consequences of the lack of association between RHOMOG and the dependent variable of Self Identification. The most important consequence is the neces- sary re -examination of the theoretical framework within which the asserted RHOMOG and Self Identification relationship was initially entertained and the specification of any modifications which are now required by this reconsideration. A Re -Examination of the Theory with Respect to RHOMOG The data have shown that RHOMOG is not related to Self Identification. Based on this lack of relationship RHOMOG must be dropped from consideration as a component referent of the RCEB construct. This deletion, however, must be accounted for theoreti- cally or the theory must be modified to account for this empirical finding. As an initial point of departure, I would like to reconstruct the theoretical justification for originally including RHOMOG as a referent for RCEB. RHOMOG was originally considered as a referent for RCEB as a logical extension of Mead's concept context of experience and behavior. I discussed RHOMOG in Chapter II as a context of interaction within which others are in a position to direct identifying behaviors toward the person. That RHOMOG is a context of inter- action within which identifying behaviors are directed toward the person remains a tenable position. That RHOMOG is a context of interaction within which religious identification behaviors are directed toward the person must not be questioned. 163 The arguments initially presented in Cahpter II held that if the person shares in common with others some object or activity, identifying behaviors could be directed by those others toward the person in terms of that object or activity. If persons share with close friends the common object of religious preference, identifying behaviors could take place in terms of that common object. It is possible, however, if not highly probable, that the person shares many Objects and activities in common with close friends . It may well turn out that identifying behaviors can take place in terms of all or most of these objects and activities. How- ever, it can be argued that the object of common religious preference might be taken for granted when all members of the group of close friends share that object and therefore no identifying behaviors are taken in terms of that object. 14 Following this line of reasoning to one logical extreme, it could be argued that religious identification behaviors would take place and thus be of consequence in terms of the person making religious Self Identification statements, only in those cases where the person and his close friends had maximum religious preference heterogeneity. By this line of reasoning, the person's idiosyncratic religious preference would stand out as an object for identification behaviors as would the idiosyncratic reli- gious preference of each other person in the heterogenous group. Thus one would expect that more persons in such an instance would make religious Self Identification statements than not. Examination of Table 7 yields no direct empirical support for this expectation. In MINRCEB, 41 per cent of the persons do make religious Self Identifi- cation statements and 54 per cent do not. On the other hand, if the 4 categories of RHOMOG under consideration are examined to determine which of the categories has the highest per cent of per- sons making religious Self Identification statements, some indirect 14‘This line of argumentation was suggested to me, directly or indirectly, by both Professors Frederick Waisanen and William Faunce when the initial proposal for the present study was under consideration. 164 confirmation of the above notion can be developed. Moving from the category of greatest RHOMOG to that of least RHOMOG, the row per- centages of persons making religious Self Identification statements are, respectively: Category 4 (N = 344), 31 per cent; Category 3 (N = 187), 35 per cent; Category 2 (N = 324), 28 per cent; and, Category 1 (N = 60), 41 per cent. An inverse relation does not exist between extent of heterogeneity and per cent of Self Identification statements made, but the most heterogenous group does make the highest per cent of religious Self Identification statements. Another tack has more obvious theoretical relevance to this consideration of RHOMOG as an RCEB component referent. I would like to compare RHOMOG as a category of interaction within which religious identification behaviors could take place, with MIG, an additional category of interaction within which religious identifi- cation behaviors could take place. The major distinction between the two contexts of inter- action is a crucial one in terms of the identification behaviors which could take place. In the case of RHOMOG, as I have already pointed out, a number of different objects and activities may be shared by the persons concerned in terms of which identification behaviors may take place. In the case of MIG the person who says his MIG is a religious group is specifying a group within which religious activities do take place, although certainly other kinds of activities may also take place. Therefore, in MIG, one can be reasonably certain that if the group is a religious group, by definition, religious activities do take place and that identification behaviors may take place in terms of those religious activities. In the observations upon which RHOMOG was constructed no information was obtained which would indicate that religious activities do take place even though the mem- bers share, to one extent or another, the object of similar religious preference. In the observations upon which placement in MIG was based no direct information was obtained which would indicate that 165 religious activities do take place. However, the person does define the group as a religious group. Based on this, the inference that religious activities do take place within a religious MIG context has a more sound basis than does the similar inference concerning the close friends group. It is instructive to note that when the extent of associa- tion between the four RCEB components and Self Identification is compared, the two contexts of interaction considered (RHOMOG and MIG) yield the lowest association with religious Self Identification. There is no association between RHOMOG and Self Identification and a slight association between MIG and Self Identification. The next highest value of association is between RBEL, a Class 2 religious activity, and Self Identification. The highest association is between RPART and Self Identification. RPART cuts across all three classes Of religious activities and provides an index of the extent to which the person engages in those activities, thereby placing himself in a position where others can direct identification behaviors toward him in terms of those activities. RBEL, referring only to Class 2 religious activities, does this to a lesser extent. MIG, as a category of interaction within which Class 1 religious activities may take place, does this to a slightly lesser extent. RHOMOG, as a category of interaction within which many kinds of activities, religious included, could take place, provides little or no Opportunity for others to direct religious identifying behaviors toward the person. The modifications required in the theory are consistent with the main emphases of the general Meadian orientation. It is more appropriate to say that the modifications are required in the logical (or in retrospect "illogical") extensions of that theory as those extensions were spelled out in Chapter II. Based on my exami- nation of the relationship between RHOMOG and Self Identification, between RHOMOG and the other RCEB components and my discussion of those relationships, I draw the following conclusions. The most 166 important aspect of any context of experience and behavior to take into account is the constituent activities in which the person is en- gaged. Participation in such activities places the person in a position where others can direct identification behaviors toward him in terms of those activities. Of least consequence are those contexts of interaction within which related activities might occur, unless it is in fact the case that those activities do occur within the particular category of interaction. This is the major distinction between the two categories of interaction considered in this study: RHOMOG and MIG. Where the person defines his most important group as religious, there is some basis for the interference in re- ligious activities take place among members of that group and that identifying behaviors may be directed toward the person in terms of those activities. That this should have some consequence for the manner in which the person identifies himself, is borne out by the data obtained in the present study. In conclusion, and this is a point which will be mentioned again in Chapter V, it is probably of greatest importance that the investigator pay attention not only to the activities in which the person is engaged that make up a particular context Of experience and be - havior, but that he carefully examine the behaviors which others direct toward the person, in terms of those activities, in relation to the identification behaviors which the person directs toward himself. That is to say, consistent with the primary proposition in Mead's notions about the self, the self is produced by the behaviors which others direct toward the person. Therefore, the theory directs our attention to the behaviors in which the person is engaged and in terms of which others direct identifying behaviors toward the person. Contexts of interaction, like mere group affiliation, are of little con- sequence for Self Identification unless we have additional knowledge concerning the kinds of activities which may take place within the context. It is those activities in which the person is engaged with 167 others that are crucial for an examination of the basic proposition con- cerning the production and maintenance of the social self. The crucial observations are those of the identification behaviors which others direct toward the person in terms of the activities in which he and they are engaged. The next most logical set of Observations is of the activities in which the person engages and/or reports from the person of the activities in which he engages, where others are in a position to direct identifying behaviors toward him in terms of those activities. Removal of the RHOMOG variable from consideration in the test of the major and sub -hypotheses requires a reconstruction of the RCEB sample so as to readmit to that sample for the test of hypo- theses those persons who were initially excluded by virtue of not being ”observable" on RHOMOG alone. I will refer to this latter sample as RCEB', or RCEB prime. I will now present the data bearing on the hypotheses tested on the RCEB' sample. Test of Hypotheses on RCEB' Sample Readmitting those cases previously excluded for lack of observational data on the RHOMOG index, the sample size is increased from 915 to 993 cases. Thus, the final test of hypotheses in the pre- sent study was carried out on 933 cases or on 65 per cent of the original sample size of l, 528. I will first examine the bivariate re- lationships and then procedd to the multivariate relationships. Again, I will refer to the hypothesis by appropriate number and will indicate whether the null or alternative hypothesis is under discussion. Hypothe sis 2 The null hypothesis specified the absence of association between persons making religious Self Identification statements and their definitions of MIG as religious or non-religious. The data bearing on this hypothesis are presented in Table 12. 168 Table 12. -— MIG and Self Identification Self Identification MIG R51 R81 TOTAL R 204 329 533 (.38) (.62) (100.0) R 106 354 460 (.23) (.77) (100.0) TOTAL 310 683 933 2 x = 25.97, p < .001 MAXRCEB = .38 MINCREB = . 77 V=.15 SI %Index=.55 The chi square test for independence yields a value of x2 = 25. 97, p < .001, 1 df, for the association between MIG and Self Identification and thus permits the rejection of the null hypothesis of no association. 5 Computation of Cramer's V yields a value of V = . 16, indicating the extent of association between MIG and Self Identification in accordance with the specification of the alternative hypothesis. Examination of these data thus indicates the presence of slight association between MIG and Self Identification. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, computation of the SI % Index indicates that 55 per cent of all cases were accounted for in the predicted cells consistent with the alter- native hypothe sis . 15 The region of rejection for the one -tailed test with 1 degree of freedom, .05 level, consists of all x2 values of 2. 71 or greater. 169 Hypothe sis 3 The null hypothesis specified the absence Of relationship between religious Self Identification and the extent of persons' state— ment of RBEL. The data bearing on this hypothesis are presented in Table 13. Table 13. --RBEL and Self Identification Self Identification RBEL R51 R51 TOTAL 4 178 217 395 (.45) (.55) (100.0) 3 38 83 121 2 48 175 233 1 46 208 254 (.18) (.82) (100.0) TOTAL 310 683 933 x2 = 65.78, p < .001 MAXRCEB = .45 hflNRCEB =.82 v = .25 51 % Index = .59 The chi square test for independence yields a value of Z x = 65.78, p < .001, 3 df, and thus permits the rejection of the null hypothesis of no association between RBEL and Self Identifi- cation. Computation of Cramer's V yields a value of V = .25, indicating the extent of association between RBEL and Self Identifi- cation in accordance with the alternative hypothesis. Based on the l 6The region of rejection for the on -tailed test with 3 degrees of freedom, .05 level, consists of all x values of 6.25 or greater. 170 per cent of cases accounted for in MAXRCEB and MINCREB, compu- tation of the SI % Index indicates that 59 per cent of all cases were accounted for in the predicted cells consistent with the alternative hypothe sis . Hypothe sis 4 The null hypothesis specified the absence of relationship between religious Self Identification and the extent of persons' parti- cipation in religious activities as indicated by their score on RPART. The data bearing on this hypothesis are presented in Table 14. Table 14. --RPART and Self Identification Self Identification RPART R51 ~ RSI TOTAL 4 138 101 239 ( 57) (.43) (100.0) 3 99 216 315 2 55 205 260 1 18 161 179 (.10) (.90) (100.0) TOTAL 310 683 993 x2: 181.19, p<.001 MAXRCEB: .57 NHNRCEB =.90 V = .42 SI % Index = .72 The chi square test for independence yields a value of x2 = 181.19, p < .001, 3 df, for the association between RPART and Self Identification and thus permits the rejection of the null hypothesis 171 of no association. 17 Computation of Cramer's V yields a value of V = .42, indicating the extent of association between RPART and Self identification. Based on the per cent of cases accounted for in MAXRCEB and MINCREB, computation of the SI % Index indicates that 72 per cent of all cases were accounted for in the predicted cells consistent with the alternative hypothesis. Table 15 presents a summary of the statistical analysis for the bivariate relationships between the RCEB components and Self Identification. The table also includes the association values for the relationships between each of the three RCEB components. A discussion of all these relationships will follow the presentation of Table 15. Three null hypotheses were tested and rejected permit- ting acceptance of the alternative hypotheses specifying relationships between MIG and Self Identification, RBEL and Self Identification, and, RPART and Self Identification. Computation of Cramer's V indicates the extent of association to be .16, .25, and .42, respec- tively, for these three bivariate relationships. The total per cent of all cases accounted for in the predicted cells, as indicated by the SI % Index was 57, 59, and 72, respectively. Thus, the three hypotheses under test specifying a positive and direct relationship between RCEB Components and the dependent variable of Self Identi- fication are confirmed. Examination of the V values for association among the RCEB components again provides support for the assertion that the three referent variables of MIG, RBEL, andRPART are related. A discussion of the relative extent of association between the RC E B component variables and Self Identification will be withheld until the multivariate relationships between RCEB and Self Identification have been examined. 17The region of rejection for the one -tailed test with 3 degrees of freedom, .05 level, consists of all x2 values of 6.25 or greater. 172 i -- -- -- om. mom a as. v a 8.4.2 emfimm a ammm -- -- -- mm. 25 m So. v a 2 .82 .22“ 554mm -- -- 1- V mm. mam M So. v a 3:; 92 u ammm Ne. ea. 3.. .3. 3a m :5. v a 2.x: E A .3265 am. mm. mm. mm. 23 m 8o. v a 3.3 Eu 15mm 3. E. 3. 3. mom M So. v a 3.3 Eu 0H2 08?: mmomzaz mmomxfiz > z E .rqu No. 663.629 .e E Em ea Ems we 3 onEmm .mmom mo 393.23% no woman .muGoGOQEOO mMOam MOM mcofimmooEmmuouGH .pcm .COSMOGEGOUH :om pad muGOGOQEoO QMOM coma/«om mGOSdUOmmaq. mummnnfim mo >Hm885m11 .mm 3an 173 All of the sub-sypotheses submitted to test on this sample of observations were confirmed. Thus, there is empirical confirmation of the relationship between the separate referents for the construct Of religious context of experience and behavior and the referent for the construct of religious Self Identification. The major hypothesis of the study, however, has not yet been submitted to direct empirical test. Examination of this hypothesis, in relation to the obtained obser - vations, requires a multivariate analysis of the relationship between the RCEB component variables, in various combinations, and the dependent variable of Self Identification. This will be achieved by means of the chi square statistic testing for the independence of those groups of persons who did and did not make religious Self Identification statements in response to the SIP with respect to their distribution on the various combinations of values for the polytomies of RBEL, MIG, and RPART. In each case the null hypothesis under test will be the same; i.e. , the hypothesis of independence or no association. In each case where the null hypothesis is rejected the alternative hypo- thesis will be the same; i.e. , the proportion of persons who will make religious Self Identification statements is increased as the combination of values for the polytomies of RBEL, and/or MIG, and/or RPART is increased. Again, I will not repeat the operationalized hypothesis but will simply refer to the major hypothesis by name and will indicate whether the null or alternative hypothesis is under discussion. Test of Major Hypothesis The null hypothesis specified no difference between the two groups of persons who do and who do not make religious Self Identi- fication statements with respect to their distribution of the values of the various combinations of the polytomies RBEL, MIG and RPART. Four sets of observations will be examined to test this hypothesis. 174‘ RBEL -togethe r -with -MIG, and, Self Identification The first set of Observations examined is the relationship under the null hypothesis, between RBEL-together-with-MIG, and Self Identification. The data for this relationship are presented in Table 16. Table 16. -- RBEL-together -with-MIG, and Self Identification Self Identification RBEL MIGW R51 R51 TOTAL R 132 141 273 (.48) (.52) (100.0) 4 _ . R 46 76 122 R 26 39 65 3 — R 12 44 56 R 22 88 110 2 — R 26 87 113 R 24 61 85 1 R 22 147 . 169 (.13) (.87) (100.0) TOTAL 310 683 993 x?- = 81.23 p < .001 MAXRCEB = .48 MINRCEB = .87 V = .28 SI % Index .61 175 The chi square test for independence yields a value of x2 2 81.23, p < .001, 7 df, for the association between RBEL- together-with-MIG and Self Identification and thus permits the re- jection of the null hypothesis of no association. l8 Computation of Cramer‘s V indicates that the extent of association between RBEL- together-with-MIG and Self Identification is V = .28. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, com- putation of the SI % Index indicates that 61 per cent of all cases were accounted for in the predicted cells consistent with the alternative hypothe sis . 18 The region of rejection for the one -tailed test with 7 degrees of freedom, .05 level, consists of all x2 values of 12.02 or greater. Recognizing that x2 has the property of additivity it is instructive to examine the cell xz value contribution to the overall xZ value in large contingency tables where direction is predicted. In the present case of the alternative hypothesis, and consistent with the same reasoning for examining the per cent of cases accounted for in MAXRCEB and MINRCEB, the corner cells in the multivariate contingency table should make the largest contributions to the total xz value for the table. This is true in the case of the table for RBEL- toget_her-with-MIG, and, Self Identification. The RSI-MAXRCEB cell and RSI-MINRCEB cell should each yield 100 per cent of this observed cases for their respective pols. Their companion cells, RSI-MAXCREB cell and RSI-MINRCEB cell, should thus yield none of the observed cases for their respective rows. For Table 16 this is Obviously not the case in terms of observed frequencies. However, dueto the fact that x2 is a measure of the observed-expected discrepancy in each cell we Should expect these four cells to make the greatest x‘2 value contri— butions to the overall xZ values: for the RSI-MAXRCEB cell the contri- bution is x‘2 = 25. 93; for the RSI-MINRCEB cell the contribution is x‘2 = 18.13; fer the RSI-MAXRCEB cell the contribution is x2 = 11.75; and, for the RSI-MINCREB cell the contribution is x2 = 8.28. These are the largest four cell x2 contributions among all cells in the table and com- bined they equal x‘2 = 64.14 which constitutes 79 per cent of the total chi Sguare value of x2 = 81.23. 176 RPART -together -with -MIG, and, Self Identification The second set of Observations examined bearing on the null hypothesis of no association between RCEB and Self Identification is for the relationship between RPART -together-with-MIG, and, Self Identifi- tion. The data for this relationship are presented in Table 17. Table 17. --RPART -together-with-MIG, and Self Identification Self Identification RPART MIG RSI R51 TOTAL R 111 76 187 ' (.59) (.41) (100.0) 4 .— R 27 25 52 R 58 128 186 3 — R 41 88 129 R 28 84 112 2 — R 27 121 148 R 7 41 48 1 R 11 120 131 (.08) (.92) (100.0) TOTAL 310 683 993 x2 = 132.70, p < .001 MAXRCEB = .59 MINRCEB = .92 v=,36 51 %Index=.73 The chi square test for independence yields a value of x?- s 132. 70, p < .001, 7 df, for the association between RPART- together-with-MIG, and, Self Identification and thus permits the 177 Table 18.--RBEL-together-with-RPART, and, Self Identification Self Identification RBEL RPART R51 R51 TOTAL 4 94 (.60) 62 (.40) 156 3 61 90 151 4 2 22 56 78 1 1 9 10 4 17 10 27 3 13 36 49 3 2 5 26 31 1 3 11 14 4 18 19 37 3 10 61 71 2 2 15 62 77 1 5 33 38 4 9 10 19 3 15 29 44 1 2 13 61 74 l 9 (.17) 108 (93) 117 Total 310 683 993 x2 = 151.29, p < .001 MAXRCEB = .60 hnNRcEB =.93 V = . 39 SI % Index . 74 178 rejection of the null hypothesis of no association. 19 Computation of Carmer's V indicates that the extent of association between RPART- together-with-MIG, and, Self Identification is V = .36. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, compu- tation of the 51 %Index indicates that 73 per cent of all cases were accounted for in the predicted cells consistent with the alternative hypothe sis . RBEL -together -with-RPART , and, Self Identification The third set of observations examined bearing on the null hypothesis of no association between RCEB and Self Identification, is the relationship between RBEL-together-with-RPART, and, Self Identi- fication. The data for this relationship are presented in Table. The chi square test for independence yields a value of x?- = 151.29, p < .001, 15 df, for the association between RBEL- together-with-RPART, and, Self Identification and thus permits re - 20 jection of the null hypothesis of no association. Computation of 9The region of rejection for the one -tailed test with 7 degrees of freedom, .05 level, consists of all x2 values of 12.02 or greater. In the contingency table for the relationship between RPART-together-with-MIG, and, Self Identification, the corner cell x‘2 contributions are as follows: the RSI-MAXRCEB cell x2 = 48.43; the RSI-MINRCEB cell x2 = 21.95; the RSI-MAXRCEB cell x2 = 21.77; and, the RSI-MINRCEB cell x?- = 10.00. Combined their x2 valued for these four cells total x2 = .02.15 which is 77 per cent of the overall x2 value of 132.70. The region of rejection for the one -tailed test with 15 degrees of freedom, .05 level, consists of all xZ values of 22. 31 or greater. In the contingency table for the relationship between RPART- together-with-MIG, and Self Identification, the corner cell x2 contri- butions are as follows: the RSI;MAXRCEB cell x2 = 41. 32; the RSI- MINRCEB cell x2 = 21.18; the RSI -MAXRCEB x2 = 18.92; and, the RSI -MINRCEB cell x2 = 9.80. One additional cell makes a larger x2 contribution than the least of the preceding. This is the R51 cell 179 Cramer’s V indicates that the extent of association between RBEL- together-with-RPART, and, Self Identification is V = . 39. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, compu— tation of the SI % Index indicates that 74 per cent of all cases were accounted for in the predicted cells consistent with the alternative hypothe sis . RBEL -together -with -RPART - together-with-MIG and Self Identification The final set of Observations examined bearing on the null hypothesis of no association between RCEB and Self Identification was the relationship between RBEL-together-with-RPART -together-with- MIG, and, Self Identification. The data for this relationship are pre- sented in Table 19. The chi square test for independence yields a value of x2 = 172.40, p < .001, 31 df, for the association between RBEL-with RPART-with-MIG, and, Self Identification and thus permits rejection of the null hypothesis of no association. 21 Computation of Cramer's V for Q RBEL and Q RPART. The observed frequency for the cell = 17. The cell xZ contribution is x2 = 10.12. Excepting this cell, the total xZ contribution of the corner cells is x2 = 91.22 which is 60 per cent of the overall xZ value of 151.29. The exceptional cell referred to above con- tribution of the corner cells is x2 = 91.22 which is 60 per cent of the overall xZ value of 151.29. The exceptional cell referred to above con- tributes 6. 6 per cent of the overall xZ value. The region of rejection for the one -tailed test with 31 degrees of freedom, . 05 level, consists of all xZ values of approximately 45. 00 or greater. The approximation is based on an interpolation of the appropriate xZ value for 31 df based on the required chi square values for 30 degrees of freedom (x2 equal to or greater than 40.26), and, for 40 degrees of freedom (xZ equal to or greater than 51. 81) . In the contingency table for the relationship between RBEL - together-with-RPART -together-with-MIG, and, Self Identification, the corner cell x‘2 contributions are as follows: the RSI-MAXRCEB cell x2 = 43.10; the RSI-MINRCEB cell xZ =43.10; the RSI-MINRCEB cell x2 = 16.68; the RSI MINRCE B cell x2 = 9.4 . The total contribution huh. I 1|1 UhflUWVFHHO\Q mm mm .ummHUNE/SE mo .nmmomxg 180 L c m. a m N _N N m N m m N m N NN 2 o m m S. 3 a m m a N m a. «N a 2 m m m o m N m w a m mm mm o m I N 3. S 2 m a. S. N N m m «3 cc - mm m Nm 2 E m a. «Na 3%; 3 3c; ow m 4.3.09 Em Em 92 954mm ammm _ 5:85:63 how rsonaosdscopa :E can 622-55-emsaaméssfimmmi .3 “03mm. 181 Nos 23 SN 132.0,; «a so; 5 so; a m , N N N m be as s m NN NN c m N ON c m m: o o m N a N m a: a N. m «N NN N m E Z N m 2. Nm a m cm om c m X EN o m em N... v m E m a m MN 3 o m a e N m 182 indicates the extent of association between RBEL-with -RPART -with- MIG, and, Self Identification is V = .41. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, computations of the SI % Index indicates that 77 per sent of all cases were accounted for in the predicted cells consistent with the alternative hypothesis . Table 20 presents a summary of the statistical analysis for the multivariate relationships between RCEB components and Self Identification. Also included in the table are the summary statistics for the bivariate relationships of the RCEB components and Self Identification. A discussion of all these relationships will follow the presentation of Table 20. of the corner cells is x2 = 86. 99 which is 51 per cent of the overall xZ value of 172.40. One additional cell makes a larger xZ contribution than the least of the four corner cells. This is the RSI cell for Q4 RBEL, Q4 RPART, and R-MIG. The observed frequency for this cell is 15. The cell x‘2 contribution is x2 = 9.14 which is 5 per cent of the overall x2 value of 172.40. 183 2. mo. we. 2.. moo am So. v a 3N: E a .22 .Hmfimm .Ammm 3. mo. oc. om. moo 2 8o. v a oNi: Eu Ransom .qmmm ms. No. om. cm. moo o So. v a ENS E a .92 .emadm 3. mm. me. mN. moo N. 8o. v a mN.E E“ .92 .qmmm me. so. S. Ne. moo m :5. v a on .2: E a em z E .nsmE ex 33323, E Em so Em so No .23 soroosssopa row was muGOGOQEOO mmom cookfiofl GOSMMUOmmAw mummnmufifidz pan mummnnsfim mo umumEEdm <1- .ON £an 184 Discussion of Summary Table 20 I will proceed in my discussion of this table by first re- viewing the results of the examination of the bivariate relationships and then I will turn to the results of the multivariate analysis. Considered separately, a rank order can be established among the bivariate relationships with respect to their extent of association with the dependent variable of Self Identification. The extent of association between MIG and Self Identification is V = . l6, and the SI % Index indicates that 55 per cent of all cases were accounted for in the predicted cells. The extent of the association between RBEL and Self Identification is V = .25, and the SI % Index indicates that 59 per cent of all cases were accounted for in the predicted cells. The strongest relationship is between RPART and Self Identification. The extent of association is V = .42, and the SI % Index indicates that 72 per cent of all cases are accounted for in the predicted cells. Thus, as I indicated earlier, all three hypotheses are confirmed concerning the relationship of RCEB components and Self Identification. Obviously, these bivariate relationships have some bearing on the major hypo- thesis. In operationalized form that hypothesis stated that the greater the magnitude of persons' ordered placement on the RCEB indices separately or in combination, the more likely those persons will make religious Self Identification statements in response to the SIP. Thus, the various combinations of RCEB components were examined in re- lation to the dependent variable of Self Identification. Examination of the multivariate relationship between RCEB components and Self Identification, as summarized in Table 20, in- dicates that the hypothesized relationship is confirmed in every instance. As additional controls are added, i.e., as additional aspects of RCEB are taken into account, the relationship between RCEB and Self Identi- fication is increased. Thus, the extent of association between RBEL- together-with-MIG, and, Self Identification is V = . 28, and the SI ‘70 Index indicates that 61 per cent of all cases are accounted for in the 185 predicted cells. As would be expected, given the extent of bivariate association between RPART and Self Identification, the extent of association between RPART -together-with -MIG, and Self Identification is V = . 36, and the SI % Index indicates that 73 per cent of all cases are accounted for in the predicted cells. Consistent with the hypo- thesis of increased relationship between RCEB and Self Identification as RCEB is increased, the extent of association between RBEL- together-with-RPART and Self Identification is V = . 39, and the SI % Indix indicates that 74 per cent of all cases are accounted for in the predicted cells. Finally, all of the RCEB components are examined simultaneously in relation to Self Identification. The extent of association between RBEL-together-with-RPART -together-with MIG and Self Identification is V = .41, and the SI % Index indicates that 77 per cent of all cases are accounted for in predicted cells in accordance with the specification of the major hypothesis . Based on an examination of all statistical analyses pre- sented in Summary Table 20 for the bivariate and multivariate re- lationships, the three sub -hypotheses and the major hypothesis are confirmed. The V values for the bivariate relationships are low to moderate, ranging from .16 to .42. The per cent of cases accounted It is obvious from inspection of the summary statistics in Table 21 that the bivariate relationship between RPART and Self Identification of V = .42 is slightly higher than the multivariate relation- ship between RBEL-together-with -RPART -together-with-MIG, and, Self Identification of V = .42. Note, however, that the total per cent of all cases accounted for in the predicted cells is greater in the latter instance. There 77 per cent of all cases are accounted for in the pre- dicted cells as compared with 72 per cent in the bivariate relationship between RPART and Self Identification. Furthermore, the reader will note that the SI % Index for all multivariate considerations begin- ning with RPART -together-with-MIG, and, Self Identification indicates a higher per cent of cases accounted for than in any of the bivariate relationships examined. 186 for in the predicted cells, as indicated by the SI % Index, range from 55 per cent to 72 per cent for the bivariate relationships. The V values for the multivariate relationships are moderate, ranging from . 28 to .41. The per cent of cases accounted for in all pre- dicted cells, as indicated by the SI % Index, range from 61 per cent to 77 per cent for the multivariate relationships. Thus, while the extent of association ranges only from slight to moderate in all relationships examined the hypotheses are all confirmed and the relationships are in the predicted direction. Although the statistical analysis testing the major and sub -hypotheses has been presented and their confirmation has been determined, one facet of the proposed analytical procedure remains to be considered. While it is obvious from examining Summary Table 20 that RPART yields the highest relationship with Self Identi- fication, followed in descending order by RBEL, and MIG, the reader will observe that such a determination is made only on the basis of considering each of these RCEB components separately in relation to the dependent variable of Self Identification. In my discussion in Chapter 111, an analytical procedure was proposed to determine the one RCEB component, among all such components considered simultaneously, which is of greatest consequence in terms of persons making religious Self Identification statements. The proposed analytical procedure centered around the use of Guttman's formal model for scale analysis . I will now present the results of that analysis . 23 In the maximum multivariate contingency table 77 per cent of all cases were accounted for as is indicated by the SI % Index. The 23 per cent deviant cases will be discussed in a subsequent section of the present chapter. 187 Imposition of the Three RCEB Components On Guttman's Formal Model for Scale Analysis The RCEB components considered are RPART, RBEL, and MIG. The imposition of the various order combinations of these three component sets of behaviors on the Guttman formal model will allow determination of the one component which is of greatest impor- tance or consequence among the three RCEB components considered simultaneously. The format for this analysis was presented in Figure 4, Chapter 111. There the formal model was presented for three behaviors, each scored dichotomously with the values of l and zero, in a hypothetical order arrangement for all possible response patterns to the three sets of behaviors yeilding total scores ranging from zero to three. There are six different possible order arrangements for the three RCEB components examined in this analysis. They are: ‘ RPART RBEL MIG RPART MIG RBE L RBEL RPART MIG RBE L MIG RPART MIG RPART RBEL MIG RBE L RPART Observations of response patterns for each of these order arrange- ments were imposed on the formal model to determine that order arrangement yeilding the highest coefficient of reproducibility. 24 The order arrangement with the highest C.R. value will indicate the rank order of importance or consequence of the three RCEB com- ponents considered simultaneously. The results of that analysis are presented in Table 21 which gives the C.R. value, the rank of the C.R. 24 A discussion of the procedures used to submit these observations to scalogram analysis and the data for the response patterns for all order arrangements considered are presented in Ap- pendix B. 188 2 value, the I.O.C., value and the corrected rank 5 of item order combinations based on an examination of both C .R. and I. O.C . values . Table 21. --Rank Order of Importance for Six Order Combinations of RCEB Components Based on C.R. and I.O.C. Values RCEB Component Order Correc- C . R . Rank Combination M. M. R. L O. C. ted Rank . 93 1 RPART RBEL MIG . 63 . 81 l . 90 2. 5 RPART MIG RBEL . 51 .80 2 . 90 2 . 5 RBEL RPART MIG . 55 . 77 3 . 86 4 MIG RPART RBEL . 51 . 71 4 . 84 5 RBEL MIG RPART . 62 . 58 5 . 83 6 MIG RBEL RPART . 53 . 64 6 Based on the results of the analysis it is clear that when the three components are considered simultaneously, themost conse- quential component is RPART, followed in order by RBEL and MIG. The complete analytical picture for this order arrangement is pre- sented in Table 22. This table includes the empirical data for response patterns to the RPART -RBEL-MIG order combination, the error 25 The reader will recall by discussion of the relationship between C.R. values and I.O.C. values in Chapter III. When two scales have the same C.R. values it is instructive to examine the mini- mum coefficient of reproducibility which each scale could obtain given the distribution of responses on each item considered in the analysis. I.O.C. , or improvement over chance, is a ratio of the minimum re- producibility to the obtained reproducibility. In the data presented in Table 21, two order combinations had the same C.R. value of . 90: RPART -MIG-RBEL, and, RBEL-RPART-MIG. The former, however, had an I.O.C. = . 77. Therefore, the corrected rank order places the RPART -MIG —RBEL order combination 2nd and the RBEL-RPART - MIG order combination third and removes the tie for these two order combinations recorded in the second column of Table 21 . 189 frequency, and, the proportionate frequency distribution for each re- sponse pattern and scale type on the dependent variable of Self Identi- fication. Using the scale analysis procedure I have demonstrated the relative consequence of the RCEB components. RPART is of greatest consequence or importance among the three components considered simultaneously, followed by RBEL and MIG. The final task is to examine the relationship of the scale types for this order arrangement of RPART -RBEL—MIG with the dependent variable of Self Identification. As I have pointed out in Appendix B, the rescoring of the three RCEB components for the purpose of the scale analysis discards considerable information. For example, the quartile score differentiation on both RPART and RBEL is reduced to a simple dicho- tomy while the religious versus non -religious differentiation on MIG remains a dichotomy. In the response pattern presented in Table 22, the dichotomy for each of the three components is represented by an X, indicating the presence of the appropriate behavior, and by a 0, indicating the absence of the behavior. Even with the loss of infor- mation resulting from this rescoring procedure the relationship between RCEB, now scored in terms of Scale Types, and Self Identi- fication should remain the same, consistent with the major hypothesis. The extent of the association, however, given the loss of information by reducing the differentiation, should be less than in the multivariate relationship between RBEL -together -with -RPART -together -with -MIG and Self Identification. This reduction should also be reflected in the total per cent of all cases accounted for in the predicted cells as indicated by the SI % Index. The response pattern, error frequency and proportionate frequency distribution for each response pattern and scale type for the remaining five order combinations are listed at the end of Appendix B. 190 uncommon >cm 0» Boom omou m wcmcwmmmm Egon/m pcm oponEoo ouoE gmwomoimummifims oat. MO 925qu 05 momma ”:5 and? swam cm. .wfismon wow you .mmmsfimcm ommom 05. mo 055mm 9.3 owcmmo Nos mmop mmafiémofioflpomsmn 03m one GESHOO Ed: 5. momoom CC on» 62:1 .omsue: S. $039309»: mm. 3 was: gnomes. ou momoaucoumm G2: o9: mason Ema EEO commune were: H mN mo.-OO.m . . . osoN mo.nmooxm . . . . ... u m . u .1. . . . ll - u E mo.-mo. OO mc mmc+mmv+§m mEE mON OOH m o moo mmc Omm mON moo 2m :N Z. O 2m O a 33 O O O O ON s mm ON ON 2 m :3 O O m m mom cmm 3. O mom O N 33 x O O m Oo mm Nm Oo Oo m m 33 O x O m o: so Nm O o: O m 33 x x O N mc Nm mm mo NO 2 a. 33 x O m N Nm 2 E Nm Nm 2 a. :3 O x x N N2 3. Om O Na O e 33 x x x m mambo. Em Em opossum N esotm ease S A: N m e 6.80m 2 38m 0:2 doom .Omadm mason sotosBEoo nopeo om2-qmmm- Hmfimm one son assumes onsoanom: .NN oEom. 191 This final test of the major hypothesis was carried out by means of examining the relationship of RCEB scale types, for the order arrangement of RPART -RBEL-MIG to the presence or absence of religious Self Identification statements. The null hypothesis specified that there will no no relationship between those persons who do and who do not make religious Self Identification statements in response to the SIP with respect to the magnitude of their scores on the RCEB scale types. The data for this hypothesis are presented in Table 23. Self Identification RCEB Scale _ Types RSI RSI TOTAL 4 125 94 219 (.57) (.18) (100.0) 3 . 84 155 239 2 41 156 195 l 60 278 388 (.18) (.82) (100.0) 310 683 993 TOTAL (100.0) (100.0) x2=111.10, p<.001 MAXRCEB=.57 MINRCEB = .82 V = .33 SI % Index = .72 The chi square for independence yields a value of x2 = 111. 0, p < . 001, 3 df, for the association between the RCEB Scale Types for RPART -RBEL-MIG and Self Identification and thus permits 192 rejection of the null hypothesis Of no association. 28 Computation of Cramer's V yields a value of V = . 33, indicating a moderate association between the scale types and Self Identification. Based on the per cent of cases accounted for in MAXRCEB and MINRCEB, computation of the SI % Index indicates that 72 per cent of all cases are accounted for in the predicted cells consistent with the alternative hypothesis. In conclusion I have shown in this section that RPART, among all RCEB components considered simultaneously, is of greatest consequence or importance. This determination was made on the basis of establishing the rank order of C.R. values for the various order com- binations of the three RCEB components. RPART -RBEL-MIG is the order combination with the highest coefficient of reproducibility and therefore indicates the ordered relationship among those three RCEB components, ranging from left to right in terms of their rank order of importance. The relationship between scale type scores for the RPART- MIG-RBEL order combination and the dependent variable of Self Identifi- cation was examined. Computation of the extent of association indicated a moderate relationship of V = . 35. The SI % Index indicated that 72 per cent of all cases were accounted for in the predicted cells consis- tent with the specifications of the major hypothesis for the relation- ship between RCEB and Self Identification. Having established the rank order of consequence of RCEB components in relation to Self Identification, I will return to a discussion of the overall results of the multivariate relationship and a consideration of selected deviant cases in that analysis. The region of rejection for a one -tai1ed chi square test with 3 degrees of freedom, .05 level, consists of all x‘2 values of 6.25 or greater. 193 Examination of Deviant Cases in MAXRCEB and MINRCEB In Table 21 the relationship of RBEL-together-with -RPART- together-with-MIG, and Self Identification was considered. In this most comprehensive inspection of the relationship between RCEB and Self Identification, the largest per cent of cases in all predicted cells is accounted for; the SI % Index = 77 per cent. However, while 77 per cent of the cases are accounted for consistent with the specifications of the major hypothesis, this leaves 23 per cent which fall outside the bounds of that hypothesis as it is derived from the explanatory frrame - work of the theory. The reader's attention is directed again to the 44 cases in RSI-MAXRCEB and the 7 cases in RSI-MINCREB in Table 19. These are the crucial deviant cases which should be accounted for. While it is obvious that other cases do not fall in the appropriate cells in this multivariate contingency table, these particular forty-four cases are the major glaring exceptions to the major hypo- thesis. Attention to the theoretical framework within which this study was developed indicates one exploratory course for "explaining" these deviant cases. An assertion consistent with the theory would hold that it should be possible to account for the presence or absence of religious Self Identification statements on the part of these deviant cases by examining the differential extent of their implication in a religious context of experience and behavior. Based on the Guttman scale analysis of the RCEB com- ponents, I have shown that RPART is the component of greatest con- sequence, )hmong all components considered simultaneously, in relation to persons making religious Self Identification statements. Furthermore, reconsidering the bivariate relationships examined in this chapter, RPART yields the greatest extent of association with the dependent variable of Self Identification. It is clear at this juncture, that of the RCEB components considered, RPART is of greatest consequence in terms of persons identifying themselves as religious. Further 194 examination of the deviant cases with RPART as an index of the extent of persons' implication in a religious context of experience should be instructive with respect to the presence or absence of religious Self Identification statements. This will be possible only if more dif- ferentiated information is available on the RPART index than that which has already been examined. I will recall here a description of the RPART index. RPART is an index of the extent of persons' participation in Class 1, 2, andi3 religious activities. It is a five item ordinal scale with total score values ranging from zero to 15. For convenience of the bi- variate and multivariate analyses the quartile ranges for the frequency distribution of scores on the index were established and quartile values were assigned to each respondent. In doing this, considerable infor- mation was discarded to facilitate the parsimony of analysis. It is possible, however, to retrieve the differentiation in this information for the forty-five deviant cases. In so doing it is then possible to determine if such additional differentiation among the deviant cases, with respect to the extent of their implication in RCEB, will account for the presence or absence of religious Self Identification statements on their part. This is accomplished by simply examining the 44 cases in RSI-MAXRCEB along with their 80 counterparts in RSI-MAXRCEB in terms of their differentiation of the fourth quartile of RPART. That differentiation consists of the six values, previously collapsed into the fourth quartile of RPART, ranging from 10 to 15. In the same manner the 7 cases of RSI-MINRCEB can be examined along with their 87 counterparts in RSI-MINRCEB in terms of their differentiation on the first quartile of RPART. That differentiation consists of the five values, previously collapsed into the first quartile of RPART, rang- ing from 00 to 05. The distribution of the deviant cases, and their non-deviant counterparts, into the appropriate contingency tables is presented below. The first table, Table 24, is for the MINRCEB deviant cases 195 at the lower end of the continuum; the 7 RSI deviant cases and their 87 non-deviant counterparts. The second table, Table 25, is for the MAXRCEB deviant cases at the upper end of the continuum; the 44 RSI deviant cases and their 80 non-deviant counterparts. Table 24. --Deviant Cases in MINRCEB: Lower Five Values of RPART Index and Self Identification Self Identification RPART R51 R51 TOTAL Cum. % 0 0 0 O4 (.00) (.00) (.00) (100.0) 03 '2 30' 32 (.29) (.34) (.34) (100.0) L 02 1 18 19 (.14) (.21) (.20) (66.0) 01 3 28 31 (.43) (.32) (.33) (46.0) 00 1 11 12 (.14) (.13) (.13) (13.0) 7 87 94 TOTAL (100.0) (100.0) (100.0) Table 25. --Deviant Cases inMAXRCEB: Upper Six Values of RPART Index and Self Identification Self Identification RPART RSI ' R51 TOTAL Cum. % 15 7 3 10 (.09) (.07) (.08) (100.0) 14 14 3 - 17 (.17) (.07) (.14) (92.0) 13 7 5 12 (.09) (.11) (.10) (78.0) 12 '25 10 35 (.31) (.23) (.28) (68.0) 11 16 11 27 (.20) (.25) ( 22) (.40) 10 11 12 "23 .(.14). .(.27). (.18) 80 44 124 TOTAL (100.0) (100.0) (100.0) 196 If the tables are examined in terms of per cent distribution along the continuous variable of RPART, it is apparent that for Table 24 the deviant cases in MINRCEB, the largest proportion of both RSI and RSI cases falls at the lower end of the continuum; i.e., there is not much differentiation between those persons who do and who do not make religious Self Identification statements with respoect to the ex- tent of their implication in RCEB as indexed by RPART. Thus, at the lower end of the continum the additional information concerning the extent of persons' participation in religious activities provides no "explanation" for the presence or absence of religious Self Identi- fication statements on their part. The per cent distribution along the continuous variable of RPART in Table 25 is less clear. Some basis is needed to provide a point of separation along that continuum. For exploratory purposes I have simply divided the continuum at the median point for the cum- mulative proportion of cases along the RPART continuum. The median falls in the category for value 12 and necessitates either the inclusion of all cases in that row in one half of a 2 x 2 table, or, the equal allocation of those cases into both halves of the 2 x 2 table. I have chosen the latter alternative and have cast 12 of the 25 R51 cases into the upper half and 13 of the 25 RSI cases into the lower half. Accordingly, I cast 5 of the RSI cases into the upper half and 5 cases into the lower half. The consequent 2 x 2 table is presented below as Table 26. Inspection of Table 26 reveals that while the largest pro- portion of persons not making religious Self Identification statements (64 ‘70) are located below the median on RPART, there is an equal distribution of persons who do make religious Self Identification state- ments located above and below the median on RPART. The chi square test for the null hypothesis of no association between RPART and Self Identification yields a value of x2 = 1.89 which is not significant at the 197 Table 26. --Deviant Cases in MAXRCEB: Median of Upper Six Values of RPART Index and Self Identification Self Identification RPART RSI RSI TOTAL 40 16 H‘ h 1g ( .50) ( .36) 56 LOW 40 28 ( .50) ( .64) 68 80 AL TOT (100.0) (100.0) 124 xZ =1.89, n.s., .05 level . 05 level. 29 The examination of this additional information on RPART is not sufficient to distinguish between those persons who do and do not identify themselves as religious in MAXRCEB. Furthermore, no additional theoretically relevant information is available to me which would allow further differentiation, and thus permit the possible ex- planation, of these deviant cases. It is consistent with the theory to argue that additional observations of the religious activities or contexts of interaction in which these persons are engaged should provide sufficient differentia- tion to account for the deviant cases. The entire pattern of analysis considered up to this point would support such an assertion. We have seen that the per cent of cases accounted for in MAXRCEB and MINRCEB. as well as the extent of relationship between RCEB and Self Identification, has increased as additional observations were taken into account con— cerning the extent of persons' implication in a religious context of experience and behavior. Were additional observations available, it would perhaps be possible to show that this pattern of relationship The region of rejection for a one -tailed chi square test with 1 degree of freedom, . 05 level, consists of all x2 values of 2. 71 or greater. 198 between RCEB and Self Identification extends to include the deviant cases just considered. Those observations are not available and this places a limitation on the tentative line of explanation which I have set forth in my attempt, and failure, to account for the deviant cases in MAXRCEB and MINRCEB. This is a limitation of the study and will be noted in my summary and conclusions to Chapter IV. Summary and Conclusions The purpose of this chapter was to submit the major and related subhypotheses to empirical test. Those hypotheses, in operational form, are: Sub -ijothe se 5 l. H : There is no difference between the two groups of 0 persons who do and who do not make religious Self Identification statements in response to the SIP with respect to the magnitude of their scores on the RHOMOG index. H : The greater the magnitude of persons' scores on the RHOMOG index the more likely those persons will make religious Self Identification statements in response to the SIP. 2. H : There is no difference between the two groups of persons who do and who do not make religious Self Identification statements in response ,to the SIP with respect to their definitions of MIG as religious. H : Those persons who define their MIG as religious are more likely to make religious Self Identification statements in response to the SIP than are those per- sons who do not define their MIG as religious. 3. H : There is no difference between the two groups of persons who do and who do not make religious Self Identification statements in response to the SIP with respect to the magnitude of their score on the RBEL index. 199 H : The greater the magnitude of persons' scores on the RBEL index the more likely those persons will make religious Self Identification statements in response to the SIP. 4. H : There is no difference between the two groups of persons who do and who do not make religious Self Identification statements in response to the SIP with respect to the magnitude of their scores on the RPART index. H : The greater the magnitude of persons‘ scores on the RPART index the more likely those persons will make religious Self Identification statements in response to the SIP. ' Majpr Hypothe sis HO: There is no difference between the two groups of persons who do and who do not make religious Self Identification statements in response to the SIP with respect to the magnitude of their ordered placement on the RCEB indices separately or in combination. H : The greater the magnitude of persons' ordered placement on the RCEB indices separately or in combination, the more likely those persons will make religious Self Identification statements in response to the SIP. The initial test of sub -hypotheses was carried out on the original sample of 1, 528 observations. The hypothesized direction of relationship between MIG, RBEL, RPART and Self Identification was confirmed. The null hypothesis of no association between RHOMOG and Self Identification was not rejected. The sample of persons on which these tests of hypotheses were conducted was not comparable, however, due to the absence of‘observations on one or both of the variables considered in the test of each hypothesis. Therefore, the hypotheses were again submitted to test on a com- parable sample composed of those persons for whom observations were available on RHOMOG, MIG, RBEL,.RPART ans Self Identification. 200 The sub -hypotheses were again submitted to test on this reduced RCEB sample of 915 observations. Again, the hypothesized direction of relationship between MIG, RBEL, and RPART and Self Identification was confirmed. The extent of association between each of the three RCEB components and Self Identification increased in that respective order. This was so for the V values and for the SI % Index. The null hypothesis of no association between RHOMOG and Self Identification was not rejected. Failure to confirm the hypothe - sized relationship between RHOMOG and the dependent variable required a re -examination of the theoretical framework within which RHOMOG had been set forth as a logically consistent referent for the construct RCEB. RHOMOG was re -examined as a context of interaction within which others should direct religious identifying behaviors toward the person. Two lines of exploration were considered. The first tack suggested that if all persons within a given context of interaction share some object or activity in common, that object or activity may be taken for granted by those persons when all persons share the common object of religious preference, that object may be taken for granted. Conversely, in a group with heterogenous religious prefer- ence the person's idiosyncratic religious preference would stand out as a focal object for identifying behaviors directed toward him by other associates. The most heterogenous groups of religious pref- erents was examined to investigate this preferred explanation. Inspection revealed that of all categories of RHOMOG considered, the largest percentage of religious Self Identification statements was made by persons in this most heterogenous category of RHOMOG (41 per cent). However, more persons than not in this category failed to identify themselves as religious (59 per cent). Therefore, no empirical support was forthcoming for this line of interpretation. 201 The second line of exploration drew a distinction between RHOMOG and MIG as contexts of interaction within which religious identifying behaviors could be directed toward the persons. I sug- gested that the persons' close friends may provide a context of interaction where all involved persons share many objects and activities in terms of which identifying behaviors can take place. While this may also be true for the persons included in a most im- portant group, the definition of that group as religious specifies that religious activities do take place within that category of inter- action and that identifying behaviors can take place in terms of those religious activities. The data reveal that MIG, as a category of interaction within which religious identifying behaviors may be directed toward the person, is related to Self Identification. The extent of that rela— tionship, however, is slight; V = .16. And, this relationship accounts for only 57 per cent of the cases in the predicted cells on this set of observations. By comparison RHOMOG yields no rela- tionship with Self Identification and accounts for only 35 per cent of the cases in the predicted cells. On the other hand, the indicies of the various classes of religious activities in which persons are engaged, RBEL and RPART, yield comparatively higher associations with Self Identification. The extent of association between RBEL and Self Identification, on this sample of observations, is only slight, V = .18. The per cent of cases accounted for in predicted cells for those two sets of relationships is 58 and 71, respectively. The modifications which are required by this re -exami- nation of the theoretical relevance of RHOMOG deal not so much with the specific emphases of Mead's theoretical orientation as they do with the "logical" extensions of that theory set forth in Chapter II of this thesis. The theory requires that the investigator attend to the activities in which the person is implicated with others. It is those activities which constitute the person's context of experience 202 and behavior. It is the identifying behaviors which others direct toward the person in terms of those activities which produce and maintain Self Identification with respect to that context. If it is not possible to observe the activities of the others, the next most logical set of observations must deal with persons' activities and/or their reports of activities relevant to the context of experience and be- havior in question. Such observations, at least, provide an index of the extent to which those persons are in a position where others can direct identification behaviors toward them in terms of the activities in which they are involved with the others. On re -examination, categories of interaction, in com- parison with observations of persons' activities within those categories of interaction, are of minimal consequence in terms of placing the person in a position where others can direct identifying behaviors toward them relevant to the context of experience and behavior in question. A general category of interaction, such as the person's group of close friends, is seen to be of no theoretical relevance in terms of placing the person in a position where others can direct specific classes of identifying behaviors toward him. On the other hand, a specific category of interaction, such as one's most im- portant group, is of some theoretical relevance in terms of placing the person in a position where others can address specific identifying behaviors toward the person in terms of the specific context of experi- ence and behavior. Empirical support is provided for both of these considerations . The deletion of RHOMOG from consideration as a referent for RCEB required the reconstruction of the sample upon which the final test of hypotheses was conducted. The new sample consisted of those persons on whom observations were available for the remaining RCEB components of MIG, RBEL, RPART and Self Identification. 203 The hypothesized direction of relationship between MIG, RBEL, RPART, and Self Identification was confirmed. For these bivariate relationships RPART and Self Identification yields the highest extent of association, V = .42. The extent of association between RBEL and Self Identification, and, between MIG and Self Identification was .25 and .15, respectively. The same order of relationships between the RCEB components and Self Identification was observed in the total per cent of cases accounted for in the predicted cells. The SI % Index for the RPART relationship was 72 per cent; for the RBEL relationship it was 59 per cent; and, for the MIG relationship it was 55 per cent. Four sets of multivariate RCEB observations were examined in relation to the dependent variable of Self Identification. Each set of observations confirmed the specification of the major hypothesis: RBEL-together-with-MIG and Self Identification yields a relationship of V = . 28; RPART -together-with-Mig and Self Identi- fication yields a relationship of V = . 36; RBEL-together-with -RPART and Self Identification yields a relationship of V = . 39; and, RBEL- to gether-with-RPART -together-with-MIG and Self Identification yields a relationship of V = .41. Furthermore, and consistent with the specification of the major hypothesis, the proportion of persons making religious Self Identification statements in predicted cells increased as additional observations of the extent of implication in RCEB were taken into account. The SI % Index for the four sets of multivariate relationships, in the order presented above, was 61 per cent, 73 per cent, 74 per cent, and 77 per cent. Finally, the RCEB components of MIG, RBEL, and RPART were examined by means of Guttman's formal model for scale analysis to determine which of the components, considered simultaneously, was of greatest importance or consequence. The rank order of ar- rangement of RCEB components which yielded the highest coefficient of reproducibility, indicating the internal rank order of the three 204 components, was that of RPART-RBEL-MIG. The scale types for this order arrangement were examined in relation to the dependent variable of Self Identification. The extent of association and the per cent of cases accounted for were significant and in the direction specified by the major hypothesis, V = . 38 and 51 %Index = . 72. This lower extent of association was attributed to the loss of infor- mation resulting from the dichotomizing of RCEB indices scores for the purpose of the scalogram analysis. The multivariate relationship of RBEL-together-with- RPART -together-with-MIG and Self Identification yielded a moderate association of .41. In this multivariate relationship 77 per cent of all cases were accounted for in the predicted cells. This left 23 per cent of the cases unaccounted for in the crucial cells. These deviant cases were further examined in an attempt to account for the presence or absence of religious Self Identification statements on their part. Consistent with the theoretical orientation is the assertion that further differentiation in the observations on the deviant cases with respect to their implication in RCEB would be instructive in terms of account- ing for the presence or absence of religious Self Identification state- ments on their part. RPART, as the most important RCEB component, was used to pursue this analysis. Whereas only quartile values of RPART had been considered in the previous analyses, the raw scores of the RPART index were now used to examine the deviant cases, and their non-deviant counterparts in MAXRCEB and MINRCEB. Despite this additional examination, the deviant cases could not be differentiated from their non -deviant counterparts in MAXRCEB and MINRCEB with respect to the presence or absence of religious Self Identification statements. Failure to account for these cases, by means of the exploratory tack pursued, is inconsistent with the specifications of the theory as well as the overall pattern of data relationships in the present study. As increasing observations of the extent of persons' implication in a religious context of experience and behavior were 205 taken into account, an increasing proportion of persons did identify themselves as religious. It was suggested that additional observa- tions of the extent of persons' implication in RCEB would consistently follow this pattern and would extend to include the deviant cases not accounted for with the additional observations on the RPART index. Unfortunately additional observations beyond those just considered were not available to the investigator and thus no support could be developed for the explanation offered for the deviant cases. The lack of such additional observations is a shortcoming of the present study in that no theoretically meaningful explanation of the deviant cases could be systematically examined. In conclusion, this chapter has presented empirical evidence confirming the major hypothesis and three of the four re- lated sub -hypotheses. The major hypothesis asserts that as the extent of persons' implication in a religious context of experience and behavior is increased, those persons will be more likely to identify themselves as religious. This increasing implication in such a context is interpreted as placing the person in a position where others can address religious identifying behaviors toward him. The primary proposition in the theoretical orientation holds that the self is produced by the behaviors which others addressto the person in the context of experience and behavior in which he and they are involved. Thus increasing implication in such a context should result in persons identifying themselves in terms of the context. Four referent indices for the construct of religious con- text of experience and behavior were examined: RHOMOG--a category of interaction referring to the extent of religious preference homo- geneity among the person and his close friends; MIG--a category of interaction referring to the person's most important group and the definition of that group as religious or non-religious; RBEL--a class of religious activities indicating the extent of persons' subscription to statements of religious belief; and, RPART--indicating the extent 206 of persons' participation in three classes of religious activities. No relationship was observed between RHOMOG and Self Identifi- cation. A re -examination of the theoretical orientation within which RHOMOG had initially been considered discarded such a category of interaction as a relevant referent for the construct RCEB. Of the three remaining RCEB components, RPART was the most consequential referent in terms of placing persons in a position where others could address religious identifying behaviors toward them, resulting in those persons' identification of themselves as religious. While the extent of association observed in both the bivariate and multivariate relationships ranged from slight to moderate, the direction of association was in the predicted di- rection confirming the specification of the hypothesis. Concomit- antly, the per cent of persons making religious Self Identification statements in specified cells increased as additional observations of the extent of their implication in a religious context of experience and behavior was increased. CHAPTER V REVIEW, CONCLUSIONS AND PROPOSALS FOR FUTURE INVESTIGATION Introduction In this chapter I will review the initial question which gave rise to the inquiry presented in the preceding pages of this thesis. I will review the theoretical orientation within which a tentative answer was developed in response to that question and the relationship of the empirical observations to that tentative answer. I will discuss those findings and the consequences of those findings for the theoretical orientation. Finally, I will comment on the advisability and proposed direction of future inquiry with respect to the thesis problem. The Problem This study was conducted to obtain crucial information re- garding a fundamental social psychological question: "What are the basic processes which produce and maintain persons' conceptions and identifications of themselves ? " The inquiry was carried out within two specific guidelines: first, the inquiry was developed within the theoret- ical orientation of George H. Mead; second, a specific class of self identifications--religious self identification--was examined within the experiential . setting upon which it should, theoretically, depend--a religious context of experience and behavior. The Theoretical Orientation Mead suggests that the person comes to identify himself in a similar manner to the way in which others identify him. He sug- gests that self identification is the process of the person behaving 207 208 toward himself as others behave toward him in the contexts of exper- ience and behavior in which both he and they are involved. Previous investigators have attempted to account for religious self identification but they have not payed attention to persons' implication in the acti- vities with others which make up a religious context of experience and behavior. Attention to such things as religious preference, affiliation with religious minority groups, etc. , has not yielded a consistent ac- counting for persons' identification of themselves as religious. The present study, on the other hand, attempted to deal with the relation- ship between context Of experience and behavior and self identification. In examining Mead's position it is quite apparent that he did not construct a systematic set of interrelated theoretical statements. While there is a great deal of consistency in the assertions which he made about a broad spectrum of human activities, he did not develop a formal or systematic theory of human behavior. Examining his asser - tions concerning self-other relationships, the present investigator attempted to specify a limited set of propositions stemming from Mead’s general orientation. Two concepts are crucial to the set of propositions. Those concepts are self, and, context of experience and behavior. Consistent with Mead's conceptualization, the self is defined as the acts of naming one's own activities. I pointed out that a clearly relevant and crucial aspect of the self as defined, is the class of self naming-self desig- nating acts in which the person engages. Therefore I defined self identification as the 'class of self designating behaviors which the per- son takes toward himself as an object. Based on Mead's discussions I defined context of experience and behavior as a construct referring to the activities in which the person is directly or indirectly implicated with other persons . There - fore, religious context of experience and behavior is a construct which refers to the activities in which the person is directly or indirectly im- plicated others where the object of the activities is religious. 209 The propositions are central to Mead's consideration of the basic processes which produce and maintain self identification and they were the basic points of departure for the hypotheses sub- mitted to test in the present study. The first proposition specifies a relationship between the identification behaviors directed toward the person by the others with whom he is engaged in interaction, and, the identification behaviors which the person directs toward himself as an Object. Proposition 1. The self, defined as the acts of naming one's ac- tivities, is produced by the behaviors which others direct toward the person. The second proposition takes into account the implication of the person in a variety of different activities or contexts of exper- ience and behavior and the consequence of this varied implication for the production and maintenance of selves and self identifications. Proposition 11. "Selves" are produced in the various "contexts of experience and behavior" by the corresponding activities others direct toward the person. The third proposition makes a statement of general rela- tionship between self and context of experience and behavior. Proposition 111. The self corresponds with the behaviors directed toward the person by others in the various con- texts of experience and behavior in which the per- son and others are implicated. The fourth proposition specifies the relationship between the specific class of self identifications and the experiential base upon which it rests that was considered in the present investigation. Proposition IV. The person who is implicated in a religious con- text of experience and behavior will identify him- self as religious. The first proposition is the primary hypothesis within Mead's theoretical orientation. A direct test of this hypothesis requires observations of the behaviors which others direct toward the person in 210 relation to the identification behaviors which the person directs toward himself. The present study was an attempt to account for empirical data at the disposal of the investigator prior to the formulation of the propositions stated above. These data did not include observations of the behaviors which others direct toward the person nor did they include the person's reports of such behaviors directed toward him by others within a religious context of experience and behavior. The data included the person's report of the circumstances in which he engages in interaction with others who are in a position to direct religious identifying behaviors toward him; observations of the person's religious activities and observations of his report of religious activities in the presence of persons who are in a position to direct re- ligious identifying behaviors toward him; and, the person's report of religious activities taken outside the presence of other persons . All of these observations, I suggested in both Chapter I and Chapter II, are logically consistent referents for the construct of "religious con- text of experience and behavior. " That construct was defined as the activities in which the person is implicated, directly or indirectly, with others where the object of the activities is religious . In my discussion of the propositions in Chapter Ithe point was made that religious context of experience and behavior obviously does not produce religious self identification. Rather, the self is produced by the behaviors which others direct toward the person. The person who is interacting with others in a religious context of experience and beha- vior is placed in a position where others should direct religious iden- tification behaviors toward him, resulting in a religious self identifi- cation as suggested in Propositions III and IV. Based on this assumption, the greater the extent of the person's implication in a religious context of experience and behavior the greater the likelihood that he will be in a position where others can direct religious identifying behaviors toward him . 211 Based on these considerations 1 extended the discussion of the logically consistent referents for the construct of religious context of experience and behavior. As that construct refers to the activities in which the person is directly or indirectly implicated with other, I attempted to specify the different classes of religious activities which could be involved. I specified three different classes of activities in which the person is implicated, directly or indirectly, with others. Class 1 activities refers to the religious activities taken by the person in the presence of or with other persons who provide, share, and sus- tain the ongoing definition of the activities as religious. Class 2 ac- tivities are those religious activities taken by the person in the pre- sence of persons who did not provide the person's definition of the activities but who may share and thus confirm that "meaning, " or, persons who may not share the person's definition of the activities. Class 3 refers to those religious activities taken by the person outside the "immediate physical presence" of other persons. In my discussion ofthe relevant others for Class lreligious ac- tivities I indicated the necessity for reconceptualizing Mead's notion of "generalized other" to refer to that set of others with whom the person is involved and must take into account in the activities in which he and they are involved. I pointed out that the relevant set of others may well vary from one context of experience and behavior to another . In my discussion of Class 2 activities I attempted to indicate that only those persons with whom the person is engaged in a context of exper- ience and behavior are in a position to define those activities and to define or identify the person in terms of his participation in the acti- vities. If the context of experience and behavior is a universal one, then the category of relevant others is large and comprehensive. If the context of experience and behavior is a more restricted one the category of relevant others includes only those persons with whom the person is involved in that restricted set of activities. I suggested that the religious context of experience and behavior is a restricted one in 212 terms of the relevant others who are in a position to define the acti- vities and the person in terms of his participation in the activities. As Mead put it, the meaning or definition of activities is as universal as the community of others whose responses make up that definition. Finally, I pointed out that the implication of others in the person's activities with a specific or general context of experience and beha- vior is not limited by time and space boundaries. Therefore, Class 3 religious activities, or activities taken outside the presence of other persons, is still defined in terms of the responses of relevant others, even though they are not physically present. I then discussed the religious others who should be in a position to direct religious identifying behaviors toward the person and his activities and/or circumstances in which those others would be in a position to direct such behaviors toward the person. This specification included the others with whom the person is implicated in the activities of religious groups and organizations as well as the others in categories of interaction where religious activities should take place. Participation in such activities places the person in a position where others can address religious identifying behaviors toward him. Four categories of interaction and/or activities were spe- cified as logically consistent referents for the construct of religious context of experience and behavior (RCEB) . RHOMOG was discussed as a context of religious interaction. It refers to the common reli- gious preference of the person and his close friends and was presented as a category of interaction where others could direct religious iden- tifying behaviors toward the person given the common object of reli- gious preference for those persons involved. The person's most im- portant group, or MIG, was discussed as a second context of interaction where others could direct religious identifying behaviors toward the person if the person defines that most important group as a religious group. RBEL was discussed as a Class 2 religious activity, referring 213 to the person's subscription to statements of religious belief. RPART was discussed as the person's participation in the formal and informal patterns of activities of religious groups and organizations such as attendance at religious activities, prayer, meditation, proselytization, etc. As such RPART cuts across all three classes of religious acti- vities discussed above. Based on these considerations of the four component re- ferents for the construct RCEB-religious context of experience and behavior -the major hypothesis and the four related hypothesis were stated as follows: Major Hypothe sis: The greater the extent of the person's implication in a religious context of experience and behavior the greater the likelihood that person will make religious self iden- tification statements. Sub -Hypothe se 5 1. The greater the extent of RHOMOG the more likely the person will make religious self identification statements . 2. The person who defines his MIG as a religious group will make religious self identification statements. 3. The greater the extent of RBEL the more likely the per- son will make religious self identification statements. 4. The greater the extent of RPART the more likely the person will make religious self identification statements. Four sets of operational procedures were specified and dis- cussed as valid and intersubjective observations of the referents for the construct RCEB as described in the preceding paragraphs. RHOMOG was discussed as an index of the extent of religious homogeneity pre- ference among the person and his close friends. MIG was discussed as an index of the person's definition of his most important group as religious or non-religious. RBEL was discussed as an ordinal Gutt- man scale indicating the extent of the person's subscription to five statements of religious belief. RPART was discussed as an ordinal 214 Guttman scale indicating the extent of the person's participation in the religious activities described in the preceding paragraph. The Self Identification Problem, or SIP, was described and discussed as the operational procedure specified for observing self identification as defined. This technique simply asks the person to make as many statements as he can in response to the question "Who Am 1?". beginning with the most important statement first. Based on the discussion of these indices the major hypo- thesis and related sub -hypotheses were then stated in operational form. I will repeat the alternative form of the hypotheses, which specifies the direction of relationship, as the basis for my discussion of the bearing of the obtained observations on the hypotheses as stated. Sub -Hypothe se 5: l. H : The greater the magnitude of persons' scores on the RHOMOG index the more likely those persons will make religious self identification statements in re- sponse to the SIP. 2. H : Those persons who define their MIG as religious are more likely to make religious self identification state- ments in response to the SIP than are those persons who do not define their MIG as religious. 3. H : The greater the magnitude of persons' scores on the RBEL index the more likely those persons will make religious self identification statements in response to the SIP. 4. H : The greater the magnitude of persons' scores on the RPART index the more likely those persons will make religious self identification statements in response to the SIP. Major Hypothe s is: 1. H1: The greater the magnitude of persons' ordered place- ment on the RCEB indices separately or in combin- ation, the more likely those persons will make reli- gious self identification statements in response to the SIP. 215 Relation of Observations to the Hypotheses The hypotheses were submitted to test on the RCEB sample constructed of those persons for whom observations were available on the RHOMOG, MIG, RBEL, RPART, and Self Identification indicies. Examination of these 915 cases indicated that RHOMOG is not related to Self Identification. The remaining sub -hypotheses involving the relationship between MIG, RBEL, RPART, and Self Identification were confirmed. The deletion of RHOMOG from consideration as a referent for the construct of RCEB required a re -examination of the theoretical orientation and a reorganization of the analytical procedure. That re-examination resulted in the conclusion that RHOMOG was not a category within which religious identifying behaviors should be directed toward the person for the reasons set forth in Chapter 11. While the person may share religious preference in common with his close friends he also shares many other activities and objects in com- mon with them which could also be the focus for identifying behaviors . As a general category of interaction, and with that information alone, it is of little relevance in terms of placing the person in a position where specific identifying behaviors should be directed toward him by others. The hypotheses were then submitted to test on that sample of persons for whom observations were available on the remaining RCEB components of MIG, RBEL, and RPART, and, Self Identification. The three sub -hypotheses specifying the direction of association between MIG, RBEL, RPART and Self Identification were all confirmed. The extent of association ranged from slight to moderate. RPART and Self Identification yielded the highest extent of association with V = . 42. Concomitantly the per cent of cases accounted for in the predicted cells by each of these RCEB components was 55 per cent for the MIG and Self Identification relationship; 59 per cent for the RBEL and Self Iden- tification relationship; and 72 per cent for the RPART and Self 216 Identification relationship. Based on both the measure of extent of as so- ciation and the 51% Index, RPART was shown to be of greatest consequence for per‘SOns making statements of religious self identification. The major hypothesis specifies that as the extent of persons' implication in RCEB is increased the proportion of those persons making religious self identification statements will increase. When the RCEB component variables were considered separately in relationship to Self Identification this hypothesis was confirmed. Four sets or combinations of these RCEB components were then examined in relation to self iden- tification. Each set of observations confirmed the specifications of the major hypothesis. When RBEL-together-with-MIG was examined in relation to Self Identification, the extent of association was V = . 28, and the per cent of cases accounted for in the predicted cells was . 61. When RPART -together-with-MIG was examined in relation to Self Iden- tification the extent of association was V = . 36, and the 81% Index indi- cated the 73 per cent of all cases were accounted for in the predicted cells. When RBEL-together—with-RPART was examined in relation to Self Identification the extent of association was . 39 and the 51% Index indicated the 74 per cent of all cases were accounted for in the predicted cells. Finally, when RBEL-together-with RPART -together -with MIG was examined in relation to Self Identification the extent of association was .41 and 77 per cent of all cases were accounted for in the predicted cells. Thus, consistent with the specification of the major hypothe - sis, as the extent of persons' implication in a religious context of exper- ience and behavior is increased, the proportion of persons identifying themselves as religious does increase. The extent of association be- tween RCEB component( 5) and Self Identification ranges from only slight to moderate. Nevertheless, all associations are in the predicted direction and the extent of association does increase as additional obser - vations of the extent of persons' implication in a religious context of ex- perience and behavior are taken into account. 217 What conclusions can be drawn having examined the rela- tionship of the observations to the hypotheses. The examination of the bivariate relationships employed to test the sub -hypotheses clearly reveal that RPART--the extent of persons' participation in the formal and informal activities of religious groups--is of greatest consequence for persons' identification of themselves as religious. When the three RCEB components of RPART, RBEL and MIG were examined simultaneously be means of Guttman's scalogram analysis procedure, RPART was again determined to be the most important component in the internal rank order arrangement of those three variables. The extent of association between scale types for this order arrangement and Self Identification was less than in previous examinations but was significant and in the predicted direction. Given the overall pattern of relationship in the data and the determination of the most important or consequential of the RCEB components in relation to religious Self Identification, the conclusions which can be drawn are quite clear if we reconsider the specifications of the theoretical orientation within which the study was developed and carried out. The primary proposition in Mead's discussion of self- other relationships holds that the self is produced by the behaviors which others direct toward the person in terms of the contexts of experience and behavior in which the person and others are involved. Consequently the specifications of the theory direct our attention to the activities in which the person is engaged and in terms of which others direct identifying behaviors toward the person. The crucial observations which should be made are of those identifying behaviors which others direct toward the person in a specific context of experience and behavior and the identifying be- haviors which the person directs toward himself in terms of that con- text. Observations Of the behaviors of the others were not available to me in the data employed for this study. The next most logically relevant set of observations were available to me. Those observations 218 included reports and/or observations of the religious activities in which the person is engaged and his reports of categories of inter- action within which religious activities should take place. When the person engaged in those activities he is directly or indirectly implicated with others who can direct religious identi- fying behaviors toward him in terms of the activities. Therefore, in retrospect, the religious participation index of RPART should yield the highest empirical relationship with the presence of religious self identification statements. RPART is an index of the extent to which the person places himself in a position where others can address religious identifying behaviors to him. I Thus, the data bearing on this relationship inferentially support the assertion that the self, defined as the act of naming one's own activities, is produced by the behaviors which others direct toward the person in terms of the con- text of experience and behavior in which the person and others are involvedJ The next highest relationship observed in this study was between RBEL--the extent of persons' subscription to statements of religious belief--and Self Identification. This pattern of relationship was observed in both the bivariate and multivariate case. These statements of religious belief take place in the presence of the inter- viewer, a non-religious other, and thus are considered as a class of activities which can be confirmed as religious activities by non- religious others but most certainly are confirmed and defined as religious activities by the person's religious others whether or not those others are physically present. Thus the person's commitment of these religious activities places him in a position where others can identify him as a religious person. Persons subscribing to these religious statements of belief are thus engaging in one aspect of a religious context of experience and behavior. The extent of their subscription is related to the presence or absence of religious self identification statements on their part. This RCEB component, 219 however, provides an index of only one aspect of persons' impli- cation in a religious context of experience and behavior whereas the RPART component provides an index of several aspects of per- sons participation in such a context. Thus the observed relationship between RBEL and Self Identification was less in both the bivariate and multivariate cases. Finally, the person's indication that his most important group is a religious group provided the lowest extent of association with the dependent variable of Self Identification. MIG, as a cate- gory of religious interaction, was discussed in comparison with RHOMOG, the only RCEB component that was not found to be related to Self Identification in the present study. No association was ob- served between the extent of religious preference homogeneity among the person and his close friends. RHOMOG was thus re-examined and redefined as a general category of interaction wherein the person and his close friends may share a number of Objects and activities in common which may be the focus for identifying behaviors. In comparison, the person's definition of his most important group as a religious group clearly specified that category of interaction as a religious one within which religious identifying behaviors should be directed toward the persons involved. Nevertheless, the knowledge that this category of interaction is a religious one was of little conse- quence in terms of accounting for those persons who would identify themselves as religious. Therefore, the logical extensions of the theoretical orientation which were developed in Chapter 11 must be carefully reconsidered. There seems to be empirical confirmation for the extension which suggested that attention to persons' participation in the activities making up a context of experience and behavior should be of consequence for their identification of themselves in terms of that context. The greater the extent of such participation the more likely those persons were, in the present study, to identify 220 themselves as relitious. There is little empirical confirmation for the extention which suggested that attention to categories of interaction, even when specifically related to the context of experience and be- havior in question, would be of consequence for persons' identifications of themselves as religious. Given this lack of empirical relationship between categories of interaction, and my subsequent examination of the logical relation- ship of such referents to the construct of context of experience and behavior, subsequent investigations of the relationship between Self Identification and any context should be formulated so as to focus upon the activities in which persons are engaged relevant to the con- text in question and the extent of persons' participation in categories of interaction that are logically related to that context. Mere know- ledge of the category of interaction or of the person's affiliation with such a category, like the mere knowledge of group affiliation, is not productive of empirical relationship with self identification in terms of the context of experience and behavior to which the category of interaction is related. The most limiting aspect of such a focus is the failure to consider the processes, or logically related aspects of those processes, which the theory indicates are productive of self identification. Attention to such attributes as group membership or interaction category affiliation deviates from the specifications of the theory which requires attention to the processes and activities in which persons are implicated with others . The present investigation presented empirical evidence which confirmed the major hypothesis under test. The basis for that empirical confirmation, in summary, has been the attention given to the activities which make up the context of experience and behavior under consideration. It is those activities which place persons in the position where others can address identifying behaviors toward them, resulting in their identification of themselves in terms of those activi- ties. 221 All cases were not accounted for in the present investiga- tion. The most exceptional deviant cases were re-examined in terms of additional available information on the extent of persons' partici- pation in the formal and informal activities of religious groups. The additional available information was not sufficient to account for those deviant cases. This was so despite the fact that the overall pattern of Observed relationships in the study suggests that additional obser- vations of the persons' implication in a religious context of experience and behavior should provide sufficient differentiation to account for those deviant cases. Despite this exception and despite the slight to moderate extent of observed association between RCEB and Self Identification, I would assert that were additional observations avail- able on the extent of persons' participation in religious activities, the majority of those deviant cases would be accounted for. That I have no way of testing this assertion is quite clear. I base the assertion on the moderate but consistent pattern of association bearing on the major hypothesis in the present study: as the extent of persons' implication in a religious context of experience and be- havior is increased the proportion of persons identifying themselves as religious also increases. Such implication places those persons in a position where others can identify them as religious. Based on the results of the present inquiry it is this set of behaviors which appears to be the crucial process producing and maintaining persons' conceptions and identifications of themselves as religious. Proposals for Future Research This investigation has demonstrated that the activities in which persons are engaged places them in a position where others can address identifying behaviors toward them, resulting in those persons identifying themselves in terms of those activities. Con- sistently, the assumption has been made that others do address identi- fying behaviors toward the person in terms of the activities in which 222 he and they are engaged. NO inquiry has been conducted which has demonstrated a consistent pattern of relationship between the identi- fying behaviors which others :12 address toward the person in a specific context of experience and behavior and the identifying be- haviors which that person directs toward himself. While the present investigation has presented empirical confirmation for propositions logically derived from and consistent with Mead's primary proposition concerning the production of the social self, additional investigation is required to directly investigate that basic statement of relationship. Subsequent investigation should be undertaken to examine the relationship between the identifying behaviors of the others with whom persons are engaged in religious and other contexts of experience and behavior and the persons' identifications of themselves. Several contexts of experience and behavior are of particular interest to the present investigator but the proposed procedures for carrying out the investigation could be extended to other contexts of interest. The religious, familial, and occupational contexts of experience and be- havior are three specific contexts which could be employed to directly test Mead's basic proposition. I do not intent,to lay out a detailed research proposal at this point but gerely to indicate some general outlines for such an exploratory investigation. I will use, by example, the occupational context of experience and behavior. The procedure would be to select a number of persons representing different occupational categories with respect to the skill, training, and comprehensivevnature of the occupation's demands. The initial observations on these persons would be to record their response to the Self Identification Problem. This could be accomplished with the conventional written format, which probably yields a larger number of responses, or with the oral format employed in the present study. Following the recording of responses to the Self Identification Problem the investigator should next seek to determine the names of the others which the respondent takes into account or with whom he 223 must deal in carrying out the tasks related to his particular occupa- tional pursuit. With this information available the investigator would thus have the identify of a set of others with whom the person is im- plicated in his occupational context of experience and behavior. The next set of observations would be taken from those persons identified by the previous respondent. Having indicated to the persons that their names had been provided by the respondent, discussed above, they would then be asked to complete the Self Identi- fication Problem for that respondent; i.e. , instead of asking them to make, or write, as many statements as possible in response to the question "Who Am 1?" they would be asked to make, or write, as many statement as possible in response to the question "Who Is He ?" Then, using a similar form of content analysis as was employed in the present study, the investigator could simply examine the occu— pational and/or work-related statements which the respondent makes to the Self Identification Problem in relation to the occupational and/or work-related identification Statements which the others make in response to the question "Who Is, He? " The outlines of this proposed investigation are quite general and admittedly have not been developed with the precision and definition which would be required in a research proposal. I have made some tentative explorations with the use of the "other identifi- cation" technique, "Who Is He ?" and "Who Is She ?" The preliminary results indicate that subsequent use and development of such an ob- servation technique may be appropriately and successfully exploited. The consequence of this exploitation and of the proposed research procedures for examining Mead's basic hypothesis concerning the production of the social self is an empirical question. Given this condition,it is appropriate to conclude these remarks with the conven- tional assertion that future investigation in the area of self-other rela- tionships is both required and desirable. APPENDICES APPENDIX A SELF IDENTIFICATION PROBELM Introduction by the Interviewer to the Re sponsent Now we have quite a different thing for you to do. Although probably new to you, it is easy and I think you will find it quite enjoyable. Everyone we have asked to do this has found it to be interesting. Now let me tell you what we have in mind. Ask this question of yourself, ”Who Am I? " Think of as many answers as you can in answer to the question, "Who Am I? " In a moment I would like you to give me the answers as if you were giving them to yourself, not to me or anyone else. Take a little time to think about it. (INTERVIEWER PAUSES HERE MOMENTARILY) Now please make what you consider to be the most important statement about yourself first. Specified Probes for the Interviewer If the respondent began to make statements in response to the question and continued spontaneously the interviewer simply re- corded those responses verbatim. If, however, the respondent made a response or responses and then stopped, the interviewer utilized the following probe procedure. Now make what you consider to be the next most important state- ment about yourself. If the respondent continued to make statements the inter- viewer simply recorded them. If not, the interviewer continued to probe in the above manner up to a limit of three minutes. The final 225 226 probe question which the interviewer could employ was, Are there any other statements you could make about yourself in answer to the question, "Who Am I? " Instructions by the Investigator to the Interviewer Record the answers to this question in the respondent's own words. Do not edit. This is the most important consideration. Differences such as "I am a student, " and "I am an average student" are significant. These would be missed if you simply recorded "student. " When possible, without interfering with the above objective, record a single statement on each line. For our purposes, a statement would be a sifle unit of thought or meanifl. The sentences "I am a student, " "I am an average student, " or just "student, " are all clear examples of what we mean by a unit of thought. However, frequently the respondent will say for example, "I am a mother, wife, and dishwasher. " While expressed in a single sentence, these are 3 units of thought. These should be recorded as follows: Line 1. I am a mother Line 2. A wife Line 3. A dishwasher On the other hand, a sentence such as "I am avery hard- working bricklayer" would be recorded on one line as a single unit of thought. The general rule then, is separate into distinct statements, sen- tences which contain 2 or more nouns or verbs. Modifiers do not produce (for our purposes) separate units of thought. An exception to the rule would be illustrated by this sentence. "I am tired, bored, and bewildered. " Again this sentence Should be broken down into three statements and recorded on 3 different lines. Don't let the above interfere with recording exactly what the respondent says. We know that this item is not as easy to administer as most others. It is new to both respondent and interviewer. However, it has been used rather widely and successfully. People will respond to it much as they will to other rather personal items provided you present it with assurance and are able to give the respondent time to consider the task demanded by the question. 227 Initially, some respondents find it difficult to form statements about themselves. Probe as often as necessary. Read or reread the item to the respondent, just as it appears in the questionnaire. In some cases, the respondent will make an opening statement such as "nothing, " "I'm nobody, " "I'm not important to anyone. " Most frequently such statements are followed by long explanation of who he is and who he is impor- tant to. To Show the respondent that anything he says regarding who he is, is important information, take down such a statement on line 1. Once the respondent sees that he is free to say any- thing he wants to about himself and your only reaction is to write it down, it will be easier for him to make statements. In general, probe enough on this item to indicate to the respondent that it is important information foxi our purposes. Heed the time, but remain flexible, so that a respondent who starts making statements and free -wheels with his statements using little or no probing, past the three -minute maximum should not be stopped. stopping him could seriously damage rapport, also. If you are successful in obtaining 5 statements (the minimum) continue to probe for up to 10 statements, but do not probe beyond that point. Only if the respondent makes more than 10 statements without probing should they appear on the page. When it is necessary to probe before the first statement is made, 'place a horizontal line above the number 1. When initials of organizations are given, ask respondent for names of organization. The may be abbreviated in understandable (e. g. Amer. for American) form. General Coding Instructions for Statements Made in Response to the Self Identification Problem The object is to code all statements as separate items. A STATEMENT CAN BE A SINGLE WORD, A PHRASE, OR A COMPLETE SENTENCE. Thus you may find more than one statement on a particular line in the questionnaire. Each such statement will be coded separately. Examples: a. carpenter; a happy fellow; I am a good father. b. I am a father and husband. (This is two statements . ) c. I am a very hardworking, unhappy father. (This is one statement.) d. I an handsome but weak. (This is two statements . ) e. I am a young single girl. (This is three statements . ) f. I am a American Negro citizen. (This is three statements . ) 228 GENERAL RULE: Any element which can be constructed as a single thought about oneself will be coded as a statement. A noun and its accompanying modifiers is usually a single thought. Adjectives modifying nouns are to be included with the noun (as in example c. above). Adjectives which stand alone as self descriptions (as in example d. and e. and f. above) will be counted as statements. Go through question 17 and mark off statements. Usually only one Statement will be found on a line. In those cases where more than one statement is found on a line, code each of the statements in order, beginning with the first mentioned state- ment. Each statement will be coded into a two column code described below. When you get to ten statements, ignore the rest. In case there are less than 10 statements, put code 88 into the unfilled columns. Coding of this item can best be done by following two procedures: a. Matching statements on the protocols with statements provided as examples for each catogory (see following pages). b. Following these General Instructions with special attention to the Arbitrary Rules (see following pages) . The titles of the categories are only suggestive of the range of statements to be included in each category. Basically, the category serves to collect statements which have some common referent. However the logic of placing certain statements in the same category may not always be apparent. Attention to such questions as "What did the respondent really mean to say?", or "How can this category placement make sense in the context of what the respondent said in another statement?", will lead to random coding, and low inter-coder reliability. Therefore, the best check on where a statement is to be coded is to try to find an example like the statement in the list of examples, or one clearly close to it. A final check should be made to make certain that the category assignment does not violate any of the arbitrary rules and exceptions. Definition of a qualified and unqualified statement. a. A QUALIFIED statement is any statement which is modified by the inclusion of an adjective or adverb. along with the noun and verb. Sometimes a verb will modify a statement, as in the statement "I have work. " 229 b. AN UNQUALIFIED statement is any statement which is not modified by the inclusion of an adjective or adverb. This would be only the noun or verb, in cases where the verb has no qualifying effect. Examples: I am a poor student. (This is a qualified statement. ) I am a father. (This is an unqualified statement. ) 7. Each statement should be coded disregarding (a) negation or affir- mation and (b) verb tense. Examples: Treat the statement "I am not a mother. " as in the same class as "I am a mother". Treat the statement "I was a foreman. " as in the same class as “I am a foreman". 8. The distinction between "consensual" and "idiosyncratic" state- ments can best be understood by reading through the list of examples provided. All of the references in the list of idiosyn- cratic statements are to events, states of mind, or processes which are private, lacking in specificity, or lacking in general agreement as to meaning. On the other hand, all consensual statements have at least one word which points to something most people would recognize, and have some agreement for the meaning of. 9. Use a 99 to code all columns in the case of a clear refusal. We define a refusal as one of the following: a. I'd rather not answer that question. b. I don't want to answer that question. c. Would you please go to the next question. "I don't know, " and "I would rather keep that to myself" and not defined as refusals to the question but as reponses to the question which should be coded as Idiosyncratic (72) . ARBITRARY RULES A. In cases of statements which fit into two or more categories, assign the statement to the category closest to the top of the list numerically, with the following exceptions: 1. Any reference to health, whether consensual or not, should be coded as a 70 or 71, depending on the matter of qualifi- cation. 2. Any reference to an organized church or religion should be coded as 53 or 52 regardless of what other codable items are mentioned in the statement. D. 230 3. Any reference to race, irrespective of other references in the statement which are codable, should be coded 50 or 51. 4. General religiosity * references frequently include words that refer to kinship, sex, work, etc. However, all religiosity references should be coded as 48. (see examples.) Code the statement as it is recorded on the schedule. Do not add or subtract words or impute meanings. The only exception is described in rule C. below. In compound sentences with a single stem, treat the clauses as if each were separate statements with the stem attached to each clause . Example 5: "I am interested in my house, in recreation, and in education. " should be treated as the following three statements: a. I am interested in my house. b. I am interested in recreation. c. I am interested in education. "I am an American Negro citizen. " should be treated as the following three statements: a. I am an American. b. I am a Negro c. I am a citizen. "I am a young single girl. " should be treated as the following three statements: a. I am young. b. I am single c. I am a girl. In cases where titles are mentioned which include several words, treat the entire title as the object of the sentences. Example 8 : lo 2. I am a U. S. citizen (Do not treat U. S. as qualifying citizen. ) I am an executive secretary (Do not treat executive as qualifying secretary) I am a retired boilermaker(Do not treat retired as qualifying boilermaker. ) I am a successful medical doctor. (Treat successful as qualifying medical doctor) Code 48 52 231 When coding initials (e. g. . B. P. O.E. , G. O. P. , I. O. O. F. , etc. ) which have not been defined by the interviewer on the schedule, all initials, whether they are recognized or not recognized by the coder, should be coded as 58 or 59 de- pending upon qualification. Example: "I am a B. P,O.E. " (This is an unqualified statement - 58) "I am a good B. P.O.E. " (This is a qualified l SIP Code Des cription Religiousity (Qualified or Unqualified) Excluding references to church, formal religion or religious organizations. I am one of God's workers put here to make this world better than when I came here. I try to serve the Lord in every way possible. I am a person who believes in God. I am a religious person. I am important to the Lord as an individual. I was blessed by the Lord in everything. I am a Christian lady. I am a child of God. 2 I am one of God's children. I lead a Christian life. I am a creature created by God. Reference to Relijion and Relipipus Orggnizations (unquali- fied) This would include all church or church-related groups, membership and participationin same. Jewish would be included in this category. Initials that cannot be recognized by the coder should be coded either 60 or 61 depending upon qualification. I am a Catholic. I go to church I'm a church member. I am Jewish I am a church goer. 1A complete listing of all coding categories and examples of statements included in those categories is on file with the Five Nations Study, Project 509, and may be obtained through the Research Laboratory, Department of Sociology, Michigan State University. 2 Where kinship referents are used here, (child, children) and sex referents (lady), disregard usual coding by kinship and sex and place all such referents to religiosity in category 48. 232 53 Qualified Reference to ReliLion and Relpgious Organizations 1 am a good church member. I am free to choose my religion. I am important to my church. I live by my religious beliefs. I am loyal to my church.3 I go to church every Sunday. I enjoy my church work. I go to church regularly. "work" is considered here as a modifier of church. Church is categorized as referring to Religious Organization and it is thus coded as 52 or 53 depending on qualification. It should not be pushed up to 42 or 43; i.e. , general or qualified general reference to work. APPENDIX B THE USE OF THE COUNTER-SORTER FOR SCALOGRAM ANALYSIS The method of scalogram analysis employed in this study was an adaptation of Waisanen's typewriter notation technique for use on the IBM counter sorter.1 Waisanen's technique is a convenient means for imposing data onto the Guttman scale pattern when the in- vestigator is working with raw data responses for any number of items and respondents. That technique was employed in constructing the RBEL and RPART indicies in the present study. The scalogram analysis of RCEB components, as discussed in Chapter 111, had to deal with the problem of employing data already punched into IBM cards in the form of total scores for each of the three indices considered. RPART was a two column score ranging from , 00 - 15. RBEL was a two column score ranging from 00 - 20. MIG was a one column score of 1 and 2. The task was to use this information as it appeared in the punched IBM cards; to transfer this information into a useable format for the scalogram analysis of all respondents' response patterns to the various item order combinations; and most important, to accomplish the above and yet remain within the guide- lines of the required criteria for scalogram analysis. Those criteria are that respondents be ranked from high to low in terms of their total score on all items, and that the items, or RCEB components in this instance, can be arranged from left to right in terms of their difficulty or importance. lFrederick B. Waisanen. "A Notation Technique for . Scalogram Analysis, " Sociological Quarterly, Vol. 1, No. 4 (November, 1960), pp. 245-252. 233 234 While the Guttman procedure permits the use of tri- chotomous responses the conventional procedure is to separate the responses to all items into the dichotomy of "correct" and "incorrect“. In the present study the latter scoring procedure was accomplished in the following manner. For RPART, a "correct" answer was arbitrarily judged to be all scores from 10 to 15, the fourth quartile for that index. For RBEL, a "correct" answer was arbitrarily judged to be all scores of 20, the fourth quartile for that index. For MIG, a "correct" answer was arbitrarily judged to be a score of 2, the score indicating the person's definition of his most important group as a religious group. This decision of "correct" and "incorrect" answers or responses to the three items resulted in the following three dichotomies . RPART RBEL MIG . 15 20 2 "Correct" - X :: :: :: {a I: I: 79— 73’ T." "Incorrect" - 0 I I :: 66 66 ’1' Thus a correct answer on all three items would give a total score of three and an incorrect answer on all three items would result in a total score of zero. With these item responses established it was then possible to initiate the scalogram analysis procedure. First it was necessary to decide on the response patterns for three items that would most closely approximate the formal model. That response pattern is re- produced here: 235 Item Item Item Scale Row Score A B C Type Errors 1 3 X X X 4 0 2 2 X X 0 4 l 3 2 X 0 X 4 l 4 2 '0— x X 3 0 5 1 0 X 0 3 1 6 1 0 ‘0— _x_ 2 0 7 1 X 0 0 l l 8 0 0 0 0 1 0 The reader can easily observe that such a pattern meets the two criteria for scalogram analysis. It maintains a rank ordering among persons from top to bottom in terms of their total score on the three items and it allows the ranking of the items under consideration fromleft to right in terms of their importance. The use to which the formal model was put in the present study attempted to determine the rank order of the various order combinations of RCEB components. Thus it was necessary to pre -establish the arrangement of items from left to right in order to examine the relative importance of the possible order combinations empirically. The next task, then, was that of determining the various possible order combinations for the three components. This was a simple task of deter mining all possible ways in which those items could rank in terms of one another. The possible order combinations for these three RCEB components are as follows: RPART RBEL MIG RPART MIG RBEL RBE L RPART MIG RBE L MIG RPART MIG RPART RBE L MIG RBE L RPART The response patterns of all respondents to each order combination were imposed onto the formal model. This was accomplished by using the data as contained in the IBM cards in the following manner. 236 The correct response for the first item (RPART 10-15) in the first order combination (RPART -RBEL-MIG) was selected on the counter sorter. All incorrect responses to RPART (RPART 00-09) 'were set aside and designated as Hold One. The correct responses on RPART (RPART 10-15) were then sorted to determine the nature of the response on RBEL, the second item in the first order combination Of RPART -RBEL-MIG. The incorrect responses to RBEL (RBEL 00-19) were set aside and designated as Hold Two. (Hold Two thus consists of all correct responses to RPART (RPART 10-15) and all incorrect responses to RBEL (RBEL 00-19) while the responses to MIG are as yet unknown.) The correct responses to RBEL were then sorted on MIG. This last determination provided the two response patterns in the first two rows of Figure 4 for the order combination of RPART -RBE L -MIG . Total Scale f Row Score RPART RBEL MIG Type Errors F Errrors 1 3 X X X 4 0 124 0 2 2 X X 0 4 1 32 32 The similar procedure was repeated on all respondents falling in Hold One; i.e., persons with a "correct" response to RPART (RPART 15) but with unknown responses to RBEL and MIG. For Hold One, the nature of the responses to RBEL were determined by sorting for the correct response on that index. The incorrect responses (RBEL 00-19) were set aside and designated as Hold Three; (i.e. ,2 persons with incorrect responses to RPART, incorrect responses to RBEL, and unknown responses to MIG). The correct responses to RBEL (RBEL 20) were then sorted to determine the nature of the responses to MIG. The results of this determination yielded the frequency of responses to response patterns in rows 4 and 5 of Figure 4. 237 Total Scale f Row Score RPART RBEL MIG Type Errors f Errors 4 2 0 X X 3 0 149 0 5 1 0 X 0 3 1 90 90 The similar procedure was then repeated for persons previously designated as Hold Two; i.e. , persons who responded correctly to RPART but incorrectly to RBEL and whose responses to MIG were unknown. These persons were thus sorted to determine the nature of their response to MIG. The results provided the frequency of response to response patterns in rows 3 and 7 of Figure 4. Total Scale f Row Score RPART RBEL MIG Type Errors f Errors 3 2 X 0 X 4 1 60 60 7 1 X 0 0 1 1 20 20 Finally, the sorting procedure was carried out for those persons previously designated as Hold Three; i.e., those persons with incorrect responses to both RPART and RBEL but with unknown responses to MIG. These persons were thus sorted to determine the nature of their response to MIG. The results provided the frequency of response to response patterns in rows 6 and 8 of Figure 4. Total Scale f Row Score RPART RBEL MIG Type Errors f Errors 6 l 0 0 X 2 0 197 0 8 0 . 0 0 0 1 0 318 0 Combining the above we have the frequency of response patterns for all persons to the items RPART, RBEL, and MIG as presented in Table 22 of Chapter IV. With this information it is then possible to compute the C.R., M.M.R. and I.O.C. for the order combination in that table. This is reproduced at the bottom of that table in Chapter IV. 238 With this information it is also possible to sort each re- sponse pattern for the other combination of RPART, RBEL, MIG, as well as all other order combinations we shall see, on the dependent variable of self identification as was reporduced in Cols 10 and 11 of Table 22. Collapsing frequencies for scale types it was then possible to produce the contingency table for the relationship between RCEB Scale Types for the Order Combination of RPART -RBEL-MIG and Self Identification as represented in Table 23 of Chapter IV. Most important, once the frequency of response patterns for any one order combination has been determined it is not necessary to repeat the sorting procedure discussed above for other order com- binations of the items. Thus the C.R., M.M.R.. and I.O.C. can be computed for all six possible order combinations with only one sorting procedure. The results of the scalogram analysis, the item response patterns, frequency distribution on scale types, frequency distribution or errors, C.R., M.M.R., and I.O.C. values for the remaining order combinations of RPART, RBEL and MIG will be presented below. The procedure described above for scalogram analysis of data contained in IBM punch cards is a convenient and useful technique when the investigator has a small number of items, knows all possible patterns of response to the items in their various order combinations and wishes to examine the frequency distribution of response patterns for those order combinations on a large number of respondents. It is appropriately suited to the task of determining the rank order or con- sequence of a number of items considered simultaneously as it was used in the present study. It allows the investigator to use the response data in the available location on IBM punch cards and it allows him to meet the relevant criteria of Guttman's formal model for using scalo- gram analysis . 239 Response Pattens, C.R., M.M.R., and I.O.C. Values for the RPART, MIG, RBEL Order Combination RPART MIG RBEL 4 3 2 1 _s_ E. F FE X x X (X) 4 0 124 0 X X 0 (X) 4 1 63 63 X 0 x (X) 4 1 32 32 0 X X (X) 3 0 149 0 0 X 0 (X) 3 1 197 197 0 0 X (X) 2 0 90 0 X 0 0 (X) 1 1 20 20 0 0 0 (X) 1 0 318 0 993 312 312 C.R. .1.00-?_’979 —.90 478 + 388 + 657 M.M.R. — 2’ 979 — .51 I.O.C. = .80 Response Patterns, C.R., M.M.R., and I.O.C. Values for the RBEL, RPART, MIG Order Combination RBEL RPART MIG 4 3 2 l s _E_ F FE x X X (X) 4 0 124 0 x x 0 (X) 4 1 32 32 X 0 X (X) 4 1 149 149 ' 0 x x (X) 3 0 63 0 0 X 0 (X) 3 1 20 20 0 0 l (X) 2 0 197 0 x 0 0 (X) 1 1 90 90 0 0 0 (X) 1 0 310 0 993 291 291 C.R. -1.00 - 2,979 - .90 680 + 388 + 565 M. . . = = . M R 2, 979 55 I.O.C.: '907'55 = .77 1.00 - .55 240 Response Patterns, C.R., M.M.R.. and I.O.C. Values for the MIG, RPART, RBEL Order Combination MIG RPART RBEL 4 3 2 1 s E F FE X X X (X) 4 o 124 o X X o (X) 4 1 63 63 X o X (X) 4 1 149 149 o X X (X) 3 o 32 o o X 0 (X) 3 1 20 20 0 o X (X) 2 o 90 o X o o (X) 1 1 197 197 o o o (X) 1 o 318 o 993 429 - 429 _ C.R. .1.oo—2’979 .86 657 + 388 + 478 M.M.R. — 2,979 — .51 .86 - .51 I.O.C. - 1.00““ .. .71 Response Patterns, C.R., M.M.R., and I.O.C. Values for the RBEL, MIG, RPART Order Combination RBEL MIG RPART 4 3 2 1 s E F FE X X X (X) 4 o 124 o X X o (X) 4 1 149 149 X o X (X) 4 1 32 32 o x x (X) 3 o 63 o o X o (X) 3 1 197 197 o o x (X) 2 o 20 o X o o (X) 1 1 9o 90 o o o (X) 1 o 318 o 468 C.R. = 1.00 .- 2,979 .84 993 468 688 + 565 + 585 M.M. . = = . R 2.979 62 I.O.C. = '84 " '62 = .58 1.00 - .62 241 Response Patterns, C.R., M.M.R., and I.O.C. Values for the MIG, RBEL, RPART Order Combination MIG RBEL RPART 4 3 2 1 x X X (X) X X o (X) x o X (X) o X X (X) o 3; 0 (X) o o X (X) X o 0 (X) o o o (x) - 499 _ C.R. - 1.00 - 2,979 .83 657 + 458 + 478 M.M.R. — 2,979 — .58 I.O.C. = ‘83 ’ '53 = .64 1.00 - .53 I—av-‘Nwwrhnbrh It!) ot-‘Or—‘OI—‘b—‘o I'Tl 124 149 63 32 90 20 197 318 993 197 499 BIBLIOGRAPHY Bain, Read. "The Self-and-Other Words of a Child. " American Journal of SocioloiX. Vol. 41 (1936), pp. 767-775. Becker, Howard S. The Outsiders. New York: Free Press of Macmillan, 1963. Blalock, Hubert M., Jr. Social Statistics. New York: McGraw Hill Book Co.. 1960. Cooley, Charles Horton. "A Study of the Early Use of the Self Words by a Child, " Psychological Review, Vol. 15 (1908), pp. 339-347, and reprinted in his Sociological Theory and Social Research. New York: 1930. Cottrell, Leonard, and Foote, Nelson. “Sullivan's Contribution to Social Psychology, " in The Contributions of Harry Stack Sullivan. Edited by Patrick Mullahy. New York: Hermitage House, 1952. Couch, Carl J. "Alienation and Self Identification. " Paper read at the American Sociological Association Annual Meet- ings, Los Angeles, California, 1963. "Self Attitudes and Degree of Agreement with Immediate Others, " American Journal of Sociology, Vol. 63 (March, 1958). PP. 491-496. "A Study of the Relationships between Self-views and Role -taking Accuracy. " Unpublished Ph. D. dissertation, State University of Iowa, 1955. Cramer, Harald. Mathematical Methods of Statistics. Princeton: Princeton University Press, 1946. Dinitz, S., Mangus, A. R., and Pasamanick, B. "Integration and Conflict in Self-Other Conceptions as Factors in Mental Illness, " Sociometry, Vol. 22 (March, 1959), pp. 44-55. 2410.2 242 Paris, Ellsworth. "The Retrospective Act, " Journal of Educational Sociology, Vol. 14 (October, 1940), pp. 79-91. Feigl, Herbert. "Operationism and Scientific Method, " in the "Symposium on Operationism, " The Psychological Review, Vol. 52, No. 5 (September, 1945), pp. 250-258. Foote, Nelson. "Identification as a Basis for a Theory of Motivation, " American Sociolflical Review, Vol. 16 (February, 1951), p. 17. Francis, Roy G. The Rhetoric of Science. Minneapolis: University of Minnesota Press, 1961. Frank, Phillip G. (ed. ) . The Validation of Scientific Theories, New York: Collier Books, 1961. Friedman, I. "Phenomenal, Ideal and Projected Conceptions of Self, " Journal of Abnormal and Social Psychology, Vol. 51 (1955), pp. 611-615. Gallup Organization. Report to the Department of Sociology and Anthropology. Michigan State University. GO/6368 GP, GO/6369 RM, and GO/6370 SW. Princeton, New Jersey, March, 1964. Garfinkel, Harold. ”Studies of the Routine Grounds of Everyday Activities, " Social Problems, Vol. 11 (Winter, 1964), pp. 225-250. Garretson, W. S. "College as a Social Object: A Study in Consensus.‘ Unpublished Ph.D. dissertation, State University of Iowa, 1961. "The Consensual Definition of Social Objects, " Sociological Quarterly, Vol. 3 (April, 1962), pp. 107-113. Goffman, Erving. Behavior in Public Places. New York: Free Press of Macmillan, 1963. The Presentation of Self in Everyday Life. Garden City, New York: Doubleday Anchor Book Co., 1959. Goodman, Leo A. , and Kruskal, William H. "Measure of Association for Cross Classifications, " Journal of the American Statistical Association, Vol. 49 (December, 1954), pp. 761-762. 243 Guttman, Louis. "The Problem of Attitude and Opinion Measurement, " in Measurement and Prediction, Vol. IV, "The American Soldier: Studies in the Social Psychology of World War II. " Edited by Samuel Stouffer. Princeton: Princeton Univer- sity Press, 1951. Hagood, Margaret, and Price,_ Daniel. Statistics for Sociologists. Revised ed. New York: Holt, Rinehart and Winston, 1952. Hartung, Frank. "Manhattan Madness, " Sociological Quarterly, Vol. 4, No.3 (Summer, 1963) pp. 261-272. Helper, M. "Learning, Theory and the Self Concept, " Journal of Abnormal and Social Psychology, Vol. 51 (1955), pp. 184-194. Hockett, Charles D. "The Origin of Speech, " Scientific American, (September, 1960). Jackson, J. M. "A Simple and More Vigorous Technique for Scale Analysis, " A Manual for Scale Analysis, Part II. Mon- treal: McGill University, 1949. (Mimeographed.) Kitsuse, John. "Societal Reaction to Deviant Behavior: Problems of Theory and Method, " in The Other Side. Edited by Howard S. Becker. New York: Free Press of Mac- millan, 1964. Klapp, Orrin E. Heroes, Villains, and Fools. Engelwood Cliffs, New Jersey: Prentice -Hall, Inc. , 1962. Kuhn, Manford, and McPartland, Thomas. "An Empirical Investi- gation of Self Attitudes, " American Sociological Review, Vol. 19 (February, 1954), pp. 68-76. Kuhn, M. H. "Self Attitudes by Age, Sex, and Professional Training, " Socioloifical Quarterly, Vol. 1 (January, 1960), pp. 39-55. "The Social Act, " in A Dictionary of the Social Sciences . Edited by Julius Gould and William Kolb. New York: Free Press of Macmillan, 1964. Lundberg, George A. "The Natural Science Trend in Sociology, " American Journal of Sociology, Vol. 61 (November, 1955), pp. 191-202. 244 Lundy, Richard M. "Self Perceptions and Descriptions of Opposite Sex Sociometric Choices, " Sociometry, Vol. 19, No. 4 (December, 1956), pp. 272-277. McNemar, Quinn. Psychological Statistics. New York: John Wiley and Sons, Inc., 1955. Mead, George Herbert. Mind, Self and Society: From the Standpoint of a Social Behaviorist. Edited by Charles W. Morris. Chicago: University of Chicago Press, 1934. Philosophy of the Act. Chicago: University of Chicago Press, 1938. McPhail, Clark. "Perceived Consensus Regarding Statements about the Self in Response to the Question, 'Who Am I?', " Paper read at the Ohio Valley Sociological Society Meet- ings, Columbus, Ohio, Spring, 1964. "Religious Preference, Religious Activity and Religious Self Identification. " unpublished study, Department of Sociology, Michigan State University, 1963. McPartland, Thomas S. Manual for the Twenty Statements Problem. Kansas City: Kansas City Mental Health Foundation, 1959. (Dittoed.) "Self Conception, Social Class and Mental Health, " Human Organization, Vol. 17, (1958), pp. 24-29. "The Self and Social Structure. " Unpublished Ph. D. dissertation, State University of Iowa, 1953. Cummings, J., and Garretson, W. "Self Conception and Ward Behavior in Two Psychiatric Hospitals, " Sociometry, Vol. 24 (June, 1961), pp. 11-24. Merton, Robert K. Social Theory and Social Structure. Revised ed. Glencoe, ML: The Free Press, 1957. Mills, C. Wright. "Language, Logic and Culture, " American Socio- logical Review, Vol. 4, No. 5 (October, 1939), pp. 670-680. Miyamoto, Frank, and Dornbusch, Sandord. "A Test of Inter- actionist Hypotheses of Self-Conception, " American Journal of Sociology, Vol. 61 (March, 1956), pp. 399-403. 245 Mulford, Harold A. , and Salisbury, Winfield W. , II. "Self-Concep- tions in a General Population, " Sociological Quartery, Vol. 5, No. 1 (Winter, 1964), pp. 35-46. Mullahy, Patrick (ed. ) . The Contributions of Harry Stack Sullivan. New York: Hermitage House, 1952. Newcomb, Theodore. Social Psychology. New York: Dryden Press, 1950. Orbach, Harold. "Operational Definitions and the Natural Science Trend: A Methodological Note, " Midwest Sociologist, (May, 1957), pp. 101-103. Reck, Andrew (ed. ) . Selected Writings of G. H. Mead. Indianapolis: The Bobbs-Merrill Company, Inc., 1964. Sherwood, John J. "Self Identity and Referent Others, " Sociometry, Vol. 28, No. 1 (March, 1965), pp. 66-81. Siegel, Sidney. Nonparametric Statistics for the Behavioral Sciences. New York: McGraw-Hill Book Co., 1956. Stewart, R. L. "The Self and Other Objects: Their Measurement and Interrelationship. " Unpublished Ph. D. dissertation, State University of Iowa, 1955. Summers, Worth. "Social Structure, Behavior, and the Meaning Component of Self-Syrnbols. " Unpublished M.A. thesis, Department of Sociology and Anthropology, Michigan State University, 1964. Tucker, Charles W. "The Dimensions of Self-Attitudes: A Working Paper. " Department of Sociology, Michigan State Univer- sity, 1965. (Mimeographed.) ”Methodological Problems of Social Self Theory, " Depart- ment of Sociology, Michigan State University, 1965. (Mimeographed. ) Turner, Ralph J. "Role Taking, Role Standpoint, and Reference Group Behavior, " American Journal of Sociology, Vol. 61 (January, 1956), pp. 316-328. Vernon, G. "Religious Self Identification," Pacific Sociological Review, Vol. 5 (Spring, 1962), pp. 40-43. 246 Videbeck, Richard. "Self Conceptions and the Reactions of Others, " Sociometry, Vol. 23, No. 4 (December, 1960), pp. 351- 359. Waisanen, Fred B. "A Notation Technique for Scalogram Analysis, " Sociological Quarterly, Vol. 1, No. 4 (November, 1960), pp. 245-252. "The Prejudice Variable: A Social Psychological and Methodological Study. " Unpublished Ph.D. dissertation, State University of Iowa, 1954. Watson, John B. Behaviorism. Revised ed. Chicago: University of Chicago Press, 1930. White, Benjamin W., and Saltz, Eli. "Measurement of Reproduci- bility, " The Psychological Bulletin, Vol. 54, No. 2 (March, 1957), pp. 85-86.