THE DEVELOPMENT OF A FUNCTIONAL MESSAGE VARTABLE: THE LOCUS OF CONTROL ' Dissertation for the Degree of Ph. D._ MICHIGAN ESTATE UNIVERSHY VIRGINIA ANNE MCDERMOTT 1974 University This is to certify that the thesis entitled THE DEVELOPMENT OF A FUNCTIONAL MESSAGE VARIABLE: THE LOCUS OF CONTROL presented by Virginia Anne McDermott has been accepted towards fulfillment of the requirements for Ph.D. degree in Communication 0 Major profesiq! Date 47/ / 71/ 74 0-7639 W T m ll T T l ll Tug" ll l T TI 3 1293 ABSTRACT THE DEVELOPMENT OF A FUNCTIONAL MESSAGE VARIABLE: THE LOCUS OF CONTROL BY Virginia Anne McDermott This thesis directed itself to the development and test of a func- tional message variable in a model of persuasive appeals. Four proposi- tions were put forward to justify the delineation of a functional message variable along three loci of control: (1) There exists a subset of human behaviors which may be described as governed by choice, (2) The determi- nation of the criteria for choice in these situations requires symbolic manipulation, (3) The self-concept is one fruitful mechanism for explain- ing criteria formation through symbolic manipulation, and (u) The self- concept suggests the presence of a message variable which distinguishes the locus of responsibility along physical, psychological or social reality. The power of this tripartite differentiation was tested through a study which utilized a multiple regression design with dummy variables to indicate message treatments. When Phase I and Phase II questionnaires were matched, 100 subjects formed the data pool for the analysis. The model hypothesized that a functional message variable which argued for the performance of some act based on physical, personal or social con- siderations which were consistent with the individual's locus of responsi- bility would exert a causal influence on one's self-concept which, in turn, ,ij . . . «3 Virginia Anne McDermott agr3would result in attitude change (contingent upon the initial mass of in- formation of the attitude). The data from the study tend to support such a conclusion. The beta weights for two of the three indices of self—con- cept were statistically significant (p .005). Nonetheless, total variance explained is low (.16). Problems encountered with scaling techniques sug- gest that improvement here is needed, and specific suggestions are made in this regard. With these improvements, a replication of the present study would provide sufficient information either to provide strong sup- port for the model or to suggest an alternative. THE DEVELOPMENT OF A FUNCTIONAL MESSAGE VARIABLE: THE LOCUS OF CONTROL BY Virginia Anne McDermott A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication 197” To my parents, whom I love very much, for always believing I could do it ii ACKNOWLEDGMENTS Thanks are due each of the four who served on my committee for pro— viding intellectual leadership, rigor, honesty and encouragement in the pursuit of this idea. Particular thanks go though to Joe Woelfel, who as committee chairman guided the ship of state, for offering invaluable assistance and advice as I struggled at first attempts at operationalizing "locus of control” -- no easy task when new ground is being broken; and Don Cushman, my strongest ally and severest critic, for his indefatigable pursuit of and demand for excellence. Finally, to the Cushman family - Don, Ellen, Kimberly and Cynthia, whose friendship has meant a great deal during the last three years. iii TABLE OF CONTENTS Chapter Page I THERE EXISTS A SUBSET OF HUMAN BEHAVIORS WHICH MAY BE DESCRIBED AS GOVERNED BY CHOICE ................ I THE DETERMINATION OF THE CRITERIA FOR CHOICE IN THESE SITUATIONS REQUIRES SYMBOLIC MANIPULATION ....... 2 THE SELF-CONCEPT IS ONE FRUITFUL MECHANISM FOR EXAMINING CRITERIA FORMATION THROUGH SYMBOLIC MANIPULATION ....................... . .................. 4 THE SELF-CONCEPT SUGGESTS THE PRESENCE OF A MES- SAGE VARIABLE WHICH DISTINGUISHES THE LOCUS OF RESPONSIBILITY ALONG PHYSICAL, SOCIAL OR PSYCHO- LOGICAL REALITY ............ . .......................... ll II PROCEDURES. ........................................... 17 Message Variable ............ . ...................... l7 Self-Concept..... .................................. 22 Mass of Information ................................ 23 Attitude Change ....... ..... ........................ 23 Design ............ . ................................ 24 III RESULTS ............................................ ... 25 IV DISCUSSION .............................. . ............. 32 APPENDIX ...................................... . ................ 33 BIBLIOGRAPHY ................................................... 58 iv LIST OF TABLES Table Page 1 Locus of Control: Varimax Rotated Factor Matrix After Rotation with Kaiser Normalization ...................... 26 2 Obtained Factor Scores for Locus of Control Items Which, Together with an Information Measure, Constituted the Measure of Self-Concept Employed............. ........... 27 3 The Correlation of Each Index with Attitude Change ...... 28 4 Self-Concept Regressed on Message Variable .............. 30 5 Standardized Partial Regression Coefficients, Multiple Correlation Coefficients, and Coefficients of Determi- nation of Attitude Change Regressed on the Predictor Variable ................ ........ ..... ..... ..... . ........ 3l 6 Change in the Dependent Variable Taking into Account the Locus of Control.... ........ ........ ........... ..... 33 7 Change in the Dependent Variable Taking into Account the Receiver ....... ....... ........... ......... .......... 33 8 Change in the Dependent Variable Taking into Account the Presence or Absence of Physical Barriers to Attending................... ..... .... ................... 3H _The_purpose of this paper is the development of a functional message variable which will be useful in a model of persuasive appeals. To that end four propositions will be put forward which will serve as the organ— izing principle of this discussion: (I) There exists a subset of human behaviors which may be described as governed by choice, (2) The determi~ nation of the criteria for choice in these situations requires symbolic manipulation, (3) The self—concept is one fruitful mechanism for explain- ing criteria formation through symbolic manipulation, and (u) The self- concept suggests the presence of a message variable which distinguishes the locus of responsibility along physical, psychological or social re— ality. THERE EXISTS A SUBSET OF HUMAN BEHAVIORS WHICH MAY BE DESCRIBED AS GOVERNED BY CHOICE A major subtype of teleological explanations are those normally la- belled "actional" explanations. Such explanations assume aiming or in- tentionality on the part of the agent. To ascribe action to a person distinguishes it from other forms of behavior by implying that a certain state of affairs came into existence: the individual intended this state of affairs (or something very close to it); and the person's actions, at least in part, are instrumental in bringing it into existence (Taylor, 1968). Human action may be regarded then as the subset of behaviors pro— duced by the individual which, in some sense, results from conscious ef- forts to reach a particular goal. Taylor's discussion suggests three 1 2 characteristics of human action: human action is organized and patterned; it involves what is subjectively experienced as choice; and it is purposive. While organization with its implications of conditionality and con- straint imposes limitations on all actions, multiple possibilities of available courses of action remain open to the individual. Not only par- ticular organizations of actions but also appropriate parts of the organ- ization are governed by choices. So that the choice itself when made is made subject to the individual's control in terms of his purposes or goals. Communication itself has been studied within this framework —- as a means to an end —- the achievement of some goal. Coorientation models, persuasion models and conflict-resolution models have all reflected this type of conceptualization. This view suggests that the behaving individual is able to recognize a goal state, is capable of receiving information about deviations from the goal state, and has the capacity to redirect itself to the original goal or to set a new goal state. Thus the behaving individual is viewed as a control system which generates goals, and strategies to reach those goals, by means of feedback information that functions to direct or cor- rect action. The person extends control over the environment through the development and use of symbolic feedback which permits the individual to extend concrete reality through symbolic representation. THE DETERMINATION OF THE CRITERIA FOR CHOICE IN THESE SITUATIONS REQUIRES SYMBOLIC MANIPULATION Vygotsky (193H, 1962) directs attention to the function of language as "instrument" through which a constellation of practical skills is in- culcated and eventually operates autonomously, and to the function of lan- guage as "object" -- once skills are acquired they continue to be exercised 3 in abstraction from the words through whose use they were initially taught. Thus, the organization of experience is hierarchical. Certain linguistic activities must be mastered before others can be learned, since what func— tions as instrument at one stage becomes object in the succeeding learning stage. Consideration of symbols as tools or instruments of learning must be coupled with consideration of symbols as the most flexible of objects to which man relates. The ability to combine and organize symbols, the ability to manipu- late symbols, makes this control over the environment possible. Symbolic control permits the individual to array a range of alternatives from which to choose in order to act. This symbolic construing of the environment constitutes each person's definition of the situation. It is the basis for action, and experience documents that people are selectively responsive to different aspects of the natural environment; identical events are seen and interpreted in divergent ways. When diverse interpretations arise this often stems from the fact that key objects, though represented by the same symbols, assume different meanings for different peOple. The nature of symbols themselves contributes to these differences. On the one hand, symbols are flexible: complex combinations whose mean— ing exceeds the mere sum of the individual symbols are possible. Much of communication behavior is thus situation specific. Yet symbols ex- hibit stability: they recur over time making shared meanings possible. The paradox of symbol systems makes possible the emergence of multiple perspectives while directing attention to the dynamics of specific sit- uations which cue the communication choices of interactants. That com- munication rests on the interaction of sources, messages and receivers u carries clear implications for consideration of the way in which partic- ipants construe environments or define situations. If the complex interrelationships of symbol flexibility and stability achieve significance in the context of specific and general patterns, then in differing patterns the significance of the symbolic units may change. Thus, any explanation of human action requires an understanding of the way the individual views the self in relation to those objects deemed relevant to a particular action. To understand what the individual's design of the environment is requires information about what he or she brings to the situation and what each extracts from it. THE SELF-CONCEPT IS ONE FRUITFUL MECHANISM FOR EXAMINING CRITERIA FORMATION THROUGH SYMBOLIC MANIPULATION Both psychologists and sociologists have suggested that the self- concept is a (if not the) principal determinant of conscious human be- havior which serves as a mediating variable in any explanation of dif— ferential human behavior within a specified situation. Despite the contributions of seminal minds like William James and George Herbert Mead.research into the self-concept has floundered. Perhaps because the early influencers argued from within a philosophic tradition, sys- tematic empirical development has been slow in coming. A brief review of the literature follows: within psychology only the phenomenalists have been examined; within sociology, only the symbolic interactionists. Three weaknesses common to both traditions are: (1) poor conceptualiza- tions, (2) faulty measuring techniques, and (3) imprecise specification of relationships. In reviewing the literature both psychologists and sociologists will be handled together; most of them can rightfully be considered social 5 psychologists. And following a discussion of each apparent weakness, a conceptualization which appears to overcome each deficiency will be put forward. (1) Poor conceptualization has plagued the self-concept literature. Definitions have been either too vague or too selective to have much util- ity. Wylie's (1961) useful review points out that the analysis of the self and its functioning has been "woefully simplistic." Allport (1943) lists several self—functions which have appeared in the literature: the ego as "behavioral system," "knower," "primitive selfishness," and "fighter for ends," but as Pepitone (1968) comments, "Although descriptively apt, such self-functions lack the operational specification that would be nec- essary to employ them in research." Shibutani (1961) variously describes the self—concept as a way of behaving and also as what a man means to himself. Curtis (1960) rules out all cognitions and admits only "as- pects of the person to which he reacts." In an effort to reduce confusion, some researchers have taken to analyzing dimensions of the self. Here single hypothesized functions of the self carry the explanatory burden for resultant behavior. Rather than improving conditions the move has only served to blur distinctions further and to glut the area still more. Research into self—esteem (Mann, 1959; Rosenberg, 1965), self—actualization (Maslow, 195a), self-blame (Doris and Sarason, 1955) and self-acceptance (Crowne and Stephens, 1961) has all grown from this perspective. While confusion surrounds what it is, even less attention has been paid to what the self—concept does. Gerth and Mills (1953) and Kinch (1963) made vague stabs in this direction but only Kuhn (1960) brought any precision to the effort. Concerning the role and function of the 6 self—concept he offers, "one behaves in terms of the kind of person he thinks he is, and for the ends such a person seeks." So, while much material on the self—concept and its related dimen- sions exists, the literature itself lacks conceptual clarity. Recently, however, a View of the self-concept which builds on Mead and is consistent with his view that the self is a symbolic construct has been proposed by Woelfel (1968) and modified and extended by Cushman and Whiting (1972). It is important to communication researchers for three reasons: it rec- ognizes the self as the most central and focal object within the symbolic environment; acknowledges that objects, including the self, are symbols whose situational and nonsituational components must be captured; and focuses on the relationship between information and behavior. In referring to man as having a self, George Herbert Mead implied that each individual can act socially towards the self. The individual may praise, blame or react to his or her own gestures. Thus it is pos- sible to become the object of one's own actions. By taking the role of others the individual is able to come to see one's self. The importance of language to this process cannot be overemphasized, for it is through language or significant symbols that the meanings and definitions of those around one are acquired. By learning the symbols of his groups the in- dividual comes to internalize the definitions of events or things, in- cluding definition of one's own conduct. The identification of one's self, as well as the knowledge about other objects of one's experience involves the establishment of relation- ships between the self and other objects. In fact, this set of relation— ships among objects and between the self and objects constitutes the total information a person possesses about reality, including one's self. 7 From out of this perspective, Cushman and Whiting (1972) propose that the self-concept be regarded as "an organized set of rules which define the relationships of objects to individuals and which are capable of gov- erning and directing human action.” Thus the self-concept becomes the composite of all the rules an individual has regarding the relationship of objects to self; as such the self-concept or rules of relationship provide the transformation of information into behavior. (2) Although operational specification has not been clear from the conceptualizations, many research instruments purporting to measure the self-concept have been spawned. Wylie (1961) and Shaver (1969, 1972) point out that the development of most instruments has been a one—shot affair; few replications have occurred which might seek to establish con- struct validity and reliability. Three problems especially plague mea- surement techniques: (1) social desirability of the response, (2) the relationship between self—disclosure and content areas, and (3) response restriction. Many self-concept instruments measure self-ideal self disparities. These instruments are frequently correlated about equally well with so- cially desirable responses as they are with each other (Crowne, Stephens and Kelly, 1961; Edwards, 1957; Ziller, 1973). Statements like "I am cool" (Strodbeck, et_al:, 1962) are assessed, then the general measure used to predict specific acts within limited situations (Rosengren, 1960; Miyamoto and Dornbusch, 1956). The relationship between self-disclosure and certain content areas is also thought to affect the precision of measuring instruments. Wylie (1961) has proposed that it may be more socially acceptable to reveal content in certain areas than in others or that areas of item content 8 may be differentially revealed because they are more or less salient to the respondent. She cites the research of Jouard and Lasakow (1960) which shows the impact of varying factors on self—disclosure in support of her position. Lastly, there have been problems in the area of response restriction. The composition of many measuring instruments depends entirely on researcher preconceptions about the phenomenal field. Yet Jones (1956) found a U— shaped curve when subjects were given a free-choice Q—sort setting sug— gesting that respondents show significant individual differences in dis- tribution form (lost, of course, when normal distributions are forced). And Levonian, gt 31. (1959) factor analyzed a self-concept inventory and found no large factors identifiable along the lines of the variables pre- sumably tapped by the test, supporting the theory that an incongruent field may be being proposed to represent the phenomenal field. The one instru- ment which appears to have overcome this problem is the Twenty Statements Test (Kuhn, 1960) which makes an important move in capturing the relational nature of self—object relationships. Woelfel develops this perspective still further. His measuring in- strument works from the fundamental premise that all human knowledge is essentially relational. Thus, to define an object, requires specifying relationships between that object and other objects. Fundamentally, this involves taking notes of similarities and differences between objects. Woelfel argues that "the continuous set of positive real numbers offers a potentially error-free language for the definition of any set of social objects" (Woelfel, 1971). A dissimilarities matrix formed from paired comparisons which permit the individual to set his own anchoring point is used as a tool for the description of any part of the self—conception, 9 such as an attitude, belief, set of beliefs, etc. The matrix S represents the static structure of the interrelationships among the set of N objects at any instant in time, and the processual character of these relation— ships can be captured in successive matrices (For fuller discussion cf. Woelfel, 1971). While in the past faulty measuring instruments have raised serious questions about what is really being captured, the work of Woelfel sug- gests that a powerful measure of self-object relationships exists which takes account of both the situational and nonsituational elements of objects. (3) Finally, imprecise specification of relationships has charac- terized the literature. The knowledge that two variables are related is of little value; to be meaningful, precise specification of relationships is needed to explain not only how changes in one variable affect another but how change in that relationship alters other relationships within the system. Yet phrases like "the self-concept is important to any consid- eration of behavior" or "the self—concept leads to behavior" dot the lit— erature. Sarbin (1962) labels the self both a series of "cognitive sub- structures" and "empirical selves," while Hilgard (1949) proposes that the construct utility of the "inferred self" rests on three hypotheses, all nontestable. 0n the other hand, Woelfel's theory is straightforward: information determines behavior. The relationship between variables is clearly spec- ified: irreversible, deterministic, sequential, sufficient, and necessary. (For discussion of specification of relationships, cf. Zetterberg, 1965). Information can be gleaned two ways: inductively and deductively. 0f the two, the more significant and powerful means of rule formation 10 is deductive. Norms, roles and positions provide an individual with a repertoire of relationships from which self-conceptions can be deduced. But whether the rule is formed inductively or deductively it is done by association and differentiation —— by categorization. An actor's rule for action contains (1) a definition of an object in terms of its rela- tionship to the individual and (2) an indication of the circumstances in which the rule is applicable. Rules and categories are therefore isomor- phic. Not only can one make inferences about his or her relationships to objects placed in the same category, but the rule provides expectations about the nature of those objects believed to be subsumed under the same rule. And, as rules develop, they form a system of rules, which provide the actor with preconceived plans of action -- a format for processing experience and initiating action: to make sense of the past, to act in the present, to cope with the future. When the actor reflects on, and using past experience, defines the situation and its relationships to the self, the action to be taken toward that situation or object is determinate. By highlighting three critical weaknesses in previous self—concept formations, and suggesting a paradigm which is corrective, we have sug- gested that actors, searching out definition, author the particulars of any cognitive-behavioral relationship. It is rules which provide the powerful tool for analyzing the relationships among elements, their log- ics and consistencies, and the implications of transforming them into other means-ends relationships. They are criteria for choice, and hence, action. "Only criteria can signify reality and criteria are variable" (McHugh, 1966). And, it is the self-concept which has been posited as a fruitful mechanism for explaining criteria formation through symbolic manipulation. ll THE SELF-CONCEPT SUGGESTS THE PRESENCE OF A MESSAGE VARIABLE WHICH DISTINGUISHES THE LOCUS OF RESPONSIBILITY ALONG PHYSICAL, SOCIAL OR PSYCHOLOGICAL REALITY When confronted by a situation initially, the actor must do three things: determine what the situation is; what alternatives are present; and what action is appropriate for the self in that situation. Like Mead, this view holds that an act encompasses a complete span of attention with its initial point as an impulse and its termination in some objective which completes the impulse. The actor is in the process of constructing and reorganizing plans for action on the basis of cues which may be physical, psychological or social. The View of the self advanced here argues for "the inseparable con- nection between rational cognition and rational purpose" (Pierce, 1969). Both James (1951) and Mead (1938) argued similarly that purpose organizes reality. James put it this way: The only meaning of thought is teleological...classi- fication and conception are purely teleological weap- ons of the mind. The essence of a thing is that one of its properties is so important for my interests that in comparison with it I may neglect the rest. While Mead noted: In the end, what we see, hear, feel, taste and smell depends upon what we are doing, and not the reverse. In our purposively organized life we inevitably come back upon previous conduct as the determining condi— tion of what we sense at any moment, and the so-called external stimulus is the occasion for this and not its cause. If this is so, then elements which are interrelated in any single situation are "fixed" only within that situation. Although an adult self- concept is relatively stable (Wylie, 1961; Brownfain, 1952; Erikson, 1959), shifts in self-object relationships should be expected to occur regularly. If multiple realities are born of one's perspective or location within 12 the set, then what must be tapped are the rules or criteria that individ— uals use to make sense out of what surrounds them. While multiple rela- tionships of information to behavior are possible, information itself comes only inductively or deductively (read, of course, against a back- drop of previous beliefs, attitudes and values). This suggests the following tripartite differentiation: When information comes deductively prepatterned goals and means can be transmitted through socialization. The research of Azjen and Fishbein (1973) supports the contention that in certain circumstances the normative component carries significant weight in the prediction of behavior. Ear- lier research by Newcomb and Charters (1958) has demonstrated that re- minding subjects of membership in salient reference groups produces be- haviors more in keeping with the perceived attitudes of the reference group. Information which an individual already has about an object or be- havior can be symbolically manipulated in key ways. Here the individual chooses the preferred behavior based on personal criteria. Schwartz (1968) and Schwartz and Tessler (1973) have suggested that personal normative beliefs make consistent, independent and substantial contri- butions to the explanation of variance in behavioral intentions. On occasion individuals run into overwhelming physical contingencies whose presence dictates means-ends relationships. If I have enough money to buy an ice cream cone but not enough to go to a movie, I may well set- tle for the ice cream. Although conceptualized along somewhat different lines, this differ- entiation is not without support. Levenson (1972) reconceptualized Rotter's (1966) Internal—External scale along three dimensions in an effort to 13 separate out the influence of powerful others from chance expectancies. And, Miller and Steinberg (1974) detail interpersonal relationships in terms of prediction and control and cite a series of examples which sug— gest that the intention to perform a task may reside in physical, social or personal choice patterns. Heider (1966) too describes how individuals both grasp reality and attempt to predict and control it by referring the transient event to more stable dispositional properties. He includes an interpretative step in which the individual asks: "What is the immediate source of x? Is it chance? Am I the cause of it? Or is another person, 2, the cause?" Rotter's conceptualization of the locus of control focuses on iden- tifying cognitive styles. Where one perceives the expectancies for con- trol constitutes one's cognitive approach for looking at the world which, in turn, leads to the arousal of diverse percepts even when stimulus con- figurations are identical. However, what this does is assess a global estimate of how individuals lggk_at the world and from that assessment attempt to predict behaviors which involve ways of being_in the world (in quite specific situations). Heider develops an intentional perspective more in keeping with the focus of this paper. His discussion centers around the assumption that there are basic constituents to an action sequence from which it is pos- sible to impute the intent (or ability) of another person to perform some task. Although he makes clear that what is said about imputing the in- tent to "the other" applies equally well to one's own actions, this per- spective is not pursued. What is crucial to the perspectives of Rotter and Heider, however, is that they (1) establish the individual as the center of prediction 11+ and control; he is not a mere passive agent, (2) do not infer the locus of control from task constraints but attempt to assess the perceived rela- tionship as elicited from the respondent, and (3) suggest there are non- situational criteria against which situational dimensions are perceived. This tripartite differentiation suggests the presence of a powerful message variable for determining choice based on information loci found in physical, social and psychological patterns. So far this paper has suggested that (1) there exists a relevant subset of human behaviors which may be described as governed by choice, (2) the determination of the criteria for choice in these situations re— quires symbolic manipulation, (3) the self-concept is a formal structure which contains criteria for choice, and (4) positing the self-concept in this context suggests the presence of a message variable which distin— guishes the locus of responsibility along physical, psychological or social reality. The framework laid out thus far has clear implications for persua- sive appeals. The task of persuasion becomes descriptive in this view. The message must describe the listener and the object of the proposed action in complementary terms. In so doing it becomes necessary to view the object of the action in terms of the audiences' self—conceptions and then to select such symbols as are necessary to make the audience aware of their relationship to that object (Woelfel and Cushman, 1968). Mes- sages can function to adjust men and their self—conceptions to objects for some predetermined reason by appeal to a choice premise which argues that the recommended behavior is consistent with the individual's per- ceptions of the locus of control for behavior. To trigger an actor's rules of relationship messages may (1) focus on can (personal choice), 15 (2) demonstrate should (social reality), or (3) encourage trying (physical reality). When confronted with choice situations where the mass of information is manipulable, the following relationships in a model of persuasive ap- peal are proposed: The relationship between the locus of responsibility message variable and self-concept is assumed to be irreversible, deter- ministic, sequential, sufficient and necessary. The relationship between the locus of responsibility message variable and attitude change is assumed to be irreversible, stochastic, sequential, contingent (upon the mass of pheno— menal relevant reality out of which the original attitude was formed and is maintained) and necessary. The relationship between the self—concept and attitude change is assumed to be irreversible, deterministic, sequential, suf— ficient and necessary. The relationship between self-concept and behavior is assumed to be irreversible, deterministic, sequential, sufficient and necessary.1 The relationship between attitude change and behavior is as- sumed to be irreversible, stochastic, sequential, contingent, (upon the mass of phenomenal relevant reality out of which the original attitude was formed and is maintained) and necessary. Diagrammatically, the influence of a persuasive appeal may be represented: X2 X1\ / XL: / \\ / III “t- \/ r” If, \ fixs/ 1This is true for any single time interval. Over time self—reflexive activity does impact the self—concept. 16 locus of responsibility message variable at one of three levels: physical, social or personal where X1 X2 = self-concept X3 = attitude change X4 = recommended behavior _-______--_____-_____——-—-----3>= Contingent causal path >= Causal path As an initial step to validating this model, the remainder of this paper focuses on a test of the model's first stage with a View to tapping the behavioral orientation of the attitude at the highest possible level of scaling so that the message is associated explicitly with an overt behavior. Pragmatic concerns underscore this decision. As the recent review of the attitude—behavior literature by Cushman and Seibold (197”) points out, there are multiple and complex reasons for the low correlations be— tween attitudes and behaviors. 0n the one hand, a series of problems exist which are connected with the relationship between verbal and non— verbal behaviors; but, beyond that, the effect of message manipulations on both verbal and nonverbal behaviors remains unclear. If the research design proposed here brings increased Operational and conceptual clarity to the relationship between messages and verbal reports of attitudes by successfully manipulating the locus of responsibility and by providing some evidence that the tripartite distinction is valid, a significant step forward will have been taken. Chapter II PROCEDURES The focus of the study itself was the design of message variables which tap each of the three loci of control: physical, psychological, and social. The remainder of this chapter will provide conceptual and operational definitions for (1) the functional message variable, (2) self-concept, and (3) attitude change. Message Variable At the conceptual level, the tripartite differentiation for the locus of control requires separate definitions. Operationally, separate messages were also constructed. However, in each instance, messages recommended attendance at a public lecture on family communication to be delivered by a faculty member from the Department of Communication at Michigan State University. The locus of control defines the situation for an individual by pro- viding the criteria for choice in that situation. As developed here, the locus of control variable posits that information grounded in three sep- arate, independent sources is available to the individual concerning one person's relationship to an object; and that for any given situation, the locus of control will predominate, causing the organization of relevant ob- jects which then constitutes that individual's definition of the situation. When the locus of control is perceived in social reality, information is derived from the normative force of prepatterned goals and means. Here, 17 18 others' expectations exert influence determining what the individual should do. Since the communicated expectations regarding family membership con— stitute a reasonably clear and persistent repertoire of relationships, and may well be the strongest available social pattern, a message grounded in social reality must make that social pattern salient. An operation consistent with this conceptual definition requires lo- cating and defining the significant others for the topic for each subject. Significant others by virtue of their location within the network and/or communication choices by the individual define the self's relationship to objects. Consequently, data regarding the mean value of incoming infor- mation as well as consensus levels within the network are essential deter— minants of what should constitute message components. From this one could derive the place of the family in the communication network, separate out the influences of parents—as—family versus a marital partner, determine the subject's centrality within the family network and the extent to which that individual's role is one of communication maintenance. Translating the expectations of others into a single message which brings the pres— sure of communicated expectations to bear on a single self-object rela— tionship poses real difficulties for mass appeals. The operation proposed does not tackle the problem head-on. However, lacking specific information on each individual, the message appeal con- centrated on making salient the role_of the individual in improving fam- ily communication, the importance of good family relationships in society today, and, by crediting the lecturer a knowledgeable and effective speaker, attempted to make of him a "significant other." Operationally, the message focused on the simplicity and stability of family life in times past and recommended that continuity in the present; 19 linked the audience with other college students concerned for the quality of family life; and concluded by contrasting the potential for family life with the dreary norm projected by the recent TV series which focused on the Loud family, An_American Family. Specific phrases included were "clear recognition of the need to foster...relationships within families"; "from Carl Rogers to Margaret Mead prominent voices have observed that...."; "fragmentation of contemporary society"; and, "if for you family life seems important to human life..." There is no doubt that the message failed to bring the communicated expectations of significant others to bear. In no way did it focus atten— tion on those important others who define self-object relationships for the individual. However, while lacking the normative power of significant other's expectations, the message did reflect a set of culturally shared expectations about family life. The attempt to make the lecturer a "sig- nificant other" for the audience was not a particularly useful move. Again, calling upon a national figure whose credibility has been culturally estab— lished would more closely approximate the criteria of being a significant other. Finally, the focus on "role” of a family member was blurred; dif- ficult to distinguish from the dimensions of personal locus; and without the pressure which shared expectations regarding a role are capable of bringing about. When the locus of control is perceived in personal reality, the in- dividual manipulates information already amassed (either by self—reflexive activity or through previous communications) in key ways to determine choices. The focus is on the controlling role of information in one's personal life. A message grounded in personal reality must stress the nature of control over interpersonal relationships which new information would make possible. 20 An operation consistent with this conceptual definition would define the individual's self-expectations by casting the appeal so as to take into account the particular hierarchy of values, beliefs and attitudes which triggers actions for that person. The information level of the individ- ual regarding the topic should be high since this locus argues that the individual now has sufficient information about the relationship of the self to the object that action is determinate. Knowing that much of the incoming information an individual receives comes from others, problems in parsing out personal and social influences were anticipated and attempts made to make them operationally independent. To this end, the message appeal for personal locus focused on making sa- lient one's personal need for information about family communication, the tools necessary to c0pe with family problems and the amount of information needed to function as a family member. Operationally, control of one's personal environment was linked with the ability to shape interpersonal relationships, and the lecture was de- picted as a rare opportunity within one's college career to glean infor- mation which would be useful. Specific phrases included: "no task more significant than obtaining sufficient information...to control one's fate"; "few opportunities ... to learn a set of principles which are relevant every day of your life"; "if...maintaining control over your personal environment is your goal...." However, no information regarding the value system of the individual was gathered. Thus, while the message did try to make the necessary ef- fort to establish a belief that the object of action stood in relation to a valued goal, it did not demonstrate the truth of this connection within the overarching value system of the individual. 21 When the locus of control is perceived in physical reality, infor— mation about environmental resources or physical contingencies which have been communicated to the individual determine choice by dictating means- end relationships. Under these circumstances the individual feels the environment is forcing choices, so the focus of a message grounded in physical reality must demonstrate that attending this lecture was an event devoid of physical constraints and, as such, worthy of note. In fact, two kinds of situational constraints operate. When the action recommended is directed toward a low mass object, physical con- straints must be overcome which inhibit action. If a lecture is sched- uled and it is raining, a low mass topic is likely to be abandoned in favor of staying warm and dry. Physical constraints of this variety are amenable to persuasive appeals which recommend efforts to overcome the constraints or demonstrate that the power of the event renders it void of all such constraints. Information should be gathered which pro- vides the researcher with direction concerning the nature of perceived situational constraints. The second kind of constraint, the exam which conflicts With the sched- uled lecture or the inflexible work schedule poses a problem which no single message appeal can overcome - the need to reorder priorities. Care should be taken to isolate those individuals from analysis who fall into this category. For, regardless of the attitude change they may demonstrate, they are unlikely candidates to carry out the recom- mended behavior. However, for this design this separation was not maintained. Opera— tionally, the lecture was hailed as an event which "merits our attention" as "an opportunity (so) devoid of physical constraints." 'The time is 22 right." "The place is right." "The situation is right." Finally, the audience was urged to join the "winner's circle", and not the loser "who persistently places himself in situations where the physical odds against success are overwhelming." In addition to the specific appeal, each message contained a state- ment of Professor Cushman's credentials both as a lecturer and as an ex— pert in interpersonal communication; a paragraph which described three specific behaviors which would be affected by the lecture; and concluded by recommending attendance at the lecture in a paragraph which recapped (for each message) the particular arguments which should serve as a basis for attending. Lastly, nine items which were thought to measure the three loci of control were constructed and factor analyzed from Time 1 data. Although these items were a part of the development of an operational index for the self—concept (and are more thoroughly treated in the following dis- cussion), the items were expected to factor along the three dimensions outlined. By comparing Time 1 and Time 2 data some insight into the stability of the factors is also available. Self—Concept The self—concept was previously defined as an organized set of rules which define the relationships of objects to individuals and which are capable of governing and directing human action. Since the self-concept is conceived of as a mechanism for criteria formation, the three loci of control are the dimensions along which the self—concept should vary. The procedure for constructing an operational index for the self- concept was threefold. First, nine items that were hypothesized to mea- sure the three loci of control were constructed and subjects were asked 23 to evaluate each dimension along a seven—point scale for its importance in deciding whether to attend the family communication lecture. In addi- tion, each of these nine items was followed by a thirteen—point scale item which asked subjects how much time they had spent in the last month talk- ing to others about the topic item under consideration (and just previously evaluated). Lastly, the data from the eighteen scales were factor analyzed and, based on the factor analyses, factor scores were estimated for each of the three loci of control to construct indices. Change scores for each index were obtained, as was a measure of the amount of information of each individual at Time 1. Mass 9f_Information Information was gathered on the knowledge individuals felt they had to relevant objects within the act along a seven-point scale which went from "know very little" to "know extremely well." Four items were in- cluded: Public Lectures, Department of Communication, Professor Cushman and Family Communication. An index was formed by transforming the re- sponse to each item into a standardized score (Z) and summing across the four items to achieve a single score. Attitude Change Consistent with the definition of self-concept, an attitude is de- fined as a rule concerning an object or course of action predisposing an individual to respond in some preferential manner. A pretest measure was obtained to evaluate the individual's attitude toward the act (attending a lecture on family communication delivered by Professor Cushman of the Department of Communication at Michigan State University) by asking for 24 judgments on three sets of polar adjectives measured along a seven-point scale. The same measure was repeated subsequent to the message manipula— tion and sums were obtained for both data sets. Attitude change was op- erationalized as the change score between post and pre-measures. Design The study was in the form of a multiple regression design utilizing dummy variables to indicate message treatments. Subjects for Phase I of the study were taken from five undergraduate classes in the Department of Communication at Michigan State University. During the period between May 6 and 8, 1974, 165 students completed the Phase I questionnaire. During the period between May 16 and May 21, 1974, the students in the same five classes were asked to participate in Phase II of the study. Subjects were randomly assigned to treatments (one of three message mani- pulations or a no-message condition) by randomly distributing the instru— ment package to all students present. In all, 130 students completed Phase II of the study. It was possible to match questionnaires for 100 students. These matched questionnaires constitute the data set for the present analysis. Chapter III RESULTS Before reporting the results of the test of the model, some explana- tion concerning the locus of control dimensions seems in order. Three dimensions were delineated and nine items hypothesized to capture the three dimensions. By factor analyzing the Time 1 data some insight into the power of the tripartite differentiation both as a reflection of real— ity and as a message variable will be made available. Table 1 contains the nine items and the dimension each item was expected to reflect. Table 2 contains a listing of the items and obtained factors. Factor scores for Time 1 and Time 2 data are also shown. While the obtained factors do not provide unequivocal support for the development of a message variable with a tripartite differentiation along the locus of control, two of the three situational items factor together, and two of the three personal and social items respectively factor together. The difficulty in differentiating clearly between personal and social dimensions was anticipated; the Azjen and Fishbein (1971) and Schwartz and Tessler (1973) works point to that problem. Certainly, more work is necessary if personal and social influences are to be clearly delineated. Two things are encouraging, however. First, the items did factor along three dimensions. The two factors whose com— position most closely reflects the hypothesized dimensions (Factors 1 and 3) show the highest correlations with the dependent variable. Two, the 25 26 factors remained stable between Time 1 and Time 2. Only one shift occurred; the one item from personal reality which factored on Factor 2 at Time 1, joined the other two personal reality and two social reality items on Fac— tor 3 at Time 2. Table l: Locus of Control: Varimax Rotated Factor Matrix After Rotation with Kaiser Normalization. Situational FACTOR l FACTOR 2 FACTOR 3 how much time and effort required to attend -.243 .065 .808 the lecture being the best source of infor- mation available .193 .720 .154 the date, time or place of the scheduled lecture .236 .081 .783 Personal what I need to know about family communication .865 .182 .001 the tools I need to cope with family problems .838 —.008 .142 how much information it takes to be a family member .550 .380 -.140 Social the importance of the individual role in im- proving family communication .803 .134 -.015 what the speaker's qualifications on the subject are .133 .891 .037 the importance of good family relationships in society today ,, T .666 .350 -.053 27 Table 2: Obtained Factor Scores for Locus of Control Items Which, To- gether with an Information Measure, Constituted the Measure of Self-Concept Employed. Time 1 Time 2 Factors Factor Scores Factor Scores Factor 1 how much time and effort required to attend .603 .650 the date, time or place of the scheduled lecture .596 .565 Factor 2 the lecture being the best source of information available .486 .560 how much information it takes to be a family member .170 **** what the speaker's qualifications on the subject are .643 .553 Factor 3 what I need to know about family communication .309 .267 the tools I need to cope with family problems .348 .324 the importance of the individual role in improving family communication .295 .282 the importance of good family relation- ships in society today .186 .236 **** This was the only item to shift factors; its factor score at Time 2 on Factor 3 is .197. 28 In order to build indices for each factor, raw scores were standard- ized for each item for Time 1 and Time 2 data. Then the simple product of the Factor Score and standardized raw score was obtained. Item by item change scores were obtained. Change scores were then summed to create three indices which corresponded to the factor dimensions obtained. Table 3: The Correlation of Each Index with Attitude Change. time and effort required to attend + the date, time or place of the scheduled lecture .30 the lecture being the best source of information available + what the speaker's qualifications on the subject are .03 what I need to know about family communication + the tools I need to cope with family problems + how much information it takes to be a family mem- ber + the importance of the individual role in improving family communication + the importance of good family relationships in society today .26 In addition, two other indices were developed to facilitate a test of the model. As demonstrated above, indices were constructed which corre- sponded to the factors on which each locus of control item had fallen. The implication of the model is that each level of the message will af- fect the corresponding dimension of the self-concept. However, due to problems with cell size, attempts to measure the differential effects adequately were not successful. Rather than violate methodological as- sumptions, an alternate procedure was devised. The message levels were collapsed to yield a dichotomy: message / no message. Then the inde- pendent factor score sums for each index were also collapsed to yield a simple measure across nine items of the change from Time 1 to Time 2. The index constructed was: Information Tl (Self-ConcepE‘ — Se1f~ConceptTl). 2 29 A relationship within the model also dictated the development of an index which would transform a contingent and stochastic relationship in such a way as to facilitate the assumption of a sufficient, deterministic relationship. Since the model argues that the impact of the message vari- able upon an attitude is contingent upon the mass of information out of which the attitude was originally formed and is maintained, a message in- dex which takes the relationship into account was created. By again di— chotomizing the message variable and utilizing the simple product of the message variable and mass-of—information index, an index was created which transformed the impact of the message variable into a value contingent upon 1 the observed values of the mass of information. Given these indices the model delineated in Chapter II hypothesizes that: (1) N3 = Blel + A (2) £4 = B42'X'2a + B42b 'X' 2b + 842C 'X' 2c + 841 'X' l where: X1 = message variable X'l = message index X2a = locus of control index (factor 1) X2b = locus of control index (factor 2) X2c = locus of control index (factor 3) 1The theory suggests that the amount of change in the attitude is in- versely proportional to the amount of information about the attitude objects which the individual has. The Optimum measure to capture this relationship would be to take the dichotomous message variable over the mass of informa- tion index since the presence or absence of the message variable represents the force advocating change. However, the mass of information is not ex- pected to take on negative values, but standardizing scores resulted in both positive and negative values in the index. Now, rather than the ex- pected small positive number in the denominator when the mass of informa- tion is low, a large negative number represents low mass. Therefore, the procedure followed for indexing is appropriate given these constraints. 30 x3 X4 Self-concept index Attitude Change In terms of equation (1) Table 4 indicates that the beta weight for the message variable is statistically different from zero (F = 4.65, d.f. = 1.98, p <.05). For equation (2) the model posits a hierarchy of relationships (i.e., deterministic relationship assumed to be stronger than the contingent re- lationship) which dictates the order of inclusion. This is a point worth noting as the order of inclusion is directly related to the size of the increment in R2. Therefore, in testing the model, the locus of control indices were included prior to the message index. The R2 for the locus of control indices alone is .14 which is sta- tistically significant (p<.005, F = 4.48, d.f. = 3.96). The addition of the message index increments R2 by .03. This increment is not statisti- cally significant (a = .05, F = 2.83, d.f. = 4.95). However, the R2 is statistically significant. Table 4: Self-Concept Regressed on Message Variable (n=100). Predictor Variable Regression 1 X1 (Message) 8 = .22* R .22 R2 05 * denotes p‘<.05. 31 Table 5: Standardized Partial Regression Coefficients, Multiple Correla- tion Coefficients, and Coefficients of Determination of Attitude Change Regressed on the Predictor Variable (n=100). Predictor Regression Regression Variables l 2 X2a (factor 1) B = .23* B = .24* x2b (factor 2) B = -.10 B = —.10 X2C (factor 3) B = .19* 8 : .21e X'l (message index) 8 = .16 Multiple R .37** .40** Multiple R2 .14 .16 * p<.05. ** p<.005. Chapter IV DISCUSSION Although the test of the model is generally encouraging, the power of the message variable is disappointing. Several factors in this regard are worth noting, however. While the use of an information index to create a message index was expected to enhance the predictive power of the message variable (since it took into account the mass of information which a sin- gle message would be expected to overcome in order to change the attitude), problems in measuring and constructing this index (reported in Chapter III) resulted in the message index reducing the strength of the message variable. When attitude change was regressed against the message variable, the re- sulting F score was just shy of statistical significance (p = .06, F = 3.63, d.f. = 1.98). In addition, an adequate test of the power of the message variable should include giving the persuasive message which is consistent with an individual's perceived locus of control. While that information was gath- ered for this study, attempts to preserve the anonymity of respondents as well as considerations of size of initial N, dictated against this proce- dure. However, some evidence to support this position is available. As Table 6 shows, when the perceived locus of control is situational or social and a message which is consistent with the perceived locus is received, the mean change is higher than that of the message variable alone. The failure of the personal locus to demonstrate this trend may reflect the failure of the locus variable to delineate clearly between personal and 32 33 social reality or it may simply reflect the fact that in this condition only two subjects received messages which were consistent with the per- ceived locus. Table 6: Change in the Dependent Variable Taking into Account the Initial Locus of Control. Mean Change when Mes- Mean Change Mean Change sage is Consistent Message/No Message by_Message with Perceived Locus .64 Situational 1.30 2.08 Personal .888 .500 Social .550 3.20 Two other message-related findings bear comment. While women show greater overall change between Time 1 and Time 2, men are more persuaded by the message. As the table below indicates, men who receive a message demonstrate more change than women, and this difference becomes more marked when the direction and magnitude of change in the control groups is taken into account. Table 7: Change in the Dependent Variable Taking into Account the Sex of the Receiver. Mean Change Message Control Males .635 1.10 —.917 Females .646 .758 .400 It was expected that differences in message effectiveness would exist between students who had some compelling activity which prevented attend- ance at either of the two lectures and those who were not constrained by 34 the physical situation. This information was gathered from students be- fore exposing them to the message manipulation. An unexpected finding was that these students who had a reason for ngt_attending showed greater change (.848) than those who indicated no physical constraints (.463). Those who said they could not attend and received the situational mes— sage variable showed a mean change of 2.29; for those who could attend the mean change was .947. Even the control groups reflect this anamoly. The respective control group means are .533 and —l.083. While the mean- ing of these figures are unclear, it may be a suggestive piece of infor- mation for sorting out the message-attitude—behavior relationship. Table 8 lays out the relationship. Table 8: Change in the Dependent Variable Taking into Account the Presence or Absence of Physical Barriers to Attending. Situational Mes- Control Overall Change sage Variable Group Reason for not attending .848 2.29 .533 No reason not to attend .463 .947 -l.083 Certainly these research results suggest the need to improve the pre— cision and power of the message variable. In part this will rest in more clearly delineating the tripartite differentiation of locus of control; the influences of personal and social responsibility must be parsed out. Conceptually the definitions are reasonably clear; as Chapter II indicates operational definitions fall short. Separate experiments which focus on a single message appeal with a large enough N to permit adequate testing of relationships are needed. In addition, the research efforts should in- clude both experimental and survey designs since the loci of control 35 antecedents do not appear to be uniformly amenable to one design. Finally, while subsequent research may parse out the independent contributions of the loci, the question of additive effects, which has not been explored, should be given attention. Also, once three stable factors have been isolated, the move to a more powerful and precise measure of self—concept, one which more closely reflects the position recommended by Woelfel (1971), should be made. Lastly, better account of the mass of locus of control is needed. Specific hypotheses must be derived which include predicted differential effects: i.e., high initial mass and a message consistent with that posi- tion will result in low locus change but high attitude change; while for an initially low locus of control, a message along any locus should be able to trigger change both in locus and in attitude. Some limited support for this position exists. When the self-concept index is grouped into high, medium and low calls, the mean change in the dependent variable for each group is: 1.583; .236; 1.750. Only further research can settle whether the mass of the perceived locus in conjunc— tion with the message variable causes this pattern of means. Three considerations about the mass of information deserve attention. If the act or object toward which an attitude is directed is itself low mass, the definition of the situation will most likely be in physical reality. Sixty-two per cent of the subjects cited the physical locus when asked to name a predominant pattern which would define their rela- tionship to the act. This raises the serious question whether a single persuasive message can overcome overwhelming physical constraints; even if the message results in a change in attitude, it may take several mes- sages to make the act or object a high mass item toward which some be- havior is likely. 36 Secondly, the very conceptualization of the locus of control in per— sonal reality implies high mass information. That is, it suggests that the individual processes existing information in new ways to determine a relationship to the act or object in question. This may well imply both a high mass attitude and a high mass locus. Finally, while the loci of control and attitude mass may need to be reasonably high to trigger the model, the information out of which the original attitude is formed and maintained must not be so high as not to be manipulable. Thus, further considerations of information mass, as plateau and as trigger, are essential. The previously delineated model hypothesized that a functional mes— sage variable would exert a causal influence on one's self-concept which, in turn, would result in attitude change (contingent upon the initial mass of information of the attitude). The data from this study tend to support such a conclusion. The beta weights for two of the three indices of self—concept were statistically significant (p<.05). In addition, the total variance accounted for in attitude change by both variables was statistically significant (p<.005). Nonetheless, total variance ex— plained is low and the contribution of the message index within the model negligible. This chapter has suggested that scaling and analysis techniques require improvement which, in turn, should increases the amount