é. . I "if! 0 ”if It" is x . :fifiuifi I «‘va ‘17:? V $| r—'| p: "" H $33: "3 _. 1Y2? adE'J’: .- , x 7 _ _ ~- “L. ~ - ‘._‘ ‘ --~~ 3.} Mugzgfig “-‘f~- A... . , ..|‘ H'y r .‘ I r . A I1 I? ‘0 NJ; L'~¢>< - r - t- a o-—. c: "I- I:\_l- .ltf‘ .—" ‘13:“ I" I‘m". ll. . ‘ .l 'H' J’hl '1‘“ '1. l'; ”MI. I ' ' l J . ‘3. " .. . u. , 1.. , . Al 'I'I I:' . J: '.‘<. " 3' ‘ W ' . 33'" J W'- ' 1E3}. ' all .’ 4. ‘ . -~ ' m ‘ ‘.u:--.'"£‘? ifléL'ifl'llVHgi 52:53 ' 1 n " ‘ n‘n.‘ .l I. I ‘ _ ' , ‘ ‘u 0‘ V':' V I .' ‘ ‘ I. 4* ’ .. ‘ . . _. ‘ "WW .. "H \5 It“. . THESiS .2 I "w" “fl.,. This is to certify that the dissertation entitled EMPATHY IN INSTRUM Al, COMMUm TEST OF EORVVM w :w— W W“: :r v' r d ~*" presented by i“ JAMES PRICE DILLARD has been accepted towards fulfillment of the requirements for Ph . D. degree in Communicati on Ma jOt’ p ssor Date October 14, 1983 MS U is an Affirmative Action/Eq ual Opportunity Institution 0-12771 itiitttiiiii PLACE IN RETURN BOX to remove thle checkout from your record. VOID FINES return on or More due due. J DATE DUE DATE DUE DATE DUE ' [MW 23‘1993‘ ' i D; Em: “—1: MSUIe eAnAfflrmdlveAcflon/Eqml om Institution WM' EMPATHY IN INSTRUMENTAL COMMUNICATION: TEST OF A THEORY By James Price Dillard A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication ABSTRACT EMPATHY IN INSTRUMENTAL COMMUNICATION: TEST OF A THEORY By James Price Dillard Recent efforts to understand the process by which a potential persuader chooses a message led to the development of an empathy based model of message selection (Hunter & Boster, l978). The primary pr0po- sition of the model asserts the existence of a negative relationship be- tween empathy and verbal aggression. The purpose of the present study was to empirically test this key proposition. Because the empathy construct admits to multiple conceptualizations, the empathy literature was reviewed in an attempt to place the model in a broader perspective. The conceptualization of empathy as a parallel emotional response was adopted for use in this study. In addition, the potential influences of Private Self-Consciousness and Other Directedness in the the process of message selection were considered. Questionnaire data were gathered on 203 Michigan State University students. A confirmatory factor analysis of the instruments intended to tap empathy, Self-Monitoring, and Self-Consciousness revealed that the empathy measure was radically multi-dimensional. Four primary factors were retained, none of which exhibited the strong negative relationship with message selection that was predicted by the empathy model. The multidimensionality of the empathy measure as well as the relationships of the factors to message selection was replicated on a second data set (N = 257). Overall, these findings disconfirm the empathy model. The Private Self-Consciousness message selection relationship was found to be zero. Other-Directedness displayed a positive coefficient with the dependent variable in one of the experimental situations and a negative value in the other. These results are interpreted in terms of the reward orientation of the Other-Directed individual. A series of path analyses were undertaken to explore the relation- ships among the variables. The eventual outcome of these analyses was to suggest the likelihood of a missing variable operating in conjunction with one of the empathy factors, Considerateness, to determine message selection. A search of the literatire revealed that the need for achievement construct possesses properties which would explain the data in the present study. A two factor model of message selection incor- porating both Considerateness and need for achievement was proposed. Despite Office Services ACKNOWLEDGEMENTS As with any piece of work, this dissertation is the product of a number of people. Although expressions of gratitude at this juncture border on injustice the project would not be complete without them. Thus,...' I would like to indicate my appreciation to all of the members of my committee for their support and direction during my years at Michigan State. Dr. Michael Burgoon was an extremely able chair and advisor. His use of a firm hand and free rein succeeded in propelling me from one ivory tower to another. This is a good thing; I owe him a great deal. Dr. Jack Hunter gave very freely of his time. His unflagging good cheer and guidance constituted a major input to this dissertation. Dr. Gerald R. Miller always asked good questions and smoked good cigars. The former had a considerable impact on my education. Dr. Judee K. Burgoon helped me in innumerable ways over the years. She kept every- body, including myself, as honest as was humanly possible. I also wish to extend my thanks to Dr. Franklin J. Boster and Mr. James B. Stiff for graciously making their data available to me. Mr. Robert Addleman, Dr. Franklin J. Boster, Ms. Julia Chrystler, Mr. David D'Alessio, Mr. Jerold L. Hale, Mr. Rodney Reynolds, Mr. Milton Shatzer, and Mr. James B. Stiff have my gratitude for serving as sorters of the empathy items. In addition to these task functions several people and places deserve mention for helping me to keep body, mind, and spirit in pretty much the same location. Jerry Hale and Jim Stiff knew when Silviano therapy could not be put off any longer. Their ministrations were well timed and much needed. Noel Doran made superlative martinis and I sus- pect that he always will. El Azteco and the original Ramon's provided me with nutrition in the style of the correct portion of the country. TABLE OF LIST OF TABLES . . . . . . . . . LIST OF FIGURES SECTION 1. Review of Literature Introduction . . . . . The Empathy Literature Empathy and Aggression Empathy and Altruism . Self-Focused Attention Self-Monitoring and Message Selection II. Methods Subjects and Procedures Independent Variables Method of Analysis of the Measurement Model Method of Analysis of the Theoretical Model III. Results CONTENTS and Message Selection Summary of the Predictions . . . . . . . . . Dependent Variable . . . . . . . . . . . . . Results of the Analysis of the Measurement Model Results of the Analysis of the Theoretical Model Path Analyses IV. Discussion . . . . . . . Empathy Unmeasured Variables Measured Variables . Conclusions REFERENCE NOTES . . . . . . . . REFERENCES 0 O O O O O O O O O 0 APPENDIX A O O O O O O O O O O 0 Additional Data and Alternative Models . . . . Search for Theoretically Meaningful Path Model List of Tables The results of three studies which examined the internal structure of Snyder's Self-Monitoring Scale . . . . . . Results of the cluster analysis . . . . . . . . . . Factor loadings (one side corrected for attenuation), means and standard deviations for each item retained in the measurement model . . . . . . . . . . . . . . Matrix of corrected intercluster correlations with Alpha reliabilities in the diagonal. . . . . . . . . Correlations between the trait measures and message 5e] ECtIon O O O O I O O O O O O O O O O O O O O O 0 Residual matrix for the path model . . . . . . . . . Matrix of corrected intercluster correlations from the combined data sets with Alpha reliabilities in the diagona] O O O O O O O O O O O O O O 0 O O O 0 0 £193 6l 62 63 67 68 69 70 I List of Figures The Marwell and Schmitt Typology of Compliance-Gaining St rategies I O O O O O O 0 O O O O O O O O O O O O O 0 O O A listing of the Self-Monitoring items common to the factor solutions across three studies: Briggs et al. (1980), Gabrenya and Arkin (1980), and Tobey and Tunnell (1981). . The path model. The path coefficients in parentheses are for the study more situation; the unadorned coefficients are for the repayment situation . . . . . . . . . . . . The first path model from the combined data . . . . The second path model from the combined data . . . . The third path mode] from the combined data . . . . A path model with one unmeasured (circled) variable A path model with two unmeasured (circled) variables Page 71 74 75 76 77 78 79 APPENDIX A Page Item Reliabilities, Inter-Item and Item-to Factor Correlations for Humanistic Orientation . . . . . . . . . . 8l Item Reliabilities, Inter-Item and Item-to-Factor Correlations for Consideration . . . . . . . . . . . . . . . 82 Item Reliabilities, Inter-Item and Item-to-Factor Correlations for Fictional Involvement . . . . . . . . . . . 83 Item Reliabilities, Inter-Item and Item-to-Factor Correlations for Emotional Contagion . . . . . . . . . . . . 84 Item Reliabilities, Inter-Item and Item-to-Factor Correlations for Private Self-Consciousness . . . . . . . . 85 Item Reliabilities, Inter-Item and Item-to-Factor Correlations for Public Self-Consciousness . . . . . . . . . 86 Item Reliabilities, Inter-Item and Item-to-Factor Correlations for Acting . . . . . . . . . . . . . . . . . . 87 Item Reliabilities, Inter-Item and Item-to-Factor Correlations for Other-Directedness . . . . . . . . . . . . 88 Item Reliabilities, Inter-Item and Item-to-Factor Correlations for Social Ability . . . . . . . . . . . . . . 89 Item Reliabilities, Inter-Item and Item-to-Factor Correlations for Extraversion . . . . . . ... . . . . . . . 90 As a subspeciality of communication, the area of persuasion has traditionaliy been concerned with the study of the effects of messages on one or more receivers. As Tedeschi, Schlenker and Lindskold (1972) note, this orientation may have been shaped, in part, by the occurrence of world War II. In response to Hitler's propaganda machine, federal funding agencies provided funds for the study of persuasive communica- tion. Thus, North American social scientists began the empirical study of message effects with hopes of developing a technology of attitude change and resistance to change. As World War II continues to recede on the temporal horizon, its impact on the scientific analysis of social influence continues to les- sen. In the recent past, the field of persuasion has begun to move away from its narrow emphasis on attitudes to encompass a broader set of phenomena associated with the persuasive act (MiIler a Burgoon, 1978). One new area, which has been receiving increased attention, is the process whereby the persuader seIects a message which He hopes will re- sult in compliance by the target. The preponderance of investigations in this area of message selection have made use of a rational message selection paradigm. That is, the decision-maker has complete informa- tion on the messages available and can estimate the effect of each. He is forced to select from a finite number of alternatives the options which satisfy his wants. This approach has been used in several substantiVely different areas. In an investigation of the causes of intrinsic and extrinsic motivation in children, Deci, Nezlek and Scheinman (1981) asked grade schooi teachers to indicate how they would use messages which differed in terms of the degree to which they controlled the child or granted autonomy. Kipnis, Schmidt, and Wilkinson (1977) empirically derived an eight category schema for classifying compliance-gaining attempts in an organizational setting. In the second half of their two part study, they showed that the type of influence tactics used differed as a func- tion of the individual's position in the status hierarchy. In a simu- lated organizational study, Ilgen and Knowlton (1980) had student supervisors rate the degree to which they would recommend different types of feedback for subordinates who differed in terms of performance, ability and effort. Among the studies undertaken by communication researchers, the most commonly employed dependent measure has been the list of message strate- gies compiled by sociologists Harwell and Schmitt (1967, see Figure 1). Like the original Harwell and Schmitt study, these.later investigations have focused on short-run compliance attempts (e.g., Lustig & King, 1980; Miller, Boster, Roloff & Siebold,_1977; Sillars, 1980). A recent attempt to provide a formal theoretical framework to this area of instrumental communication is the work of Hunter and Boster (1978, 1979). They offer a model of the psychological decision process of message selection based on the concept of empathy. In brief, Hunter and Boster contend that any given message is likely to produce some affective reaction in the receiver. For example, a threat to strike someone is likely to evoke a negative reaction, such as fear or anger, in the message recipient. It is argued that a persuader's decision about what types of messages might be used in a situation is determined by his expectation of the affective consequences of those messages. Hunter and Boster suggest that individual senders have observed the im- pact of their messages in the past, and on that basis, make predictions (rightly or wrongly) about what the affective outcomes of their message behavior will be in the present. One point at which Hunter and Boster part company with other researchers in this paradigm is with their treatment of the set of mes- sage strategies. Rather than treat each strategy as separate and inde- pendent, they suggest that messages and message strategies can be ordered on an unidimensional continuum which varies only in terms of the direction and degree of affect that will be produced by a given message (messages are considered to be specific instances of more general mes- sage strategies). Strategies such as threat and aversive stimulation fall near the negative end of the continuum, while the positive end is populated by tactics such as reward and positiVe expertise., The model specifies that the set of strategies form a Guttman scale such that for each persuader there is a point on the continuum which separates all those strategies he would be willinggto use in a partic- ular situation from those he would not.‘ That point is called the per- suader's ethical threshold. The Hunter-Boster model differs from the majority of message selec- tion research in another important way; it posits the existence of a theoretical process which may explain strategy selection. It is assumed that individuals experience an emotional response as a result of per- ceiving the affective state of others. The empathy model asserts that it is the anticipation of experiencing vicarious affect that is funda- mental to determining the messages chosen for use in a compliance- gaining situation. 'Thus, the most essential statement of the model is that the range of messages a persuader is willing to use is a function of the anticipated emotional consequences for the receiver and in turn for himself. To date no study has directly assessed the functioning of the empathy mechanism as it relates to message selection. The purpose of the present study is to provide a test of the theoretical process sug- gested by Hunter and Boster, as well as to explore some elaborations and alternatives. Since a well-developed body of research exists on the topic of empathy, a review of that literature will assist in‘delineation of-the construct. The Empathy Literature Early approaches to the study of empathy tended to view the construct as a cognitive skill. Empathy was thought of as the relative ability of an individual to adopt the phenomenology of another. Such a perspective is very much in keeping with the lay use of the term; "to put oneself in another's shoes,” or to take on the perspective of another. Stotland (1969) labeled this "predictive empathy" because the methods employed by such empathy researchers required subjects to fore- cast the state of mind of their experimental partner. Work by Dymond (1949, 1950) typifies the procedures employed by researchers operating within the predictive empathy framework. In her studies, she had pairs of subjects follow a four-step procedure. First, each subject A was asked to rate himself or herself on six, single-item trait measures (self-confidence, superior/inferior, selfish/unselfish, friendly/unfriendly, leader/follower, sense of humor). Next, subject A rates subject 8 as A sees him. Third, A judges B as he thinks B would judge himself. Finally, A rates himself as A supposes B would rate him. A measure of empathy for A was then calculated by observing how closely A's predictions of 8's ratings (steps 3 and 4) correspond to 8'5 actual ratings (steps 1 and 2 for subject 8). These procedures came under attack when Cronbach (1955) demon- strated that data produced by such methods may be, in part, artifactu- ally determined. Hhile matching responses for A and 8 might result from empathic processes, it is also possible that the match was brought about by (1) sharing the same response bias as the other, (2) stereotypical knowledge of the other person, or (3) actually having common character- istics as the other and assuming that the other ratings are like one's own. Since no methodology has yet been devised which enables the would- be empathy researcher to parse out these variance components, empirical research from this perspective pretty much ground to a halt. Empathy re-emerged somewhat later in an altered form. These second generation empathy studies shifted the emphasis from the cognitive arena to the affective. Stotland's (1969) definition is paradigmatic. He states that empathy is ”an observer's reacting emotionally because he perceives that another is experiencing or is about to experience an emo- tion” (9. 272).' This conceptualization fits nicely with the Hunter- Boster model since it specifically includes empathic emotion as an anticipatory state. Stotland considers an emotion to be composed of two parts; the physiological and the subjective. Physiological arousal occurs first and is labeled by the individual as an affective response. Both physio- logical and subjective components are necessary for an emotion to exist. It is noteworthy that Stotland's definition does not require that there be a high degree of similarity of emotion between observer and ob- served. As a practical matter it seems likely that most individuals will experience at least the same direction of emotion, i.e., positive or negative. Hoffman (1977) suggests that we can expect a reasonable match due to the fact that people all have basically the same nervous system and they have in common a large number of affective experiences as a result of socialization. However, the definition does not preclude the possibility of some sadistic individual deriving genuine pleasure from observing another writhing in pain. . A final point is that Stotland (1969) sees empathy as the conse- quent of a cognitive process. This is consistent with a great many modern day approaches. For example, Miller and Steinberg (1975) empha- size that stimulus discrimination must occur prior to an emotional. response. That is, the empathizer must perceive that the target is ex- periencing distress before making an emotional response to the target. The upshot of this line of thinking is that perspective taking has been treated experimentally as an independent variable which brings about -change in an individual's emotional state. An exemplar from Stotland's (1969) research illustrates the implications of his definition. Emotional empathy has been examined as a dependent variable most frequently operationalized by measures of vasoconstriction and palmar sweating. Questionnaire measures were also usually taken, but were seen as a means of clarifying physiological \ responses, rather than as direct evidence of empathy in their own right. Stotland and his colleagues conducted a series of studies in which sub- Jects watched a confederate experience pain, pleasure, or no emotion at all. Participants were variously instructed to (1) imagine how the ”demonstrator" (the confederate) was feeling, (2) imagine how they would feel if they were the demonstrator, or (3) to simply watch the demonstrator. While data from the gross physiological measures did not always support the hypotheses, the questionnaire data did. Subjects in the imagine-self conditions reported more emotional arbusal than respondents in any other condition. Overall, the results provide rea- sonable support for the notion that individuals may experience emotions vicariously and that the perspective-taking orientation (i.e., cognitive set) of the observer affects the extent of emotional arousal. . Several writers have attempted to bring coherence to the empathy literature by proposing models which specify the substantive domain of empathy. Feshbach (1975) offers a three dimensional conceptualization of empathy. She defines empathy as 'a match between the affective response of a perceiver and that of a stimulus person" (p. 26), but does not specify the degree of match necessary for empathy to be present. Empathy is treated as consisting of one affective and two cognitive com- ponents. The capacity to discriminate both the role and the perspective of another individual are cognitive abilities. These competencies are of interest to Feshbach since much of her work is done with children. In populations of youngsters these abilities may be highly variable, depending on the relative development of the child. Role- and ‘ perspective-taking as an ability probably differs very little among normal adults (Hoffman, 1977). The third necessary element in Feshbach's (1975) approach is emo- tional capacity and responsiveness. That Feshbach emphasizes this com- ponent over the others is reflected in statements such as ”the follow- ing; ”while the cognitive dimension of empathy is important, it is the. affective component that gives the empathy construct its unique proper- ty" (p. 26). A more behavioristic orientation to empathy is advanced by Miller and Steinberg (1975). In keeping with their transactional view of interpersonal communication, these writers stipulate that two steps are necessary for empathy to have occurred. First, the would-be empathizer must accurately predict the other's attitudes and values. Following these social perceptions, the empathizer must present behaviors which the other perceives as rewarding. As can be seen, this conceptualiza- tion of empathy emphasizes the behavioral, rather than psychophysical, response to the other. Hoffman (1977) presents a developmental theory of empathy which asSumes the existence of cognitive, affective, and motivational processes. He suggests that among adults the perdeption of another in need activates cognitive processing. An affective response, empathic distress, occurs next. This may in turn create other forms of arousal all of which join together to motivate some behavidr. Presumably, the resulting behavior reduces empathic distress. The treatment of empathy in the present paper draws in a loose way on all three of the previously presented models. First, it is assumed that some form of cue discrimination and the attachment of meaning to that cue is necessary for empathy to take place. However, that is not the focus of this investigation. Rather, the Hunter-Boster model im- plies that the key determinant of message selection will be emotional response. Consequently, that will be the primary emphasis in this paper. Like Hoffman, it will be assumed that emotional response acts as amotivating force. If the model is correct, this response, or the anticipation of it, should produce a behavioral'outcome, i.e., an indi- cation of willingness to employ a given range of message strategies in an instrumental communication situation. Empathy and Aggression Two areas to which social psychologists have devoted relatively large amounts of time and energy, are altruism and aggression. While these two areas are not often examined together, the frequency with which the empathy construct appears in models of altruism and models of aggression, indicates that it may be useful to do so. Another hint that an analysis from both perspectives might prove beneficial comes from the Hunter-Boster model. Their unidimensional conceptualization of message behavior suggests the possibility of a parallel treatment of social be- havior. That is, social actions may be placed on a continuum which varies in degree and ranges from positively valued to negatively valued. At one end we find selfless acts of altruism, and at the other pole, in- stances of aggression for its own sake. This section will review a selective sampling of the literatures or pro- and anti-social behavior as they relate to empathy. Aggression will be considered first. . Clark (1980) casts empathy as the counterforce to individuals' desires for power. He suggests that it is these egoistic desires for power which result in murder and mayhem of all varieties. In contrast, the empathic person is one who is more inclined to experience the feelings of others. The ”functionally empathic" individual is compelled to help and support other individuals. We need more functionally empathic people, says Clark. In fact, "the survival of the human species now appears to depend upon a universal increase in functional empathy" (p. 190). While other social scientists accord empathy somewhat less importance, it is a construct which figures very prominently in many studies of aggression. Similar to Clark (1980), S. Feshbach (1964) has 10 proposed a model of aggressive behavior in which empathy serves as a regulator. He suggests that it is fruitful to limit the use of the term aggression to these acts which are intentional. In keeping with this constitutive specification, he borrows from Sears, Maccoby, and Levin (1958) to distinguish between two forms of aggression. Instrumental aggression is that which is directed toward obtaining some nonaggressive goal. Conversely, the purpose of hostile aggression is to cause pain or injury. This distinction is an important one because Feshbach suggests that the two types of aggression operate quite differently from one another. When one person attacks another, it is usually the case that he perceives the distress of the victim. Feshbach's (1964) model suggests that this may have two possible effects. First, if the aggression was hostile, exposure to the pain of the target may be reinforcing to the .aggressor since he has been successful in obtaining his goal. Presum- ably, this is gratifying, and consequently reinforcing. For instrumental aggression Feshbach asserts that One would expect these painful, vicarious responses to function as inhibitors of the child's [aggressor's] own aggressive responses. These inhibitory tendencies should occur even if the instrumental aggressive act is success- ful in achieving some nonaggressive goal. Thus, children high in empathy should, on the whole, manifest less overt aggression than children low in empathy. (p. 260) N. Feshbach and S. Feshbach (1969) set themselves to the task of testing S. Feshbach's (1964) hypothesis of a negative correlation be- tween empathy and aggression. They assigned children to high or low empathy groups on the basis of their scores on the Feshbach and Roe (1968) empathy measure. This index requires that the subject view a series of slides depicting emotion-producing situations for the actor ll (e.g., a boy loses his dog forever). When subject describes his own emotional reaction to the slides, the responses are coded as empathic if they match Feshbach's and Roe's notion of the appropriate emotional response, and nonempathic if they do not. Gender of subject and age (4- and 5-year-olds versus 6- and 7-year olds) filled out the 2 x 2 x 2 design. The dependent variable was aggression. The results indicated no differences in aggressiveness for girls of differing empathic ability at either age group. For the younger boys, those high in empathy were rated as more aggressive than the low empathy boys. For the older boys the data showed just the reverse; the more empathic, the less aggression. While only this last finding comports with the hypothesis, Feshbach and Feshbach conclude that the data partially support the predicted relationship. A more accurate interpretation might be that little can beconcluded from these data. The study is fraught with problems, not the'least of which are measurement error (in both the empathy and aggression tests) and samp- ling error (some cells contained as few as eight subjects). Despite these problems this study is often cited as support for the existence of a negative correlation between empathy and aggression. Fortunately, other investigations exist which circumvent many of the problems of the previous study. Most of these studies utilize a ‘ learning situation guise reminiscent of Milgram (1965) to create condi- tions of instrumental or hostile aggression. The work of Baron (1971a) typifies this line of research. In this study subjects were told that they would be using electric shock to "teach" the learner (confederate) a series of nonsense syllable pairs. Prior to the learning task the subject met the confederate and received either a positive or negative )2 evaluation from him. The positive evaluation was intended to create a condition of no anger, while the negative evaluation was supposed to provoke the subject. To act as teacher, the subject was seated in front of a Buss (1961) aggression machine. This consists of a metal box on the face of which is mounted ten shock buttons which alledgedly vary in intensity. The subject was allowed to select a shock level whenever the learner made an error, and was free to depress the button for as long as he thought appropriate. Both intensity and duration data were collected. The victim's pain cues were made available to the subject via a bogus 'psychoautonomic pain meter” located on top of the aggression machine. In this study, as in most of Baron's work, empathic processes are assumed to be operating but are not measured directly. Analysis of the Baron (1971a) data revealed that signs of suffering hadian aggression-inhibiting effect for both instrumental (nonangry) and hostile (angry) situations. Although similar results are reported by Baron (1971b) and Geen (1970), later work showed support for the predic- tions of Feshbach's (1964) model. Bandura (1971) suggested that the anger manipulations were too weak to really make the subjects hostile. Baron (1974) tested this speculation using a stronger anger manipula- tion. He found that pain cues acted to inhibit aggression when the sub- ject was acting instrumentally, but facilitated attack among subjects in the high anger condition. Rule and Nesdale (1976). after reviewing over 60 investigations, came to the same conclusion. Thus, it appears that information indicating that a target person is experiencing negative affect acts to reduce the aggressiveness of the attacker. The Hunter-Boster model makes a similar prediction for verbal )3 aggression. The case for empathy acting as a mediator in this process would be a good deal more convincing had measures of empathic response been taken. Unfortunately, very little data exists which bears on this point. Some evidence for the role of empathy as a mediator comes from the work of Mehrabian and Epstein (1972). These researchers developed a 33- item scale which they contend measures a dispositional tendency to be emotionally empathic. Mehrabian and Epstein undertook several experi- ments in an attempt to demonstrate the validity of their instrument. Two of these closely resemble Baron's investigations. Subjects were brought into the lab for the purpose of acting as a teacher for another supposed subject. The learning task was again pairs of nonsense syllables. The equipment used was the aggression machine. The learner was either in close proximity to the teacher or in a nearby room. The teacher received visual and vocal information about the learner's suffering in the first case, but only vocal in the second. Subjects were classified on the emotional empathy measure as either high or low empathics. ' The results showed a gender effect (females were less aggressive and more empathic than males) and an empathy by immediacy interaction. Aggression was lowest for high empaths in the immediate condition. A second study showed the same pattern of results. Given the cumulation of empirical and theoretical evidence it is reasonable to conclude that empathy plays an important part in the process of instrumental aggression. This review of the literature has demonstrated the existence of a body of data on the general topic of aggression which closely parallels the predictions made by the empathy 14 model of message selection. As such, this provides indirect support for the generality of the model. ,Empathy and Altruism Some differences of opinion exist among researchers as to the nature of the motivational mechanism which brings about altruistic behavior.‘ Several suggest that empathy with a victim's plight causes empathic arousal which is distressing to the observer. It is thought that sub- jects in helping behavior studies act altruistically as a result of a desire to reduce their own aversive state. The motivation in these models is primarily egoistic (Piliavin, Rodin, & Piliavin, 1969; Gaertner & Dovidio, 1977). Hoffman (1977) has extended this thinking to include another form of distress. He contends that individuals who are aware of the suffering of another first experience a parallel response to the victims feelings (empathy) which is then followed by feelings of concern or compassion. He labels these latter emotions sympathic distress. This secondary response is brought about by the initial feelings of empathic distress and is distinct from empathy in his model. Another result of empathic distress, in Hoffman's model, may be guilt. When individual's perceive themselves as the source of distress they may transform their empathically aroused emotions into feelings of self-blame. Others have suggested the existence self-punishing emotions which arise from a violation of internalized moral standards (Berkowitz, 1972; Staub, 1974). Analogous to the conceptualization of empathy as an anticipation of an emotional response, Rawlings (1970) has put forth the notion of "anticipatory guilt." She suggests that simply anticipating suffering may be sufficient to cause helping behavior. In contrast to the egoistic forces suggested by some writers, Coke, Batson, a McDavis (1978) have posited the existence of a purely altruistic motivation. They suggest that individuals possess a genuine desire to sacri- ' fice for their fellow humans and that this motive is activated by empathic arousal. While Coke et al.'s work deals exclusively with situational determi- nants of empathy, Ardher, Diaz-Loving, Gollwitzer, Davis, and Foushee (1981) have sought to extend the model to include social evaluation and dispositional differences in tendency to be emotionally empathic. In sum, several different reasons for engaging in prosocial behavior have been suggested. First an individual confronting another person in need may himself experience aversive affect as a result of empathic processes. The ob- server might then act altruistically to reduce his own discomfort. Second, persons may help to avoid guilt feelings. This type of emotion is presumably aroused when an individual finds himself acting in a way that is not in accor- dance with internalized standards of helping. Third, pedple may help others because of a concern for the well-being of other people (sympathy). The common theme to all three motivational schemas is empathy. In each case it is taken to be a prerequisite to the activation of motivation to behave altruistically. Three empirical studies offer convincing support for the notion that empathic distress motivates helping. Krebs (1975) used both physiological and self-report indices of the degree of empathic arousal subjects experienced while watching a model undergo a shock. A measure of altruism was taken after the empathy trials which required subjects to choose between helping the model at a cost to themselves or aiding themselves at the model's expense. 0f the two experimental conditions, the one in which the subjects showed the most empathy was also the one in which they behaved most altruistically. )6 Gaertner and Dovidio (1977) also provide convincing support for the operation of empathic processes. Subjects were presented with a situ- ation in which a confederate apparently suffered physical harm (had a stack of chairs fall on her). Arousal was measured by heart-rate acceleration. The major finding was that the more quickly the subject exhibited cardiac acceleration, the more quickly she acted to help the confederate. In this case, as with the Krebs (1975) study, the temporal ordering of the components of the empathic process are clear; a stimuli occurs which triggers an empathic response, which in turn brings about a prosocial behavior. Additional evidence for this conclusion comes from Mehrabian and Epstein (1972). Five separate individual difference measures including empathic tendency were taken on all of the female subjects. Participants 'were given an opportunity to help a female confederate who was trained to exhibit anxiety and depression over a class. The only significant effect found in the helping data was for emotional empathy. By way of sup- porting evidence for empathic arousal, Mehrabian and Epstein note a moderate correlation between selfereports of empathic tendency and characteristic arousal level. i All in all, the empathy model of verbal aggression has considerable support from two literatures of social behavior. By considering aggres- sion and altruism as opposing forms of social behavior in a way suggested by the Hunter-Boster model, it was demonstrated that much of the extant literature on social behavior is congruent with the predictions of the model for verbal behavior in instrumental aggression situations. Empathic processes were shown to have an inhibitory effect on aggression and to enhance prosocial behavior. )7 In the case of both pro- and anti-social behaviors, data exist which lend credence to hypothesized empathic processes. No corresponding data exist for the model of message selection. In light of the previous analysis and the untested prediction of the model, it is expected that a measure of emotional empathy, such as that offered by Mehrabian and Epstein (1972), will correlate negatively with a willingness to use a wide range of compliance-gaining strategies. Self-Focused Attention and Message Selection In 1972 Duval and Wicklund proposed a theory of self-awareness which has subsequently spawned a great deal of research. Although the theory itself has few direct implications for an empathy-based model of communi- cation behavior, much of the research based on the theory does. _This section will survey the portions of the theory and empirical literature on self-focused attention which are relevant to the study of verbal aggression and instrumental communication. The key term in Duval and Wicklund's (1972) theory is objective self-awareness. This is defined as "a state in which the person takes himself to be an object" (Wicklund, 1975, p. 234). The theory assumes that at any point in time an individual's attention is focused entirely on himself or entirely on the environment. The proportion of time spent in self-reflection is defined as the extent of self-awareness. The theory states that any stimulus which serves to remind a person of his object status will act to increase self-awareness. Other stimuli will tend to focus attention outward, thereby decreasing self-awareness. Having subjects view themselves in a mirror or listen to their own tape- recorded voices are two experimental manipulations which have been used to magnify self-awareness (Wicklund & Duval, 1971). The presence of 18 either a television camera (Wicklund a Duval, 1971) or an audience - (Scheier, Fenigstein, & Buss, 1974) has also proven effective in boosting self-awareness. Buss (1980) has modified Duval and Wicklund's theoretical framework and extended it to include individual differences in self-focused atten- tion. He distinguishes between the state of self-awareness and the trait of self-consciousness. The latter is defined as “the consistent tendency of persons to direct attention inward or outward" (Fenigstein, Scheier, a Buss, 1975). Buss further subdivides the period of inward directed attention. A person who tends to reflect upon his own internal states is high in Private Self-Consciousness (PRSC). Conversely, an individual who spends a high proportion of his time considering observable aspects of himself is defined as being high in Public Self-Consciousness (PUSC). Fenigstein, Scheier, and Buss (1975) report the development of a scale to measure both forms of self-consciousness. -Analyses of the ° internal structure of the test indicate that the subscales are unidimen- sional and that they exhibit adequate internal reliability. Validation studies of the scales have provided evidence of convergent and divergent validity (Carver & Glass, 1976; Turner, Scheier, Carver, a Ickes, 1978). The domain of the private self includes affective as well as cogni- tive events. For example, while at rest one may be dispassionately aware of a desire to succeed in a sporting event. This desire may become a highly charged feeling in the heat of actual competition. This is worth noting because the theory assumes that increases in attention focused on aspects of the private self will bring about changes in both. 1. Private self-focus is assumed to make all private events, both affectively charged and neutral, clearer and more distinct. )9 2. Attention to the private aspects of oneself is assumed to intensify the affective charge of bodily stimuli, moods, motives, fantasies, and self-esteem (Buss, 1980, p. 13-14). That is, persons who are high in PRSC, or who are made self-aware, will experience transient emotional states more strongly than persons who are low in PRSC or do not receive a self-awareness manipulation. Thus, to the extent that the empathy model is valid we would expect persons high in PRSC to experience stronger affect and consequently to exhibit less- willingness to aggress verbally. Several studies exist which support the proposition that self- focused attention increases degree of affect. Scheier and Carver (1977) demonstrated that undergraduate males who viewed slides of nude women found the women more attractive when self-awareness was induced. In a second experiment, subjects high or low in PRSC were exposed to slides of either nude women or mutilated bodies. An interaction between PRSC and type of stimulus was found such that higthRSC subjects reported more in- tense affect (either attraction or repulsion) than did the low PRSC par- ticipants. Two additional experiments reported in the same Scheier and Carver (1977) article provide further support for effects of self-focused atten- tion on affective response. Participants in both cases were subjected to a mood induction procedure which involved reading a series of affectively positive or negative statements. The presence of a mirror produced more extreme reactions in the subjects. Similar results were found in a different sample when participants were split using their PRSC scores. A later study (Scheier, Carver, Schulz, Glass & Katz, 1978) revealed the existence of a sympathy effect which was particularly strong among 20 high PRSCs. Persons were asked to evaluate a stigmatized and non- stigmatized target. High PRSC subjects rated the stigmatized person more favorably than did low PRSC subjects. The two groups did not differ in their ratings of the normal target. This study demonstrates the intensi- fication of emotion among high PRSCs. When subjects are provoked, PRSC correlates positively with self- reports of anger. Scheier (1976) had a confederate verbally derogate a subject and then gave the subject an opportunity to aggress against the accomplice. High PRSC subjects reported the greatest degree of anger and aggressed significantly more than the low PRSC participants. This is consistent with Feshbach's (1964) assertion that increased emotional response in hostile aggression situations tends to enhance the aggressive response of the subject. In summary, these studies uniformly support the assertion that private self-consciousness intensifies emotion.. They show that high PRSC individuals are likely to experience a range of emotions more strongly than their less self-conscious counterparts and that these emotional responses are likely to inhibit their aggressive tendencies in instru- mental aggression situations and to enhance it in hostile aggression situations. Considered in the framework of the empathy model, this sug- gests that PRSC should exhibit an inverse relationship with persuasive [strategy selection. An additional possibility exists. That is, PRSC may interact with emotional empathy. If subjects are classified according to their scores on emotional empathy and PRSC, then persons high in both should exhibit the least verbal aggression. For people low on empathy, PRSC should have little effect. That is low empathy/high PRSC persons should differ lit- tle, if at all, from low empathy/low PRSC persons: if the individual is not particularly emotionally responsive it should matter little if his awareness is heightened or not. Thus, if an interaction should appear, its form would be expected to be ordinal. Self-Presentation and Message Selection An alternative to the notion that empathy is the limiting factor on verbal aggression is provided by a number of psychologists who have developed theories of the self (e.g., James, 1890). For example, one recent proponent of this approach to the study of human behavior is Goffman (1959,_1967). He has developed a theatrical metaphor as a basis for the study of human communication. Following Goffman's lead, Snyder (1974) has defined a construct called ”self-monitoring” and developed a. corresponding scale. He argues that individuals may be differentiated in terms of their "concern for social appropriateness, sensitivity to the expression and self-presentation of others in social situations as cues to social appropriateness of self-expression, and use of these cues as guidelines for monitoring and managing self-presentation and expressive behavior” (p. 529). This notion has proved intriguing to many researchers and conse- quently has generated a good deal of research (see Roloff, 1980 or Snyder, 1979 for reviews). In general, support for the social percep- tiveness aspect of the construct has been mixed. Two studies offer sup- port (Geizer, Rarick, a Soldow, 1977; Kraus, Geller, & Olson, Note 1) while three others present disconfirming evidence (Cunningham, 1977; Ickes & Barnes, 1977; Snyder, 1977). Briggs, Cheek, and Buss (1980) point out that the first two studies tested the ability of the self- monitor to detect deception on the part of another. The nonsupportive 22 , studies all examined subjects' ability to decode a much wider range of nonverbal behavior. Thus, high self-monitors may be particularly sensi- tive only to social cues involving deception. Evidence for the self-presentation ability would seem more clear. Lippa (1976) reports that high self-monitors are seen as more outgoing and extraverted than are low self-monitors. Other evidence shows that self-monitoring correlates positively with a tendency to initiate a con- versation. Also, as would be expected for individuals who are socially adaptive, there is some data to support the notion that high self- ' monitors exhibit greater variance as a function of situational differ- ences than do low self-monitors (Rarick, Soldow, a Geizer, 1976; Snyder & Monson, 1975; Danheiser & Graziano, 1982). I Unfortunately, recent analyses of the internal structure of the 25 item scale reveal the existence of at least three distinct factors (see Table 1). Two studies report very similar three-factor solutions. Factor one they label Extraversion (Briggs, et al., 1980) or Sociability (Tobey & Tunnell, 1981) because it is composed of reverse-coded items which appear to tap anxiety in social situations. The second factor con- sists of questions concerned with individual's estimates of their acting ability and consequently is labeled Acting in both studies. A third fac- tor, named Other-Directedness, appears to measure the self-regulation aspect of the self-monitoring construct as proposed by Snyder (1974). A recurrent theme in the items which comprise the Other-Directedness factor is an instrumental orientation (see Figure 2 for the items). These items are permeated with the flavor of an amoral social operator. A desire for personal gain without regard for maintaining correspondence between one's true feelings and behavior comes through strongly. This interpretation 23 is consistent with the thinking of Danheiser and Graziano (1982) regarding the motivation of the self-monitor. These writers have sug- gested that self-monitors' concern for social appropriateness arises out of a desire to maximize their own outcomes. Data from their study make two points. First, high self-monitors exhibited greater variance in their behavior than did low self-monitors. Second, in the prisoner's dilemma gaming situation, high self-monitors tended to use more coopera- tive strategies when they anticipated future interactions with their partner than when they did not. Moreover, high self-monitors were less cooperative than lows when they believed that there would be no future meetings. Since self-monitoring correlated (r a .26) with Eysenck and Eysenck's (1964) measure of extraversion, the analyses were repeated blocking on extraversion. The results were replicated. Thus, although the Self-Monitoring Scele was treated as unidimensional in the covariance analysis, it is safe to assume that the effects of the Extraversion fac- tor were removed. While there is no way to rule out the possibility that the Acting factor is responsible for the results, it seems unlikely since there is no clear theoretical reason for supposing that to be the case. Conversely, the Other-Directedness factor appears to be a very likely candidate for such an effect. It contains at least one item (13) which specifically taps behavioral variability as well as several which suggest that for the high Other-Directed individual there is no ethical dilemma associated with acting instrumentally even if it violates one's own feel- ings, attitudes and beliefs. Other data provide additional information about the high Other- Directed individual. Briggs et al. examined the relationships between Other-Directedness and Sociability (.05). Neuroticism (.48), Shyness (.37), Self-Esteem (-.49), Public Self-Consciousness (.28), and Private Self-Consciousness (.13). Tobey and Tunnell (1981) report a similar value for the correlation between Other-Directedness and Public Self- Consciousness (.31). Gabrenya and Arkin (1980) found somewhat lower correlations for Other-Directedness with both Public Self-Consciousness (males 8 .13 and .16, females 8 .14 and .16) and Private Self- Consciousness (males and females 8 -.03 and -.07). The lower correla- tions for Public Self-Consciousness in the latter study are probably due to the fact that Gabrenya and Arkin's Other-Directedness factor was com- posed of a smaller number of items than either Briggs et al. or Tobey and Tunnel. Thus, there were some content differences (omissions) for that factor, as well as a potentially depressed internal reliability due to the smaller number of items. The picture that emerges of the Other-Directed (00) person is some- what different than, though not wholly inconsistent with, Snyder's self- monitor. The consistent positive correlations with Public Self- Consciousness suggest that the prototypical 00 person is concerned with social appropriateness. Perhaps because of this predilection to attend to the self as a social object 005 suffer from low Self-Esteem and Neuroticism (cf., Fenigstein, Scheier & Buss, 1975; Wicklund, 1975). The positive relationship with Shyness may also spring from the anxiety- provoking nature of social situations for self-conscious persons. Although the high 005 may find social interaction cause for some distress, this does not seem to interfere with either their desire to get what they want by skillful presentation of their own behavior. They seem to feel that deception in general, and ingratiation in particular, are justifiable modes of behavior when they are necessary to maximize rewards. 25 The orientation toward personal gain that must be attributed to high 005 on the basis of this analysis, suggests a person who is relatively uncon- cerned with the outcomes of their actions for others. This implies that, in instrumental communication situations, high 005 would be willing to engage in a greater degree of verbal aggression in order to obtain their goals than would low 00s. Therefore, a positive relationship is pre- dicted between Other-Directedness and compliance-gaining message selec- tion. Summaryfof the Predictions Three possible predictors of compliance-gaining message strategy selection have been examined. The first follows most directly from the Hunter-Boster empathy model. It posits a negative relationship between emotional empathy and strategy selection. I Self-consciousness theory provides the background for the second prediction. Persons high in Private Self-Consciousness are defined as being particularly aware of their own internal states. Thus, a high PRSC individual should be especially sensitive to empathic arousal. On these grounds, it is expected that PRSC will inhibit a willingness to use a wide range of message strategies. This may be evidenced by an ordinal interaction between the individual differences, emotional empathy and PRSC. Finally, the self-monitoring construct was scrutinized in relation to message selection. Several studies which examined the internal struc- ture of the measure found it to be multi-dimensional. The Other- Directedness factor appears to embody a number of aspects of the Self- iMonitor but with some differences. The Other-Directed person seems to I>ossess an instrumental orientation, i.e., an amoral perspective that 26 would certainly permit the use of Other humans for personal gain. Thus, Other-Directedness will show a positive correlation with strategy selec- tion. Methods Subjects and Procedure Students enrolled in undergraduate communication courses at Michigan State University served as subjects (N = 203). All questionnaires were administered in class after advising students of the voluntary nature of the study. No one refused to participate. After the questionnaires were collected, the nature and purpose of the research was explained to the participants. Independent Variables Emotional empathy was measured using the 33-item scale developed by Mehrabian and Epstein (1972). The four-point response scale was anchored with Strongly Disagree, Disagree, Agree, and Strongly Agree. Private self-consciousness was assessed using the instrument developed by Fenigstein, Scheier, & Buss (1975). The four-point response scale was anchored with False, Mostly False, Mostly True, and True. The 25-item scale used to assess self-monitoring (Snyder, 1975) employed identical anchors. Items designed to gather data on gender, age, racial background, religion, political views and affiliation, and academic background were also included. 27 Dependent Variable Two situations used in previous research were chosen for inclusion in the present study. Half the subjects were randomly assigned to one situation, half to the other. The first originally appeared in Harwell and Schmitt (1967) while the second was developed by Kaminski, McDermott, and Boster (1977). The text for the two situations was as follows: Your teenage son, Dick, who is a high school student, has been getting poor grades. You want him to increase the amount of time he spends studying from 6 to 12 hours a week. How willing would you be to use each of the following messages in order to persuade your son to study more? The second situation was: Your best friend has a habit of borrowing money from you and not repaying it for long periods of time. Recently s/he borrowed S20 from you. Because of some expenses you have incurred, ydu need the money back quickly. ~You want to persuade her/him to repay the loan very soon. Which of the following messages would you be willing to use to persuade your friend to repay the loan? A Data from the pretest and manipulation checks in Williams and Boster (1981) revealed that subjects' perceptions of these two situations differ in terms of self- and other-benefit. Items included in the questionnaire to measure compliance-gaining message selection were taken from Williams and Boster (1981). Thirty-two messages, two per strategy, were used for each situation. In accordance with Hunter and Boster's (1980) assertion that the process of message selection is one of making Bernoulli judg- ments, subjects were asked to indicate whether they "would use" or "would not use” each message. 28 Analysis of the Measurement Model A necessary condition for a fair test of the theoretical predictions is the establishment of a valid and reliable measurement model. Confirmatory factor analytic methods, alternately called multiple groups analysis, were used for that purpose in the present study. This pro- cedure involves placing items in clusters determined a priori, then sub- jecting each item to three tests (Hunter, 1980). The first of these tests is homogeneity of content. That is, the items in a cluster must all have similar substantive meaning. . Second, the variables in a cluster must exhibit internal consis- tency. The intra-cluster correlation matrix should contain only positive values (after reflection of negatively worded items). Ideally, these items should be of uniform quality, i.e., the matrix should be flat or of unit rank. The third method of assessing the unidimensionality of a set of items is to compare their correlations with other clusters. This is called external consistency or parallelism. If two variables measure the same thing, then not only should they show a positive correlation with each other, but they should correlate with other variables in a similar fashion. Some variability must be expected as a result of differences in individual item reliabilities and sampling error. Each of these three tests constitute a necessary condition for the inclusion of an item in a cluster. Together they are sufficient. If an item fails any one of the three tests it is placed in a residual cluster. The items which remain in the original clusters, or are recombined in some way to form new clusters, make up the purified measurement model. The next step in the analytic procedure is to examine the relationships among the clusters. 29 Analysis of the Theoretical Model A simple correlation coefficient was sufficient to test the empirical validity of the predictions for emotional empathy, private self-consciousness and other-directedness. Scatterplots provided infor- mation regarding the functional form of each of these relationships. A test of the interaction between emotional empathy and private self- consciousness required the use of multiple regression. The product vec- tor of the two terms, which represented the variation due to the inter- action, was entered into the regression equation after the two main effects components. The size of the slope for the product term provided an indication of the importance of the interaction in determining message selection. Results Analysis of the data followed the two step procedure outlined in the previous section. First, the measurement model was validated. The mul- tiple groups analyses were accomplished using PACKAGE (Hunter, Cohen, & Nicol, 1975). Communalities were placed in the diagonal in all cases. The number of subjects for which pairwise data was available ranged from 203 to 195. Tests of the theoretical relationships were performed using the multiple regression subprogram of the Statistical Package for the Social Sciences (Nie, Hull, Jenkins, Steinbrenner, & Bent 1975). Results of the Analysis of the Measurement Model Since this investigation was conducted using measures for which some data about the validity and internal structure of each teSt already existed, items were placed into clusters using that information. Follow- ing the procedure recommended by Mehrabian and Epstein (1972, p. 527) the Emotional Empathy Scale was initially treated as unidimensional. That 30 is, after reflection, the 33 items were summed. Items used to tap private and public self-consciousness were grouped according to Fenigstein, Scheier, and Buss (1975). Clusters consisting of self- monitoring items were formed after reviewing the results of earlier ex- ploratory analyses. If an item was common to two of the three solutions reported in Table 1, it was placed in that cluster. Items which did not meet this criterion were placed in a residual cluster. The initial analysis also included clusters representing strategy selection, gender, and political orientation. The latter two single-item measures were used as reference variables. The initial cluster analysis revealed that many of the items'failed to meet one of the three tests described earlier. The Emotional Empathy Scale was particularly troublesome. The interitem correlation matrix for that measure exhibited a large number of negative and near zero correla- tions. Since Mehrabian and Epstein (1972) assert that the instrument is composed of seven subscales it was thought that using the 33 items grouped into seven clusters might provide a workable solution. Unfor- tunately, Mehrabian and Epstein did not assign all the items to subscales which appear in their article. Rather, they provided two examples for each subscale from the 33 item pool. Since this left over half the items unassigned a plan for placing the items into clusters was developed. Nine communication scholars agreed to act as judges in a sorting task. They were presented with the names of the subscales and the corresponding example items. Using this information they attempted to assign the remaining items to the appropriate factor. The results were not pretty. There was 1002 agreement on only three of the 19 unassigned items; 882 agreement on two items; 772 agreement on two items; 662 agreement on 3) three items; 552 agreement on four items. For the remaining five items the judges agreed less than half the time. Since these findings indi- cated that cluster development on the basis of content was not feasible strategy, a blind multiple groups factor analysis was performed on the full set of 33 items. This analysis yielded three factors which provided a basis for beginning the confirmatory analyses. These three clusters, in addition to the self-consciousness and self-monitoring clusters, were successively modified until the measurement model fit the data. The names of the resulting clusters and the original source of each item appear in Table 2. See Table 3 for the items which make up each cluster and their corresponding factor loadings. Four distinct factors emerged from the Emotional Empathy Scale. The first, labeled Humanistic Orientation, appeared to tapping a generalized appreciation of others and concern for their emotional welfare. As a whole these items seem to measure a value or deep-seated perspective on humankind. The importance placed on the dignity of others is evidenced by items such as ”I get very angry when I see someone being ill-treated." A generally positive orien-tation toward others is shown by the reverse- coded item “Lonely people are probably unfriendly." In short, this clus- ter gets at some very basic assumptions about the worth of others and how they should be treated. Persons high in humanistic orientation seem to be manifesting an affection for living things in general, as well as an interest in and concern for, their affective states. A second factor, Considerateness, is somewhat less abstract. These four items measure regard for others that is reflected in a willingness to help (the reverse coded item "when a friend starts to talk about his problems, I try to steer the conversation to something else") and a 32 tendency to adopt the perspective of another (the reverse coded item “It is hard for me to see how some things upset people so much." Where the previous cluster tapped a value, this set of items leans more toward the cognitive, emotional, and behavioral accompaniments of that value. Involvement with Fictional Others was the third factor. The common thread running through these three items was a tendency to become en- grossed in a media experience, i.e., a book or movie. The last cluster of items drawn from the Mehrabian and Epstein (1972) scale was given the name Emotional Contagion. All of the items in this set dealt with the impact of others' emotions on the respondent. The two items which Mehrabian and Epstein offer as defining items for their “Susceptibility to Emotional Contagion" subscale are included in this factor. Of the four "empathy" clusters, this cluster is closest in meaning to the empathy construct developed earlier. The set of items appear to assess the extent to which an individual is prone to make a parallel vicarious emotional response to another person. . The four clusters which were formed from the Emotional Empathy scale appear to be substantively different from one another. The pattern of correlations with other clusters confirms this perception. Inspection of, the intercluster correlation matrix (Table 4) reveals considerable differences in how four clusters correlate with other variables. This is evidence for the multidimensionality of the scale. These results show that treating the Emotional Empathy Scale as unidimensional will blur the causal relations between dimensions of empathy and other variables such as message use. I The two self-consciousness clusters held together fairly well. Al- though some items were dropped from the PRSC scale due to their failure 33 to meet internal or external consistency (See Table 2 for specification of those items), these deletions did not alter the substantive meaning of that cluster. All PUSC items were retained in that cluster. Analysis of the clusters of self-monitoring items extends the find- ings of previous studies. It was necessary to drop five items to purify the Other-Directedness factor (see Table 2 for item numbers). In con- trast, all four items used to form the initial Acting cluster were re- tained. Four items intended to measure extraversion broke into two separate but highly correlated two-item clusters. They were separated on the basis of their differing pattern of correlations with other clusters (external consistency), particularly with Other-Directedness and Acting. One cluster, labeled Social Ability, assesses comfort and skill in social situations. The second cluster was more clearly an Extraversion measure. These items focused on actualrbehaviors in social situations, as opposed O to cognitive or affective states. : Next, two subsamples were formed on the basis of the situational manipulation. Ninety-one subjects responded to the high other benefit situation while 112 received the low other benefit scenario. Missing data reduced these subsamples to 87 and 109 respectively. Clusters analyses were conducted on each subsample with the expectation that the corresponding correlations in each analysis would not differ by more than sampling error. For the most part this expectation was borne out. How- ever, the pattern of correlations with message strategy selection was substantially different between situations (see Table 5). Since this was an indication that situational differences were interacting with indi- vidual differences, the relationships between the trait measures and compliance-gaining strategy selection were examined within each situa- tion. 34 Results of the Analysis of the Theoretical Model Scattergrams were run which plotted each of the ten clusters against message selection in each situation. There was no evidence of a non- linear relationship in any case. Table 5 shows that in the high other benefit condition none of the “empathy” clusters, i.e., Humanistic Orientation, Considerateness, and Emotional Contagion, exhibited nontrivial relationships with message selection. While the Contagion message selection relationship was no different in the low other benefit condition, both Humanistic Orientation and Considerateness showed weak negative correlations. This pattern of correlations suggests two conclusions. First, Emotional Contagion as measured by this cluster of items, does not directly impact strategy selection. Second, Humanistic Orientation and Considerateness may be associated with message selection only under certain conditions. Of the ,two hypothetical scenarios used, Humanistic Orientation and Considerate- ness show sizable relationships with message?selection only in the ”repayment" situation. This scenario has been rated by subjects in previous studies as both low other benefit and high self benefit (Williams a Boster, 1981). The other situation, i.e., "study more," has been rated as high other/ low self benefit. Data reported by Williams .and Boster (1981) show that differences in strategy selection attribut- able to the self-benefit dimension are trivial. However, the impact of the other benefit dimension is not. In fact, those researchers report effect sizes on the order of.: - .28 for other benefit. Thus, although the self/other benefit dimensions are confounded in the present study, it is reasonable to conclude that the differences are likely a result of differences in perception of other benefit. 03 Fictional Involvement exhibited a nonsignificant positive correlation with strategy selection. Since Fictional Involvement did not appear to interact with situation its.£ values were averaged across the two subsamples. This prbvided a more stable estimate of the degree of association by reducing sampling error. The averaging procedure yielded §_- .10, p < .10. When corrected for attenuation due to measurement‘ error the estimate rose to‘ge- .16. This analysis suggests the likeli- hood of a weak tendency for individuals who tend to empathize with characters in books and movies to exhibit greater verbal aggression. . Such individual differences may have implications for measures of com- munication behavior that require the subject to imagine himself in a hypothetical situation. PRSC demonstrated no appreciable covariation with strategy selection in the low other benefit condition. A nonsignificant, negative relation- ship obtained in the high other benefit condition. Thus, the predicted relationship between PRSC and message selection was not obtained. PUSC yielded near zero‘[ values in both situations. The four clusters developed from the Self-Monitoring Scale displayed a highly variable pattern of correlations with message selection. Other- Directedness measure showed a weak negative correlation (5" -.12, p < .20) in the high other benefit condition and a weak positive correlation (g = .16, p < .20) in the low other benefit condition. Since these correlations were an indication of an interaction be- tween Other-Directedness and situation, a regression analysis was con- ducted to formally test for that effect. The results of that analysis revealed a sizeable multiple correlaton.g = .42,.§(3,194) = 13.51, p < .0001, 5.adjusted for shrinkage . .40. Two-tailed tests of significance of the partial slopes showed no effect for Other-Directedness (g = .03, 36 .5 s .49, p 8 .62), a main effect for situation (3 = .93,‘t|= 3.34, p . .001) and a situation X Other-Directedness interaction (9 8 -.56, t_- -2.01, p = .046). This latter finding confirmed the existence of a disordinal interaction, such that Other-Directedness was positively associated with message selection in the repayment situation and nega- tively associated in the study more situation. _ The Acting cluster showed a significant positive correlation with strategy selection (5_- .25) in the high other benefit condition, but a null relationship in the alternate scenario. Thus, self-reported acting ability is positively associated with verbal aggression when the per- suader perceives the outcome of the situation as primarily benefiting the target. This suggests that individuals seeking the compliance of other in this sort of situation may ask themselves whether or not they can per- form convincingly as a persuader. Social Ability exhibited a nonsignificant positive association with message selection in the high other benefit situation (§_- .18, p < .10). This finding suggests that when compliance is seen as having a positive outcome for the target, persons who are comfortable and effective in their social interactions are prone to use a broader range of persuasive strategies than their less graceful counterparts. The corresponding coefficient in the repayment situation was .02. A correlation of .21 was obtained between the two item measure of Extraversion and strategy selection in the low benefit other condition. A near zero value was found in the high other benefit situation. These data show that under conditions of low benefit to the target, persons who view themselves as outgoing are likely to exhibit greater verbal aggres- sion. 37 To test the interaction prediction regression analyses were performed using Emotional Contagion, PRSC, and the product vector of those two terms as predictors. A separate analysis was conducted for each situation. In the study more situation, the full equation yielded an.R of .20,.[(3,81)=1.09, p = .35. The‘R adjusted for shrinkage equaled .06. A similar situation obtained in the repayment situation. The full equation produced an.5 of .09, fj3,101)=.26, p a .86, R_adjusted for shrinkage .00. Both findings indicate no support for the interaction prediction. In summary, the predictions advanced earlier in this paper were not con- firmed., To the extent that Emotional Contagion may be considered an index of empathy, as it was conceptualized earlier in this paper, the expected negative correlation did not obtain. PRSC showed a nonsignificant negative relationship with message selection in the high other benefit condition, but a near zero re- lationship in the alternate scenario. Thus, the anticipated uniform negative relationship did not materialize. An interaction was predicted between empathy and PRSC such that PRSC would magnify the effects of empathy. This prediction .was not supported. The Other-Directedness cluster showed an interaction with situation, thereby disconfirming the main effect prediction. Three other clusters formed from the Mehrabian and Epstein (1972) Emotion- al Empathy Scale were included in the analysis. Two of these, labeled Humanis- tic Orientation and Considerateness exhibited low, but nontrivial, correlations with message selection in the low other benefit condition; when corrected for measurement error they were -.22 and -.32, respectively. While only the co- efficient for Considerateness is significant at the 5% level, the two co- efficients are not significantly different from one another. Another cluster of items drawn from the Emotional Empathy Scale, labeled Involvement with Fictional Others, showed a small positive association with strategy selection across situations. 38 The Self-Monitoring clusters, for which no predictions were made, exhibited some of the highest correlations with verbal aggression in the matrix. Both Acting and Social Ability were positively associated with message selection in the high other benefit situation. Extraversion showed a positive coefficient in the low other benefit situation. Path Analyses A series of path analyses was conducted to explore the relationships among the set of variables. Wicklund's (1975) theory of self-awareness and Buss's (1980) theory of self-consciousness were both drawn on for conceptual guidance. However, it should be emphasized that the analyses conducted here were not tests of either theory. Since a number of models might have fit the data, the value of the models developed in this sec- tion is primarily heuristic. A path model appears in Figure 1. Because that model was tested for each situation, it actually repreSents two models; one for the repayment situation and one for the study more situation. Since data were avail- able for the full sample on each of the individual difference measures, those paths were estimated using 203 observations. The paths to message selection were estimated using the smaller number of subjects who responded to each of the two situations. Ordinary least squares procedures were used to estimate the path coefficients for the two models from a matrix of correlations corrected for attenuation due to measurement error. The method used to evaluate the fit of the models to the data was inspection of the residual matrices. This is formed by calculating a matrix of predicted correla- tions from the observed correlations and the path coefficients. This matrix is then subtracted from the matrix of observed correlations. This 39 yields a matrix of residual correlations which can be used to identify problems in the path model. The residual matrices for each situation are presented in Table 8. Assuming the null hypothesis, = 0, the 95% confidence interval around the full sample correlations is :_.07. For the subsample correlations the 952 con-fidence interval is approximately :_.20. It is clear from inspec- tion of the residual matrices that this model yields predictions which deviate only slightly from the observed values. These deviations may be accounted for by sampling error. Therefore, it may be concluded that the model fits the data. As can be seen from Figure 1, Humanistic Orientation is the sole exogenous variable. The model specifies that this general high regard for others produces an increased awareness of one's own internal workings (PRSC) as well as an appreciation of and willingness to help others. Both PRSC and Consideration are outcomes of this positive regard for humankind. In the case of PRSC this regard is manifested in the form of increased attention to the self. Consideration is a reflection of that same regard for people other than oneself. Self-focused attention has two results. It positively influences involvement in media events. Involvement with fictional others is also enhanced by consideration. It seems likely that perspective taking ability with real people increases the tendency to become immersed in fic- tional characters. The second outcome of PRSC is PUSC. Wicklund's (1975) work in self- awareness bears on this relationship as well as the PRSC-Fictional In- volvement path. The theory of objective self-awareness (Wicklund, 1975) Specifies that the major outcome of self-focused attention is negative affect. This occurs as a result of the individual comparing some aspect 40 of himself against where or what he would like to be. Wicklund suggests that whatever dimension along which an individual examines himself, he will find himself wanting. The recognition of a discrepancy between desired and obtained states produces negative affect. The intensity of the affect is a function of the size of discrepancy. The theory states that the individual is motivated to reduce this state of self-directed negative affect. One way to do this is to attempt to meet the standard. Another is to forget about the discrepancy by focusing attention else- where. Yet another alternative may be to adjust the standard so as to make the discrepancy smaller. Thus, the individual may ask himself “Is this the correct standard?" Given that all standards of behaviors have not yet been set by some government regulatory agency, many remain ambiguous and ill-defined. To facilitate clarification or adjustment of a standard the individual may seek information about what is an appropriate standard from media sources or by adopting the perspective of others toward himself, i.e., viewing himself as a social object. These assertions are.clearly not new to the social sciences. Festinger (1954), in his theory of social comparison, suggested that people often seek information from others about how to evaluate attitude objects when standards are ambiguous. PUSC has emotional and strategic consequences. An awareness of the self as a social object means attending to oneself in relation to others. That is, the interdependence of the actors in a social system is magni- fied. Thus, when one actor displays affect, the emotional effects on the other may be concommitantly magnified. Consideration also positively im- pacts Emotional Contagion. 41 Increased awareness of the self as a social object may lead to the suspicion that others are perceiving the same discrepancy between aspira- tions and achievements that the self-focused person is orienting on. Fur- ther, the individual may assume that everyone else reacts to the discrep- ancy in the same way and evaluates him negatively. This produces a stra- tegic orientation labeled Other-Directedness. It represents a concern with presenting only the agreeable aspects of the self in order that others will judge them favorably. Other-Directedness also taps a willing- ness to deceive others about the self in order to be liked. The concern is for others' evaluations of the self, not others' well-being. Not sur- prisingly then, Consideration negatively impacts Other-Directedness. Con- cern for the welfare of others, as well as attempts to take their perspective, reduces the strategic, instrumental orientation of the Other- Directed individual. 1 The model specifies that message selection is determined by three . forces; Considerateness, Fictional Involvement, and Other-Directedness. Consideration act consistently to inhibit verbal aggression. The direc- tion of influence of Other-Directedness depends on the situation. Fictional Involvement has a uniformly positive effect on message selection. This indicates that persons who become engrossed in fictional events tend to report a tendency to be more verbally aggressive. Although the effects of Fictional Involvement appeared trivial by simple inspection of the zero order correlations, it effects are relatively potent when Con- siderateness and Other-Directedness are accounted for. Other-Directedness caused a mild increase in verbal aggressiveness in the low other benefit situation and a decrease in the high other benefit situation. The key to explaining this finding lies in the reward orienta- tion of the Other-Directed person. In the repayment situation (low 42 other), gaining compliance from the target resulted in $20.00 for the persuader. In the study more situation there were no such payoffs. While convincing the offspring Dick to improve his study habits would benefit Dick, it produces no tangible rewards for the persuader. In fact, the persuader who uses a broad range of strategies in the study more situation runs the risk of ineurring Dick's dislike. Since the high 00 person values “getting along” more than "doing the right thing," his goals are I best met by less verbal aggression. Additional Data and Alternative Models During the time the present study was being written, data of a simi- lar nature was being collected by Boster, Stiff, and Reynolds (1983). Working to establish the validity of the empathy model, these researchers had 257 students respond to Mehrabian and Epstein's (1972) Emotional Empathy Scale and to two compliance-gaining scenarios. One situation was the repayment situation used in the present study. The text of the other situation (hereafter the roommate situation) appears below: You live down the street from a retired couple. Noticing that they have been trying to clean up their yard you decide to help them on Saturday. You also attempt to get your roommate, Lee, to spend next Saturday helping the older couple finish their yard work. Boster and Stiff (1982) made their data available for inclusion in the present study. While their data set does not include PRSC, PUSC, or Other-Directedness, combining the two data sets does permit the test of a reduced model. Since the estimates are made using N = 460, sampling error is greatly reduced. A matrix of correlations among the "empathy“ clusters and message selection (averaged across situations) is presented in Table 7. 43 Several models were attempted (see Figures 4 & 5) using the "empathy" clusters to predict message selection. The model in Figure 4 was the first model tested. It is a reduced version of the original model (Figure 3) with direct paths inserted where variables are missing. This model failed due to deviations greater than sampling error between Fictional In- volvement and Emotional Contagion (.10) and between emotional contagion and message selection (.18). Residuals of that magnitude deviate from chance with a probability of less than .05. Figure 5 shows the next model attempted. That model differs from the previous one only by the inclusion of a path from Emotional Contagion to .message selection. Insertion of that link was not adequate to reduce the discrepancy between the predicted and obtained-Es for Fictional Involvement-Emotional Contagion. Insertion of a path representing Fictional Involvement causing Emotional‘Contagion (Figure 6) was sufficient to fit the model to the data. However, inclusion of so many paths exacts a heavy toll. Mathematically, the model makes only one prediction; it asserts that the correlation between Humanistic Orientation and message selection is zero. Thus, the model is nearly saturated. Search for a Theoretically Meaningful Path Model There are three links in the combined data path model (Figure 6) which were determined by the data rather than theoretically derived: the links from Fictional Involvement and Emotional Contagion to message selec- tion and the link from Fictional Involvement to Emotional Contagion. Since there seemed to be no explanation in terms of direct or indirect causal processes in the specified direction, a search was made for vari- ables that might explain these links as "spurious," i.e., as the result of a common causal antecedent variable. 44 First, it was shown that the data could be fit by introducing a variable that is causally antecedent to Fictional Involvement, Emotional. Contagion, and message usage. The three puzzling links then disappear. The path coefficients from the missing variable are algebraically deter- mined. Second, a search was made for a variable that would be antecedent to both Fictional Involvement and message selection. One plausible variable is need for achievement (McClelland, 1961). However, as will be shown be- low, the theory that explains the linkage from need for achievement to message selection and fictional involvement does not explain the link from need for achievement to Emotional Contagion. Third, a search was made for indirect causal processes which might lead from need for achievement to Emotional Contagion. One possible in- direct path introduces the intervening variable "nervousness" and explains the link from need for achievement to emotional contagion as a two step link from need for achievement to nervousness and from nervousness to Emotional Contagion. The resulting path model is shown in Figure 7. The fit is identical to that of the model in Figure 6 since they are mathematically equivalent. The first step in producing the path model in Figure 7 was to look for a variable which might be a common antecedent to both Fictional Involvement and message selection. This search began by looking for causal deter- minants of Fictional Involvement other than sympathy for others (i.e., Considerateness).i The clue that led to need for achievement was the fact that the theme in most fiction is striving and achievement. People who are high on need for achievement should identify with striving more than those who are low on need for achievement. The link from need for ' achievement to message selection stems from the assumption that people 45 with high need for achievement will be more likely to interpret the persuasion situation as one in which there is a winner and a loser. The person with high need for achievement will then be more likely to counter- balance anticipated negative feelings in the other with anticipated nega- tive feelings in the self stemming from losing. The theory above does not explain the link from Need for Achievement to Emotional Contagion required by the path model. The search for an ex- planation began with an attempt to generate the causal determinants of Emotional Contagion. Since many of the items in the Contagion cluster deal with nervousness, a plausible hypothesis would be that people who are more nervous in general would be more easily influenced by nervousness in others. The link from need for achievement to nervousness follows from the assumption that people with high need for achievement are continually assessing their behavior in terms of winning and losing. Since clear cut evidence of high achievement is rarely present,,this would lead to uncer- tainty and hence to nervousness. Discussion Overall, the findings of this study raise as many questions as they answer. First, the conceptualization of empathy that fits best with the model is not the "tendency to vicariously experience the affect of another" definition suggested by much current literature. A more human- istic definition which includes attitudinal components reflecting a posi- tive regard for others appears most consistent with the model. Second, empathy was shown to play somewhat more limited part in the process of message selection than the empathy model would predict. The data show that empathy should share the stage with individual difference variables such as Fictional Involvement and Other-Directedness. 46 Explanation of these effects introduces hypotheses about other variables such as need for achievement and willingness to manipulate. Empathy A great many empathy researchers adopt a stance which is heavily in- fluenced by learning theory perspectives. Empathy is typically defined as the process of experiencing affect induced by knowledge of the emotional state of another. Most authors contend that there must be some similarity between the emotional states of the observer and the observed (usually the more, the better) to say that empathic processes have been at work. The Emotional Contagion factor which was distilled from Mehrabian and Epstein's (1972) scale, shows the closest correspondence to this defini- tion. However, in the combined data set the correlation between Emotional Contagion and message selection is positive rather than negative. Thus, to identify “empathy” with Emotional Contagion suggests that empathy en- hances verbal aggression rather than inhibits it. Another cluster of items drawn from Mehrabian and Epstein's (1972) scale does behave as the empathy model would predict. Its overall corre- lation with message selection is negative across the three different situ- ations in the combined data sets. That cluster was named Considerateness because its constituent items reflect a tendency to make an altruistic cognitive, emotional, or behavioral response to others. In the four item cluster, one item appeared to be tapping a predisposition to adopt the perspective of another (number 12, Table 3). Two items assessed emotional response (numbers 11 a 13) and another behavioral response (number 10). Thus, Considerateness includes many elements of current definitions of em- pathy. However, its flavor is more value-laden than that of Emotional Contagion. This suggests that another dynamic such as personal standards 47 may be inhibiting aggression rather than egoistic concerns. Recall that the treatments of empathy examined earlier in this paper specified that empathy regulated instrumental aggression because the aggressor experi- enced aversive affect in direct proportion to the amount of suffering he caused the victim. That is, aggression is limited to the extent that it has painful consequences for the aggressor. The negative relationship between Consideration and verbal aggression implies a different process. Rather than individuals restricting their aggression for purely egoistic reasons, the Consideration factor suggests that a more altruistic concern is functioning. All of the path models which fit the data show Humanistic Orientation as a causal antecedent of Consideration. While that alone may be considered minor support, it should be noted that the models which reversed the direction of causal flow were also the models which failed. Thus, while there is support for the humanistic conceptualization of empathy, the more behavioristic alternative is disconfirmed by the data. . Future research might do well to pursue two parallel avenues to im- prove on the measurement model in the present study. First, the develop- ment of additional items which reflect a wider range of responses would add to the reliability and clarity of each of the empathy clusters. Second, existing instrumentation should not be overlooked. For example, Crandall (1975) offers a measure of Adler's concept of social interest which incorporates many aspects of Considerateness. Summarizing Adler's earlier work on personality, Crandall (1980) states that ”...the core of social interest is a valuing of things other than the self. Such valuing is based on the human capacity to transcend the limits of the self and to identify with the needs and concerns of others" (p. 481). He goes on to say that social interest is manifested cognitively, affectively, 48 motivationally, and behaviorally. This construct may represent a more developed notion of the processes and values suggested by Considerateness and Humanistic Orientation. Unmeasured Variables The pattern of correlations between the “empathy" clusters and mes- sage selection suggests the existence of one or more missing variables. One likely candidate, the one included in both speculative path models, is need for achievement, sometimes abbreviated as nAch. This motivational construct was developed primarily by McClelland (1961) and later refined by Atkinson (1964). While nAch may be considered a stable and enduring aspect of personality, it describes a desire for success which may be triggered by situational cues. More formally, the study of nAch is synonymous with the investigation of the persistence, vigor, and frequency of gbal-oriented behavior. Persons high in nAch have been characterized as less concerned with the feelings of others (McClelland, 1965) and as high in a desire to control the environment. These characteristics sug- gest that nAch may causally precede verbal aggression. In addition, the work of de Charms and Mueller (1962) on the relationship between achieve- ment themes in literature and economic striving suggests that nAch may also bring about Fictional Involvement. The second model suggests the existence of another unmeasured variable which mediates the effects of nAch on Emotional Contagion. That variable is nervousness (Guilford a Martin, 1948), defined as a tendency toward excitability. Persons high in nervousness are often characterized as highly charged and frequently seen as apprehensive. Nervousness may plausibly result from nAch, since persons who are goal-oriented spend a good deal of time being concerned about attaining those goals. This state of heightened arousal may in turn bring about a susceptibility to the' moods of others. Although nAch and nervousness were the only constructs included in the models, others are possible. For example, dominance may function similarly to nAch. Dominance might reasonably be expected to bring about increases in verbal aggression. Its relationship with Fictional Involve- ment would be positive by the same reasoning as for nAch: identification with the striving theme in a great deal of literature and visual media. Dominance might also positively influence nervousness. The models developed in this paper which contain missing variables are obviously speculative. They are posed as hypothetical answers to questions advanced by the limited data in the present study. The untested assertions they contain are amenable to empirical tests. Additional research will determine the efficacy of these hypothesized processes. Measured Variables . In one of the most interesting findings of this study, Other- Directedness showed an interaction with situation such that a positive relationship obtained in the repayment situation and a negative relation- ship in the study more situation. It was argued that high 00 persons tend to view compliance-gaining situations in terms of the potential benefit to themselves. As a result they aggressed more in the repayment situation to obtain money. In the study more situation they aggressed less because they were faced not only with the absence of the possibility of any imme- diate physical or social gain, but with the real likelihood of engendering ill will if they chose to employ a wide range of messages. This account I of the interaction effect fits with the picture of the high 00 as a per- ceptive social creature who is unemcumbered by ethical concerns. This 50 interpretation stands in contrast to previous research which has found no effect for self benefit (Williams & Boster, 1981). Private self-consciousness was shown to have only an indirect effect on message selection. The path model constructed from the first data set suggests that the effect of PRSC is mediated by PUSC and Other- Directedness. Given the length of this causal chain, the cumulative im- pact of PRSC on message selection is quite small. It may reasonably be concluded that the PRSC message selection relationship is so weak as to be of no real consequence. Some significant results were obtained for clusters for which no theoretical predictions were advanced. Both Acting and Social Ability showed small positive associations with verbal aggression in the high other benefit situation only. The Acting coefficient, which was the stronger of the two, was interpreted as an indication that people may ask 'themselves whether or not they can perform convincingly in the compliance- gaining situation. Apparently those individuals who are more self-assured about their ability to behave as an actor are also more willing to display a larger degree of verbal aggression. Similarly, persons who are com- fortable and self-confident in social situations also report a tendency to select a broader range of messages. However, these findings are limited to the high other benefit situation. Thus, it may be that some confidence in one's own communicative ability is necessary to argue strongly for a cause that is not one's own, i.e., that primarily benefits the persuadee. This same confidence is unrelated to message selection in the low other benefit conditions because the motivational base is different. In the latter case, the persuader is arguing for his own interests. Thus, he does not need to act to appear convincing, he should be persuasive because it is in his best interests to be so. 5) Conclusions The major conclusion to be drawn from this study concerns the nature of the empathy construct as it relates to message selection.) It has been argued that the conceptualization of empathy as a parallel emotional response has little, if any, predictive or explanatory value in the study of verbal aggression in compliance-gaining situations. Rather, analysis of two different data sets suggested that a factor, labeled Considerate- ness, behaved in a way predicted by the empathy model. This factor con- tained items which tapped perspective-taking and regard for others. Thus, while Considerateness measures aspects of common definitions of empathy it is also a value laden individual difference measure. As such it seems to tap an orientation towards others at the cognitive level. Additional research using improved instrumentation is necessary to clarify this con- struct. The theoretical explanation offered to account for the data relied on .two unmeasured variables. It was suggested that need for achievement has a direct impact on message selection.- This is so because individuals high in need for achievement tend to consider persuasion situations as competi- tive events in which there are winners and losers. Further, it was sug- gested that need for achievement may cause nervousness, due to a lack of clear standards, which in turn renders the individual susceptible to the ‘nervous behavior of others. This two-factor model of message selection, incorporating Considerateness and need for achievement, remains to.be tested. REFERENCE NOTES Hunter, J.E., Cohen, S.H., & Nicol, T.S. PACKAGE: A set of corre- lation and factor analysis programs. Unpublished manuscript, Michigan State University, l975. Boster, F.J., Stiff, J.B., & Reynolds, R.A. 00 persons respond dif- ferently to inductively-derived and deductively-derived lists of compliance gaining message strategies? A reply to Wiseman and Schenck-itamlia. Paper presented at the annual meeting of the International Communication Association, Dallas, May, 1983. Hunter, J.E. & Boster, F.J. An empathy model of compliance-gaining message strategy selection. Paper presented at the annual meeting of the Speech Communication Association, Washington, D.C., November, 1978. & . Situational differences in the selection of compliance-gaining messages. Paper presented at the annual conven- tion of the Speech Communication Association, San Antonio, November, l979. Kaminski, E., McDermott, S., & Boster, F. The use of compliance-gaining strategies as a function of machiavellianism and situation. Paper presented at the annual convention of the Central States Speech Association, Southfield, MI, April, l977. Kraus, R.M., Geller, J., & Olson, C. Modalities and cues in perceiving deception. Paper presented at the annual meeting of the American Psychological Association, Washington, D.C., September, l976. Williams, D.L. & Boster, F.J. The effects of beneficial situational characteristics, negativism, and dogmatism on compliance-gaining message selection. Paper presented at the annual meeting of the International Communication Association, Minneapolis, May l98l. REFERENCES Archer, R.L. Diaz-Loving, R., Gollwitzer, P.M., Davis, M.H. & Foushee, H.C. The role of dispositional empathy and social evaluation in the empathic mediation of helping. Journal of Personality and Social Psychology, 1981, 39, 786-796. Atkinson, J.W. An Introduction to Motivation. New York: .Van Nostrand, 1964. Bandura, A. Social learning theory of aggression. In J.F. Knutsen (Ed.) Control of Aggression: Implicatidns for Basic Research. Chicago: Aldine - Atherton, 1971. Baron, R.A. Magnitude of victim's pain cues and level of prior anger arousal as determinants of adult aggressive behavior. Journal of Personality and Social Psychology, 1971a, 11, 236-243. . Aggression as a function of magnitude of victim's pain cues, level of prior anger arousal, and aggressor-victim similarity. Journal of Personality and Social Psychology, 1971b, 18, 48-54. . Aggression as a function of victim's pain cues, level of prior anger arousal, and exposure to an aggressive model. Journal of Personality and Social Psychology, 1974,129, 117-124, Berkowitz, L. Social norms, feelings, and other factors affecting help- ing and altruism. In L. Berkowitz, Advances in Experimental Social Psychology, Vol. 6. New York: Academic Press, 1972. Briggs, S.R., Cheek, J.M., & Buss, A.H. An analysis of the Self- Monitoring Scale. Journal of Personality and Social Psychology, 1980, 38, 679-686. Buss, A.H. .The Psychology of Aggression. New York: Wiley, 1961.. . Self-Consciousness and Social Anxiety. San Francisco: Freeman, 1980. Carver, C. a Glass, 0. The self-consciousness scale: A discriminant validity study. Journal of Personality Assessment, 1976, 49, 169- 172. Clark, K.B. Empathy: A neglected topic in psychological research. American Psychologist, 1980, §§, 187-190. Clark, R.A. The impact of self-interest and desire for liking on the selection of communicative strategies. Communication Monographs, 1979,.16, 257-273. Coke, J.S., Batson, C.D. & McDavis, K. Empathic mediation of helping: A two-stage model. Journal of Personality and Social Psychology, 1978,‘3§, 752-766. Cronbach, L. Processes affecting scores 6n ”understanding of others" and ”assumed similarity.” Psychological Bulletin, 1955, 52, 177- 193. Cunningham, M.R. Personality and the structure of the nonverbal -communication of emotion. ”Journal of Personality, 1977, fig, 564- 584. Danheiser, P.R. & Graziano, W.G. Self-monitoring and cooperation as a self-presentational strategy. Journal of Personality and Social Psychology, 1982, 42, 497-505. de Charms, R. & Moeller, G.H. Values expressed in children's readers: 1800-1950. Journal of Abnormal and Social Psychology, 1962, 65, 136-142. Deci, E., Nezlek, J. & Sheinman, L. Characteristics of the rewarder and intrinsic motivation of the rewardee. Journal of Personality and Social Psychology, 1981, 59, 1-10. Duval, S. a Wicklund, R.A. A Theory of Objective Self-Awareness. New York: Academic Press, 1972. Dymond, R. A scale for the measurement of empathic ability. Journal of Consulting Psychology, 1949, I}, 127-133. . Personality and empathy. Journal of Consulting Psychology, 1950, 11, 343-350. Eysenck, H.J. & Eysenck, 5.8. Manual for the Eysenck Personality Inventory. San Diego: Educational and Industrial Testing Service, 1968. ' Fenigstein, A., Scheier, M.F., & Buss, A.H. Public and private self- consciousness: Assessment and theory. Journal of Consulting and Clinical Psychology, 1975, 33, 522-527. Feshbach, N.D. Empathy in children: Some theoretical and empirical considerations. Counseling Psychologist, 1975, 5, 25-29. a Feshbach, S. The relationship between empathy and aggression in two age groups. Developmental Psychology, 1969, I, 102-167. 8 Roe, K. Empathy in six and seven year olds. Child Development, 1968, 32, 133-145. Feshbach, S. The function of aggression and the regulation of aggres- sive drive. Psychological Review, 1964, 11, 257-272. 56 Gabrenya, W.K. a Arkin, R.M. Self-Monitoring Scale: Factor structure and correlates. Personality and Social Psychology Bulletin, 1980, _6_, 13-22. Gaertner, S.L. & Dovidio, J.F. The subtlety of white racism, arousal, and helping behavior. Journal of Personality and Social Psychology, 1977,.35, 691-707. Geen, R.G. Perceived suffering of the victim as an inhibitor of attack induced aggression. Journal of Social Psychology, 1970, 8;, 209- 215. Guilford, J.P. a Martin, H.G. Guilford-Martin Inventory of Factors: GAMIN Manual. Beverly Hills: Sheridan Supply, 1948. Hoffman, M.L. Empathy, its development and prosocial implications. In C.B. Keasey, Nebraska Symposium on Motivation, Vol. 25. Lincoln: University of Nebraska Press, 1977, pp. 169-217. Hoffman, M.L. Toward a theory of empathic arousal and its development. In M. Lewis a L. Rosenblum (Eds.) Affect Development. New York: Plenum, 1978. Ickes, W. a Barnes, R.D. The role of sex and self-monitoring in un- structured dyadic interactions. Journal of Personality and Social Psychology, 1977, 35, 315-330. Ilgen, D.R. 8 Knowlton, A. Performance attributional effects on feed- back from superiors. Organizational Behavior and Human Performance, 1980, 25, 441-456. James, W. The Principles of Psychology (Vols. 1, 2). New York: Holt, 1890. 57 Kipnis, D.R., Schmidt, S.M. a Wilkinson, I. Intraorganizational in- fluence tactics: Explorations in getting one's way. Journal of Applied Psychology, 1980, 33, 440-452. Knapp, R.H. N achievement and aesthetic preference. In J.W. Atkinson (Ed.), Motives in Fantasy, Action, and Society. Princeton, NJ: Van Nostrand Co., Inc., 1958. Krebs, D. Empathy and altruism. Journal of Personality and Social Psychology, 1975, 33, 1134-1146. Lippa, R. Expressive control and the leakage of dispositional introversion-extraversion during role-playing teaching. Journal of Personality, 1976,{33, 541-559. Lustig, M. 8 King, S. The effect of communication apprehension and situation on communication strategy choices. Human Communication Research, 1980, 1, 74-82. ‘ Marwell, G. 8 Schmitt, 0. Dimensions of compliance-gaining behavior: An empirical analysis. Sociometry, 1967, 39, 350-364. McClelland, D.C. The Achieving Society. New York: Van Nostrand, 1961. . N achievement and entrepreneurship: A longitudinal study. Journal of Personality and Social Psychology, 1965, 1, 389- 392. , Atkinson, J.W., Clark, R.A. & Lowell, E.L. .153 Achievement Motive. New York: Irvington, 1976. Mehrabian, A. a Epstein, N. A measure of emotional empathy. Journal of Personality, 1972, fig, 525-543. Milgram, S. Some conditions of obedience and disobedience to authority. Human Relations, 1965,‘33, 57-75. 58 Miller, G., Boster, F., Roloff, M., & Siebold, D. Compliance-gaining message strategies: A typology and some findings concerning effects of situational differences. Communication Monographs, 1977, 11, 37-51. & Burgoon, M. Persuasion research: Review and commentary. In B. Ruben (Ed.) Communication Yearbook 2. New Brunswick, NJ: Transaction Books, 1978. Miller, G.R. a Steinberg, M. Between People. Chicago: Science Research Associates, 1975. Mitchell, T. & Wood, R. Supervisor's response to subordinate poor per- formance: A test of an attributional model. Organizational Behavior and Human Performance, 1980, 33, 123-138. Piliavin, I.M., Rodin, J. & Piliavin, J.A. Good samaritanism: _An underground phenomenon? Journal of Personality and Social Psychology; 1969, 1.3.» 289-299. Rarick, 0., Soldow, G., & Geizer, R. Self-monitoring as a mediator of conformity. Central States Speech Journal, 1976,‘31, 267-271. Rawlings, E.I. Reactive guilt and anticipatory guilt in altruistic be- havior. In J.R. Macaulay & L. Berkowitz (Eds.) Altruism and Helping Behavior. New York: Academic Press, 1970, pp. 163-177. Roloff, M.E. Self-awareness and the persuasion process: Do we really know what we're doing? In M.E. Roloff and G.R. Miller (Eds.) Persuasion: New Directions in Theory and Research. Beverly Hills, CA: Sage, 1980. Rule, 8.6. a Nesdale, A.R. Emotional arousal and aggressive behavior. Ppychological Bulletin, 1976,.33, 851-863. 59 Scheier, M.F. Self-awareness, self-consciousness, and angry aggression. Journal of Personality, 1976, 99, 627-644. 6 Carver, C.S. Self-focused attention and the experience of emotion: Attraction, repulsion, elation, and depression. Journal of Personality and Social Psychology, 1977, 33, 625-636. , , Schulz, R., Glass, D.C., a Katz, 1. Sympathy, self-consciousness, and reactions to the stigmatized. Journal of Applied Social Psychology, 1978, 3, 270-282. , Fenigstein, A. a Buss, A.H. Self-awareness and physical aggression. Journal of Experimental Social Psyghology, 1974, 39, 264-273. Sears, R.R., Maccoby, E.E. a Levin, H. Patterns of Child Rearing, New York: Holt, Rinehart, & Winston, 1958. Sillars, A. The stranger and the spouse as target persons for compliance-gaining strategies: A subjective expected utility model. Human Communication Research, 1980,‘9, 265-279. Snyder, M. Self-monitoring of expressive behavior. Journal of Personality and Social Psychology, 1974, 39, 526-537. & Monson, T. Persons, situations, and the control of social behavior. Journal of Personality and Social Psychology, 1975, 33, 637-644. Snyder, M. Self-monitoring processes. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology, Vol. 12. New York: Academic Press, 1979. Staub, E. Helping a distressed person: Social, personality, and stimu- lus determinants. In L. Berkowitz, Advances in Experimental Social Psychology, Vol. 7. New York: Academic Press, 1974. 6O Stotland, E. Exploratory investigations of empathy. In L. Berkowitz (Ed.) Advances in Experimental Social Psychology, Vol. 4. New York: Academic Press, 1969. Tedeschi, J., Schlenker, B., & Lindskold, S. The exercise of power and influence: The source of influence. In J. Tedeschi (Ed.) .193 Social Influence Process. Chicago: Aldine, 1972. Tobey, ELL. & Tunnel, G. Predicting our impressions on others: Effects of public self-consciousness and acting, a self-monitoring sub- scale. Personality and Social Psychology Bulletin, 1981, Z, 661- 669. Wicklund, R.A. Objective self-awareness. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology, Vol. 8. New York: Academic Press, 1975. & Duval, S. Opinion change and performance facilitation as a result of objective self-awareness. Journal of Experimental Social Psychology, 1971, 1, 319-342. 6T .aupprnoppme Pecemucp m~Pspxme cu woumpee swamp Emuun .eouumu one easy egos :o umumo_ mema_ cozy—emucze mcopaspom mmegu o~.m~.m.m ma m~.-.¢~.NH co ceased msmu_ m~.m~.afi.e~ mec.a=_om o3» o~.mfi.w.m .mH.mfi.a.~.m.~ m~.-.-.¢~.~H cu ceased asaoH «on u z a a a a a a a “Naafi Afim@fiv -uinrui ow m~ mg m m o~ mg Ma na.~ m~.-.-.¢~.md poccap a amnoh amm.c¢~.ou¢.o- u z Aomafiv a.m.¢.~ o~.wfi.m.o.m 8H.AH.8.K.m m~.-.¢_.- =.xa< a aacaeaaa .1. mam.omm u z in. m~.m~.a~.a~ .i. .I. Acmaflv ------- ¢~.c~.w~.m.m .e~.mfi.mH.K.m...~ m~.-.- ac~.¢~.~# ._a co mom.am »B.P.a< m=anaam ac.uo< mmaceauuaaeo-aaeuo »u__.aa_oom \copmem>meuxm husum meouuom Am_pwu :.;up3 Leanna memae=c emuwv m—mum mcpcouecoZTEPom m.emu»:m mo meauuseum poceoacp we» umcvsmxo cope: mmpvzum mean» we mapsmee ugh H m—awh 62' Table 2 Results of the cluster analysis Original Scale Cluster Itemsa Emotional Humanistic 19,1,15,30, Empathy Orientation 14,27,2,16,29 (Mehrabian & Epstein, 1972) Considerateness 21,23,26,22 Involvement with 31,18,28 Fictional Others Emotional 32,20,9,25, Contagion 13,5,10 Residual 3,4,6,7,8,11, 12,17,24,25,33 Self-Consciousness Private 20,15,5,18,13,7 (Fenigstein, Scheier, ' & Buss, 1975) Public 14,11,6,19,21, Residual 1,3,9,22 Self-Monitoring Other-Directedness. 13,19,16,25,17 (Snyder, 1974) Acting 8,20,18,5 Social Ability 23,14 Extraversion 12,22 Residual 1,2,3,4,6,7,9, 10,11,15,21,24 aItems numbers are taken from the article in which the scale first ap eared. They are presented in descending order of the magnitude of the corre ation of the item with its clusters after reflection of negatively worded items. 63 Table 3 Factor loadings (one side corrected for attenuation), means and standard deviations for each item retained in the measurement model. Items requiring reflection are followed by (R). Humanistic Orientation 1. I get very angry when I see someone being ill-treated. 2. It makes me sad to see a lonely stranger in a group. 3. I am very upset when I see an animal in pain. 4. Lonely people are probably unfriendly. (R) 5. I become more irritated than sympathetic when I see someone's tears. (R) ' 6. I like to watch people open presents. 7. People make too much of the feelings and sensitivities of animals. (R) 8. Seeing people cry upsets me. 9. It upsets me to see helpless old people. Considerateness 10. When a friend starts to talk about his problems, I try to steer the conversation to something else. (R) 11. Sometimes at the movies I am amused by the amount of crying and sniffling around me. (R) 12. It is hard for me to see how some things upset people so much. (R) 13. Another's laughter is not catching for me. (R) Involvement with Fictional Others 14. 15. 16. I become very involved when I watch a movie. I really get involved with the feelings of the characters in a novel. Becoming involved in books or movies is a little silly. (R) Loading, '7 sd .58 3.4 .69 .56 3.2 .65 .49 3.2 .77 .45 2.6 .58 .45 2.4 .64 .41 3.3 .65 .46 3.2 .79 .39 2.8 .80 .35 3.3 .70 .61 2.4 .61 .51 1.7 .78 .47 1.6 .82 .44 2.3 .66 .60 3.0 .66 .56 2.9 .81 .57 2.3 .64 64 Table 3 (Cont.) Emotional Contagion 17. 18. 19. 20. 21. 22. 23. I often find that I can remain cool in spite of the excitement around me. (R) I tend to lose control when I am bringing bad news to people. I am able to remain calm even though those around me worry. (R) ‘ I cannot continue to feel OK if people around me are depressed. I don't get upset just because a friend is acting upset. (R) I become nervous if others around me seem to be nervous. The people around me have a great deal of influence on my moods. Private Self-Consciousness 24. 25. 26. 27. 28. 29. I'm alert to changes in my mood. I'm constantly examining my motives. I reflect about myself a lot. I sometimes have the feeling that I'm off somewhere watching myself. I'm generally attentive to my inner feelings. I'm often the subject of my own fantasies. Public Self-Consciousness 30. 31. 32. 33. I usually worry about making a good impression. I'm self-conscious about the way I look. I'm concerned about the way I present myself. I'm concerned about what other people think of me. Loading 7' sd .62 1.4 .72 .54 2.2 .79 .55 1.2 .74 .46 2.6 .66 .44 1.2 .71 .43 2.3 .72 .42 2.9 .72 .56 3.3 .76 .55 2.7 .85 .52 3.0 .83 .43 2.0 .95 .38 3.3 .67 .35 2.7 .98 .71 3.0 .88 .69 3.2 .84 .63 3.4 .71 .58 3.1 .78 65 Table 3 (Cont.) 34. I'm usually aware of my appearance. 35. One of the last things I do before I leave the house is look in the mirror. 36. I'm concerned about my style of doing things. Other Directedness 37. In different situations and with different people, I often act like very different persons. 38. In order to get along and be liked, I tend to be what people expect me to be rather than anything else. 39. I'm not always the person I appear to be. 40. I may deceive people by being friendly when I really dislike them. 41. I would not change my opinions (or the way I do things) in order to lease someone else or win their favor. (RI Acting 42. I would probably make a good actor. 43. I have never been good at games like charades or improvisational acting. (R) 44. I have considered being an entertainer. 45. I can make impromptu speeches, even on topics about which I have almost no information. Social Ability, 46. I feel a bit awkward in company and do not show up quite so well as I should. (R) 47. I am not particularly good at making other pe0ple like me. (R) Loadingy 'X sd .51 3.4 .58 .45 2.8 1.0 .36 2.9 .98 .64 2.5 .99 .50 1.9 .82 .56 2.5 .95 .38 2.5 .91 .27 1.1 .78 .71 2.5 1.0 .62 2.0 .92 .56 2.1 1.2 .36 2.3 .97 .67 2.0 .85 .67 2.4 .70 66 Table 3 (Cont.) Loading 7 sd Extraversion 48. In a group of peOple I am rarely the center of attention. (R) .60 1.6 .81 49. At a party I let others keep the jokes and stories going. (R) . .60 1.6 .88 67 Table 4 Matrix of corrected intercluster correlations with Alpha reliabilities in the diagonal. 1 2 3 4 5 6 7 8 9 10 11 1. Humanistic Orientation 71 2. Considerateness 50 58 3. Fictional Involvement 27 56 60 4. Emotional Contagion 37 48 3O 69 5. Private S-C 28 12 36 38 62 6. Public S-C 14 15 23 47 56 76 7. Acting 18 6 26 -12 23 28 58 8. Other- Directedness -21 -26 -01 17 24 4O 03 65 9. Social Ability 13 21 21 -20 -22 -9 36 -38 60 10. Extraversion 14 16 8 -20 -10 -01 60 10 75 51 11. Gender (male = 1, female = 2) 31 56 27 30 19 18 10 -14 15 10 100 12. Message Selection (repayment) '-22 -32 16 -O3 -05 04 05 23 02 33 06 13. Message Selection (study more) 14 02 15 -02 -21 O9 38 -17 28 04 17 14. Message Selection (average) -04 -15 15 -2 -13 O7 22 O3 15 16 12 -03 68 Table.5 Correlations between the trait measures and message selection (r columns are corrected for measurement error) Situation Study More . Repayment Variable r r r r Humanistic 14 11 ~22 ~15 Orientation Considerateness 02 01 ~32 ~25** Fictional In- 15 09 16 10 volvement Emotional ~02 ~02 ~03 ~02 Contagion PRSC ~21 ~15 05 05 PUSC O9 05 O4 O3 0ther~ ~17 ~12 23 16 Directedness Acting as 25** 05 05 Social 28 18* 02 02 Ability Extraversion 04 03 33 21** N a 87 N = 109' Ho:r=o,p(~.21