MSU RETURNING MATERIALS: Place in book drop to remove this checkout from _;::::::::L_ your record. FINES wil] be charged if book is returned after the date a. y 1.. ' stamped below. \”“ A-,.u it??? 9537§€9 7 r t»: , ‘9" “4 MAGS? (Q9 FEM é. “" EFFECTS OF VERBAL AND NONVERBAL AFFILIATION ON INITIAL INTERACTIONS BETWEEN HANDICAPPED/ NONHANDICAPPED PERSONS DURING EMPLOYMENT INTERVIEWS BY Gordon L. Dahnke A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication 1982 61/7787 ABSTRACT EFFECTS OF VERBAL AND NONVERBAL AFFILIATION ON INITIAL INTERACTIONS BETWEEN HANDICAPPED/ NONHANDICAPPED PERSONS DURING EMPLOYMENT INTERVIEWS BY Gordon L. Dahnke Purpose of this discourse is development and use of effective communication strategies to address the pressing problem of unequal employment opportunities for physically handicapped persons. It represents an initial attempt to build substantive, deductive theory from which to conduct empirical research focusing on dyadic interactions between handicapped/nonhandicapped persons. Based on current developments in interpersonal communication, a first approx- imation of axiomatic theory and a recursive model are pro- posed, ad0pting an uncertainty reduction paradigm. Fifteen theorems deduced from the theory were tested via exploratory research in an actual employment interview context. Four verbal and four nonverbal behaviors were manipu- lated simultaneously at three levels of congruency to form a composite index of the primary independent variable, ver- bal/nonverbal affiliation. Two types of dyadic interaction, nonhandicapped/nonhandicapped and handicapped/nonhandicapped, comprised the second independent factor, suggestive of a two-by-three factorial design. Two—way ANOVA procedures were used to assess main and conjoint effects of these Gordon L. Dahnke predictors on five independent variables: communicative certainty, state anxiety, interpersonal attraction, per- ceived interpersonal attraction and perceived state anxiety. One hundred and forty-four participants were exposed to a videotape record of the stimulus message during one hour, standardized employment interviews. Four of five theorems predicting a main effect from verbal/nonverbal affiliation received full or partial support. Predicted linear effects attributable specif— ically to the presence of a handicapped person in dyadic contexts did not typically occur. Several conjoint effects emerged contrary to prediction, indicating that relations among these variables are more complex than the linear associations hypothesized. Some evidence was found suggesting that nonhandicapped receivers do not respond in the same manner to incongruent verbal/nonverbal messages in general or nonverbal affiliative cues in particular when the source is a handicapped versus nonhandicapped counterpart. Accepted by the faculty of the Department of Communication, College of Communication Arts and Sciences, Michigan State University, in partial fulfillment of the requirements for the Doctor of Philosophy degree. .fiflmfid fl W Director of Dissertation Guidance Committee: flQ/(flE/Q flgbdégfl’ , Chairman *— / Kira-‘14,: T £31." I: V flifi/ [Zyd‘gcigfl To ones I love, whose loving sacrifice I cannot requite . . . Dottie Debra Jeff Bruce ii ACKNOWLEDGMENTS Indeed, . . . no man stands alone! Several persons deserve more than this page can provide for their indelible marks on my person and education. If no man is an island, Dr. Gerald Miller is a mountain . . . of intellectual acumen and professional achievement. Most graduate stu- dents need an academic hero; few are privileged to sit at the feet of Gamaliel; none have had a better example to emulate. Dr. Miller, I am proud to have been your student, advisee and friend. Also, thanks to you and Pearl Ann for the pleasantries of home and company. To the extent that this work reflects quality, it reflects the spartan influence and tireless patience of Dr. Norman Fontes. It is difficult to state, more so to overstate, my indebtedness to Norm. I thank the Department of Communication for providing the resources that made this work possible and the person singly, most responsible for its academic excellence, Dr. Erwin Bettinghaus. His knowledgeable commentary made the difference between an acceptable dissertation and publish- able research. In a similar vein, I am indebted to Dr. Judee Burgoon who conscientiously poured over these pages and lengthy preliminary papers, often on short notice at the expense of private time. I have learned from her iii critical insights on methodology, logic and style. Special thanks are due Dr. William Crano, Department of Psychology, for his willingness to serve on my commit- tee. As scholar, teacher and person the man has class. This project might never have been completed in lieu of his positive comments and encouragement during preliminary examinations. I regret that he was unable to serve the duration due to a well—deserved leave of absence. Dr. Keith Adler did what you would hesitate to ask, except of a trusted friend. Thank you, Keith, for stepping in at the last moment to bring this marathon to closure. We will not forget you and Carol. Today, as I write, retired Ruth Langenbacher is typing to help meet a deadline. What can I say, Ruth, to let you know how much I appreciate your skillful arts and crafts these many years. My grandchildren will hear of you. We could not have done it without your help. It has been difficult to order these acknowledgments. As all will know, the last wine is not always least. Dr. Arnold Schuring, lifelong friend and surgeon, deserves the utmost credit for this accomplishment. He believed when I wouldn't; he called when I couldn't; he gave when a man shouldn't! Arn, you know my sentiments. Many thanks to you and your otolgic group for financial and other more important means of support undergirding this research. I am deeply grateful to all of you. iv TABLE OF CONTENTS Chapter Page I THE THEORY . . . . . . . . . 1 Introduction. . . . . . . . l The Domain . . . . . . . . 8 Behavioral Determinants of Social Interaction . . . . . . . 22 Theoretic Moorings . . . . . . 29 Testable Theorems . . . . . . 64 II THE METHOD . . . . . . . . . 74 The Independent Variables . . . 74 Nonverbal Indicators of Affiliation . . . . . 75 Verbal Indicators of Affiliation . . . . . 77 The Dependent Variables . . . . 84 The Design . . . . . . 93 An Overview of Procedures . . . 95 The Standard Stimulus Interview . . 98 .The Videotaped Stimuli . . . . 99 The Live Interview . . . . . 102 The Questionnaire . . . . 111 Recruitment and Selection . . . 113 A Summary . . . . . . . 116 III THE RESULTS . . . . . . . . . 118 The Sample . . . . . . 118 Assessment of Experimental Manipulations . . . . . 123 Factor Analyses on Dependent Measures . 132 Tests of Theorems . . . 142 Effects on Communicative Certainty . 142 Effects on State Anxiety. . '. 145 Effects on Interpersonal Attraction . 152 Effects on Perceived Interpersonal Attraction. . . . . . 158 Effects on Perceived State Anxiety . . 161 Chapter Page IV DISCUSSION . . . . . . . . . 168 Hypothesized Relations . . . . 169 Significant Results Contra Prediction . . . . . . 176 Nonsignificant Trend . . . . 181 Limitations. . . . . . . 187 Directions For Future Research . . 192 APPENDICES . . . . . . . . . . . 194 A. Standardized Employment Interview . . . 194 B. SRA Employment Application . . . . . 205 SRA Employment Interview Question- naire, Part I . . .j . . . . 206 SRA Employment Interview Question- naire, Part II . . . . . . . 218 BIBLIOGRAPHY . . . . . . . . . . 219 vi Table l. 10. ll. 12. LIST OF TABLES Nonverbal and Implicit Verbal Behaviors Defining Affiliation, a Primary Dimension of Social Interaction. . . . . . Means Resulting From Manipulation Check Data-Ratings of Sociability . . . . Means Resulting From Manipulation Check Data-Ratings of V/NV Congruency . . . Means Resulting From Manipulation Check Data-Ratings of Source Credibility . . Factor Loadings - Communicative Certainty Scale . . . . . . . Factor Loadings - State Anxiety Scale. . Factor Loadings - Interpersonal Attraction Scale . . . . . . . Factor Loadings - Perceived State Anxiety Scale. . . . . . . . Factor Loadings - Perceived Interpersonal Attraction Scale . . . . . . . Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affili- ation and Interaction Type on Commun- icative Certainty. . . v . . Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affili- ation and Interaction Type on State Anxiety - Factor 1 . . . . . . Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affili- ation and Interaction Type on State Anxiety - Factor 2 . . . . . . vii Page 27 125 128 129 135 136 138 140 141 144 148 150 Table Page 13.. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affili— ation and Interaction Type on State Anxiety - Factor 3 . . . . . . . 152 14. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affili- ation and Interaction Type on Inter- personal Attraction - Social Attraction Scale. . . . . . . . . . . 154 15. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affili- ation and Interaction Type on Inter- personal Attraction - IJS Scale . . . . 157 16. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affili- ation and Interaction Type on Perceived Interpersonal Attraction . . . . . . 160 17. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affili- ation and Interaction Type on Perceived State Anxiety - Factor 1 . . . . . . 164 18. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affili- ation and Interaction Type on Perceived State Anxiety - Factor 2 . . . . . . 167 viii Figure 1. LIST OF FIGURES Flow Chart Illustrating a Logical Model of the Theory Proposed to Represent Initial Interactions Between Handicapped/Nonhandicapped Persons From an Uncertainty Reduction Perspective . . . . . . Representation of a Factorial Design Crossing Three Levels of Verbal/Non- verbal Affiliation With Two Types of Dyadic Interaction. . . . . . . Pattern of Means Showing Rank Order of Effects and Interactions Predicted in Theorems 11, 14 and 15 . . . . . Pattern of Means Showing Rank Order of Effects and Interactions Predicted in Theorems 12 and 13 . . . . . . Graphic Representation of a 2 x 4 Factorial Design Crossing Four Levels of Verbal/Nonverbal Affiliation With Two Types of Dyadic Interaction . . . Graphic Representation of Communicative Certainty as a Function of Affiliation and Type of Dyadic Interaction. . . . Graphic Representation of Interpersonal Attraction as a Function of Affiliation and Type of Interaction . . . . . Graphic Representation of Perceived Interpersonal Attraction as a Function of Affiliation and Type of Interaction. . Graphic Representation of Perceived State Anxiety as a Function of Affili- ation and Type of Interaction - Factor 1 . ix Page 41 70 73 73 96 146 156 162 166 CHAPTER I THE THEORY Introduction A recent survey conducted by the President's Committee on Employment of the Handicapped indicates that one of every eleven Americans has some disability. These persons presently constitute a minority group subject to adverse educational, economic and employment discrimination. A significant segment of this minority is involuntarily rele- gated to a dependency role in which they become wards of the state. The plight of the handicapped is exacerbated by attitudinal factors which impede meaningful social dis- course with nonhandicapped persons, and communication problems tend to restrict their complete integration into the productive mainstream of society. Recent literature addressing problems encountered by physically handicapped persons (e.g., Goffman, 1963; Comer & Piliavin, 1975; Park, 1975) differentiates physical and psychological barriers which inhibit such individuals during interactions with nonhandicapped persons. Physical barriers such as architectural impediments are relatively easy to remove. Psychological barriers may arise in: (1) the handicapped person's perception of him/herself; (2) the handicapped person's perception of nonhandicapped counter- parts; and (3) the nonhandicapped person's perception of l 2 the handicapped. These mutual perceptions appear to influ- ence the manner in which handicappers and nonhandicapped persons interact (Goffman, 1963). Their interactions may, in turn, reinforce or alter existing attitudes of both parties. Similar effects may also accrue from a decision not to interact. Two general research foci are apparent in the litera- ture: (1) assessment of attitudes held by each group of persons, and (2) investigation of the interaction process between these two individual types. Review of the litera- ture suggests that considerable attention has been devoted to the former issue. For example, extensive attitudinal studies by Siller, Chipman, Ferguson and Vann (1967) con- clude that current attitudes toward handicapped persons con- tribute to a serious problem involving "strain in social interaction" during fact-to-face encounters. The strain may result from both participants being uncertain about the most appropriate communication strategy to employ in order to maximize the quality of their interactions. The impli- cation of this research is that "reduction of the 'sticky' problems in initial and subsequent encounters would be a major help in reducing anxiety about dealing with the handicapped" (Siller 33 21,, 1967:65). Regarding the interaction process more limited research to date indicates that during interactions with handicappers, nonhandicapped persons tend to: (1) increase their physical 3 distance (Kleck, 1969; Worthington, 1974); (2) terminate the interaction more quickly and exhibit less variance in verbal behavior (Kleck, Ono & Hastorf, 1966); (3) demon— strate greater motoric inhibition, i.e., restrict their nonverbal behavior (Kleck, 1968), and (4) exhibit more anx- iety (Marinelli & Kelz, 1973; Marinelli, 1974). Furthermore, nonhandicapped individuals report more emotional discomfort following interaction (Kleck, 1966). Four of eleven different areas suggested for future research by Richardson (1976) concern the problem of strain between handicappers and nonhandicapped persons during social interaction. Specifically, little is known about factors influencing various behavioral distortions which develop on the part of both participants during such inter- actions. Moreover, it is intuitively reasonable to expect that identification of specific anxiety-producing (or reducing) communication behavior which may originate with either party would be a significant first step toward elimi- nation of these problematic distortions. Siller gt 31. (1967) report anecdotal evidence indicating the heuristic promise of this approach. Among the host of relevant variables, few communication scholars will deny that interpersonal transactions are largely dependent upon the effectiveness of verbal/non- verbal cues exchanged. Furthermore, certain nonverbal cues (e.g., illustrators, regulators, and affect displays) are 4 believed to be closely associated with, and auxiliary to, linguistic behavior (Harrison, 1974; Ekman & Friesen, 1969). Under normal circumstances these nonverbal counterparts of language may serve to repeat, substitute for, complement, accentuate, contradict or regulate verbal communication (Burgoon & Saine, 1978). Assuming their validity in this context, these assertions bear serious implications for physically handicapped persons who may also be communica— tively handicapped particularly in the nonverbal realm, i.e., they may be unable to effectively encode appropriate nonverbal cues as a result of impairment. Consequently, handicappers with normal verbal capabilities may frequently display nonverbal behaviors which are perceived to be incon- gruous with their verbal communication. Such verbal versus nonverbal discrepancies, as well as the sheer absence of supporting nonverbal cues in some instances, may be major determinants of communication problems arising when handi- capped/nonhandicapped persons interact. Previous research indicates that visible impairments per se tend to provoke anxious reactions from nonhandicapped interactants (Marinelli & Kelz, 1973; Marinelli, 1974). This anxiety may be manifest via the verbal (Dibner, 1956) or the nonverbal (Mehrabian, 1969) channel of communication. For example, the nonhandicapped individual may exhibit verbal nonfluencies, nervous fidgeting, anxious facial affects, restricted body movements, and so on. 5 Given the transactional nature of human communication, anxiety revealing cues and/or discrepant verbal/nonverbal messages emitted by either interactant may be perceived as betraying the source's ambivalent emotional state, negative social evaluations, or communicative uncertainty toward the counterpart. In turn, these perceptions are expected to affect the level of ambivalence, social evaluation and un- certainty experienced by the counterpart and, thus, to elicit a similar affect-revealing response. Ongoing trans- actions of this type may result in an escalation process that continues until a termination threshold has been reached by one or both interactants. Theoretically, such an escalation process provides an explanation reasonably consistent with several previous findings (e.g., Kleck, Ono & Hastorf, 1966; Kleck, 1968, 1969; Worthington, 1974; Marinelli, 1974). Obviously this line of reasoning assumes, for example, that discrepant verbal/nonverbal behaviors are perceived and interpreted by both participants in the general manner described above. It further suggests that the exhibition of negative or ambiguous communicative behaviors is more probable, at an accelerated pace, in interactions with handicapped persons who are restricted in their ability to encode nonverbal cues. As a result of atypically high uncertainty, it is also conceivable that handicapped/nonhandicapped interactants tend to monitor the nonverbal channel more intensely than 6 the verbal, in order to assess the emotional states of the counterpart. While there is general support for the domi- nant impact of nonverbal information particularly when dis- crepant messages occur simultaneously (e.g., Argyle, Alkema & Gilmour, 1971; Fujimoto, 1971; McMahan, 1976), nonverbal messages are not always pre-eminent. Situational factors, personality types, as well as the type and degree of message components apparently tend to complicate this issue (e.g., Fujimoto, 1971; Shapiro, 1968; VandeCreek & Watkins, 1972). To date, little empirical research has attempted to determine which major channel exerts the greater impact on transactions of the present type. Knowledge of the verbal/ nonverbal relationship and the extent to which these chan- nels differ as salient sources of information would likely prove invaluable for development of effective communication strategies in the context of handicapped/nonhandicapped interactions. More specifically, the ensuing discussion anticipates development and use of such strategies to address the pressing problem of unequal employment oppor- tunity for handicapped persons. Two methodological impulses that guided the conduct of this research deserve comment. First, it should not be necessary to argue the general need for substantive, deduc- tive theory in our discipline. Certainly not in this area. Social scientists are well aware that facts do not explain themselves. If observations are to be interpretable, sound 7 theoretic information is necessary (Heise, 1975). More- over, there is an obvious need for testable theory which is sufficiently complex to provide informative insights that supercede those gleaned from common sense and practi- cal experience. Hence, this paper represents an initial attempt to build theory defined as a set of interrelated "systematically organized, lawlike propositions" that can be supported or falsified via empirical evidence (Zetterberg, 1965:22). Second, the attempt to build sufficiently complex yet testable, deductive theory poses a serious problem partic- ularly in the context of a young science. To develop such theory, the builder must begin with relatively simple sets of relationships that serve as first approximations of deductive theory. Typically, the model includes only a few relevant variables (at best) and is far too simple to represent the complexities of the real world under study. Moreover, at this stage it is extremely difficult not to prioritize those variables and relationships that are more easily observed or tend to "fit" conveniently into the simple theory while eliminating relevant others that do not (Blalock, 1969). The theoretical statement to follow attempts a happy medium between the goals of complex- ity and testability. It is a first approximation of deductive theory that invites reformulation toward more realistic representations of the real world via inductive research processes. Given prior seminal ideas, the subsequent theoretic formulation addresses this general research question: In an initial employment interview, what effects do varying levels of verbal and nonverbal communication emitted simultaneously by handicapped interviewees, have on the nonhandicapped interviewer's (1) commun- icative uncertainty, (2) state anxiety level, (3) interpersonal attraction toward the interviewee, (4) perception of the interviewee's state anxiety level, and (5) perception of the interviewee's interperson- al attraction toward the interviewer? While present interests focus on an employment interview context, the theoretic statement developed to address this question is sufficiently general to apply to any initial encounter between handicapped and nonhandicapped persons. ~é-«a The Domain This discourse presupposes an interdependent relation- ship between verbal and nonverbal components in ordinary human communication. It does not imply a subordinate role for either channel, nor deny that each can function effec- tively alone. Their ordinary interdependency does imply that selection of appropriate variables for comparison across the two channels will depend, in part, on a con- ceptual view of communication theoretically capable of con- sistent application in both verbal and nonverbal areas. Communication per se, in distinction from other human behavior, minimally requires (1) a socially shared signal system, that is, a code; (2) an encoder actively making 9 some experience public via that code; and (3) a decoder able to respond systematically to the same code (Wiener, Devoe, Rubinow & Geller, 1972; Burgoon & Saine, 1978). While this definition provides a useful point of departure, it leaves several key questions unanswered particularly in the nonverbal area. First, the sign versus symbol issue. For some scholars signs are distinguished as a natural extension of the events they herald. Thus, the occurrence of cumulus clouds is considered a £192 of impending rain, or a flushed face may naturally signal an underlying emotional state. On the other hand, symbols have arbitrarily assigned, socially shared referent meanings. While verbal symbols such as "cumulus cloud" are clearly understood by English speakers, their association with the real event is neither natural nor necessary. Nonverbal symbols are believed to serve essentially the same communicative function. Although com- plex differences between verbal/nonverbal systems are be- coming increasingly apparent, these distinctions are not relevant here. Another approach allows the term sign to represent any attribute or human behavior from which an observer can infer some significance or glean any information about the source. Presumably, the term sign may then include such diverse and idiosyncratic behaviors as lint picking, head scratching or weight watching activity. The nonverbal 10 literature is replete with similar examples. Signs, in this latter sense, do not qualify as communicative events by present definition. While they may be informative in some broad sense, idiosyncratic behaviors and more general stimuli serving only as a basis for inference are excluded here. However, the pathological interest of this paper recommends inclusion of nonverbal signs as well as symbols which meet either of two criteria: (1) natural behaviors that reflect underlying emotional states, provided they are known to bear socially shared meanings and (2) learned be- haviors that reflect the enculturation process and are known to bear socially shared referent meanings. Second, intentionality in the nonverbal area is equal- ly problematic. Who decides what is intentional and the required level of consciousness involved are closely related, complicating issues. Several prominent investi- gators (e.g., Ekman & Friesen, 1969; Wiener g£_al., 1972) insist on conscious intent by the message source as a pre- requisite for communicative events. Others are willing to relax this stipulation to include messages that are per- ceived as intentional by either the receiver, an outside observer, or the source (Miller & Steinberg, 1975). Still others accept an even broader view involving messages that are either (1) subconsciously or unintentionally sent, if interpreted as intentional by any observer, or (2) subcon- sciously received by an observer, if consciously and intentionally sent (Burgoon & Saine, 1978). 11 This already complex issue is exacerbated in the con- text of interactions with handicapped persons who may be unable, due to physical impairment, to encode their inten- tions apprOpriately via the nonverbal steam. Previous research suggests the extraordinarily high information potential of affect-revealing cues when handicapped/non- handicapped persons interact. Nonverbal signs of this type, whether natural or learned, which may occur at a relatively low level of awareness are quite germane to present concerns. The following discussion is a modest attempt to avOid some of the confusion present in extant nonverbal literature by establishing clear conceptual parameters for the class of events under study here and the terminology used to describe it. Wiener, Devoe, Rubinow and Geller (1972) expose con- ceptual problems associated with the now prevalent decoding perspective and argue convincingly for greater emphasis on the encoding subject in current conceptualizations of non— verbal communication. In that regard, the authors offer a useful distinction between informative nonverbal behavior which may be emitted at a relatively low level of awareness and nonverbal communication which they define as a more dis— tinctly intentional act by the message source. Although use of the term behavior is eventually elimi- nated as a label for informative events here, the prior con- ceptual distinction proposed by Wiener 31 a1. (1972) 12 suggested that theory construction involving both verbal and nonverbal communication might ideally begin at a higher level of abstraction, using a larger generic category that will allow consistent classification of informative events across the two channels. That class of events, to be defined momentarily, is labeled information transfer. The more familiar term, information exchange, would have been preferred. However, as the introductory comments warn, it is imperative to begin with simple models. The term exchange implies a dynamic, nonrecursive process with which the present formulation is not designed to deal. The genus labeled information transfer will require careful definition. The concept, information, is drawn from general infor- mation theory. It is defined as any observation, perception or cognition that reduces uncertainty for the observer. A more thorough explication of this notion is provided later and will not be repeated here. Basically, uncertainty arises in a communication context when an interactant is confronted with several alternatives, either for encoding or decoding a message, and does not know which to choose. Information eliminates alternatives from the response field reducing the probability of error proportionately and, thus also, the level of uncertainty associated with the choice. For present purposes, information also necessarily involves a socially shared set of referent meanings. That is, un- certainty reducing behaviors based on rules or referent 13 meanings unique to a specific interacting dyad do not qual— ify as informative actions by this definition. The term transfer refers to mediated or face-to-face transmission of signs or symbols, with some awareness, from a human source to a human receiver. Consistent with the recursive communication model (cf. Figure 1, below) under consideration here, transfer implies only a one-way flow of signs or symbols from source to receiver at a single point in time. Information transfer, then, minimally requires a source manipulating signs or symbols with socially shared referent meanings in the presence of an interacting receiver. That is, information associated with the signs and symbols must be common to some specifiable set of individuals who systematically interpret them in a similar manner (Ekman & Friesen, 1969). Hence, the observation, perception or cog- nition of these signs and/or symbols muSt be informative, i.e., reduce the interacting receiver's uncertainty. As suggested earlier, an informative act implies some decoded meaning. It does not mean that the act of the source in question was intended to convey any information at all, although it does in fact do so. The source must, however, be aware that he/she has performed the act. This raises a critical distinction between intentionality and awareness, most germane to the classification of informative acts soon to be proposed below. 14 Much of the ensuing'discussion relies largely on the seminal work of Ekman and Friesen (1969) in their attempt to order the nebulous realm of nonverbal communication. In order to avoid a potentially unproductive entrenchment in problematic issues involving intentionality, the authors simply argue that a source can be aware of his/her nonverbal behavior whether or not he/she engages in the act with intent to communicate a specific message. In support of this con- tention, the following definitions are provided. Awareness or internal feedback refers to the person's knowledge that he/she is engaging in a particular nonverbal act at the moment it is done, or the ability to recall with ease what he/she has done (1969:53). Relevant examples are provided later. Intentionality implies deliberate use of a nonverbal act to communicate a specific message. This definition does not include behavior which is considered to be subconsciously intended. Intentional nonverbal acts must be within aware- ness and the source must purpose to send a message via his/ 'her act (1969:53-54). Obviously, prior definitions and distinctions do not solve all problems associated with levels of intentionality particularly with respect to operationalization in the non- verbal realm. Grey areas remain between those nonverbal acts which are clearly intentional and those which lie some- where within the source's stream of awareness but fall short of a conscious effort to communicate. One example is the 15 flushed face. Neither has the line been definitively drawn between acts that occur within and those that occur without awareness. The writer does contend that these modest con- ceptual beginnings will enable a clear identification of many, though by no means all, informative nonverbal acts. Thus far, our discussion has necessarily focused pri- marily on conceptual problems arising in the nonverbal area. The writer's intent from the outset was to provide a set of parameters sufficiently general to be applied with relative consistency to the verbal domain as well. Given prior cri- teria, it seems intuitively reasonable to subsume at least .four types of informative events under the conceptual um- brella labeled information transfer: (1) verbal communica- tion, (2) informative verbal acts, (3) nonverbal communica- tion, and (4) informative nonverbal acts. These four categories will be defined primarily in terms of the pre- vious intentionality/awareness distinction and the type of information conveyed in each case. Two additional types, verbal and nonverbal behavior, will be defined in general terms to clarify their exclusion via parameters established for the domain of events presently under description. verbal communication is defined as use of an elaborate code system or language by which an encoding source is actively and with deliberate intent attempting to make his/ her experience known to some other person(s). Use of a code system means manipulation of written or spoken word-symbols 16 with arbitrarily assigned, socially shared meanings. It is assumed that these symbols are addressed to receivers who are privy to the code system and thus able systematical- ly to interpret them. Informative verbal acts refer to implicit characteris- tics exhibited by a source in the process of encoding a verbal message. They support and contribute to, but do not independently constitute, verbal communication by our defi- nition. While these implicit dimensions of the encoded message are ordinarily produced without deliberate intent to inform, the source is usually aware of his/her actions at the moment of execution or is unable to recall them at a later point in time. Verbal immediaCy (Wiener & Mehrabian, 1968) illustrates the type of implicit verbal characteristics to which this category refers, e.g., systematic use of the first person plural pronoun ("we" as Opposed to "you and I"). Brief vocalizations expressing agreement or understanding (e.g., "uh-huh," "hmm") called vocal reinforcers (Mehrabian, 1971a) provide another example. Limited evidence to date suggests that verbal characteristics of this type are inform- ative regarding attitudes or emotional states of the source (e.g., Mehrabian, 1971a; Mehrabian & Ksionzky, 1972). Similar to its verbal counterpart, nonverbal communica- tion involves deliberate use of an elaborate code that func- tions much the same as any ordinary language system. Accordingly, the more rigorous definition adopted here re- stricts nonverbal communication to highly stylized gestural l7 movements called emblems (Ekman & Friesen, 1969). Emblems are defined as "those nonverbal acts which have a direct verbal translation or dictionary definition, usually con— sisting of a word or two, or perhaps a phrase. This verbal definition or translation of the emblem is well known by all members of a group, class or culture" (1969:63). These gestures may serve to repeat, substitute for, or contradict some part of the accompanying verbal message. Emblems are used in an intentional effort to communi- cate. The source knows when he/she is using an emblem, can repeat it if asked to do so, and will accept communicative responsibility for it. Many emblems are arbitrarily coded, e.g., the sign language used by the deaf. Others are icon- ically coded (visually resembling their meaning) as illus- trated in signal systems used by officials during sports events. While emblems may be exhibited in any area of the body, they are expressed primarily via the hands and face in this culture. The projected thumb of a hitchhiker is a familiar example of the former. Informative nonverbal acts have several characteristics in common with their verbal counterparts. They are intrins- ically related to the concomitant verbal message, are emitted quite spontaneously and ordinarily occur within the source's awareness. Similarly, they support and contribute to, but do not independently constitute, nonverbal communication. This class of events includes use of three additional l8 gesture types described by Ekman and Friesen (1969): illustrators, affect displays and regulators. In common with all informative acts previously defined, these non— verbal signs have agreed-upon, decoded meanings that elicit similar interpretations from a specifiable set of observers. Illustrators are defined as body movements which directly accompany and serve to illustrate what is being said verbally, for example, gestures which (1) time out, accent or emphasize a particular word or phrase; (2) sketch a path or direction of thought; (3) point to a present object; (4) depict a spatial relationship; or (5) draw a picture of the referent. Illustrators augment the verbal band "on a moment-to-moment basis; they are directly tied to content, inflection, loudness, etc." (1969:69). Affect displays are movements which reveal the state of one's emotions. While some bodily motions such as trem- bling are considered affective, the site of these nonverbal acts is more particularly the face. Distinctive movements of facial muscles are generally associated with the primary affects believed to be universal: happiness, surprise, fear, sadness, anger, disgust and interest (e.g., Tomkins, 1962, 1963; Tomkins & McCarter, 1964; Ekman & Friesen, 1967; Frijda & Philipszoon, 1963; Izard, 1968). These affective states are quite easily distinguished by observers in this culture; hence, it is assumed that they bear socially shared, decoded meanings. Affect displays may accompany use of the l9 verbal channel or they may meaningfully occur alone. The former usage is of central concern here. Regulators are nonverbal acts used to maintain or regulate the back-and-forth flow of speaking/listening be- havior between two or more interactants. They indicate that the speaker should continue, repeat, elaborate, in- crease speech rate, give up the floor, and so on. Regu- lators may also signal the listener to give special attention, wait or assume the encoding role. The most common regulator is the head nod; others include eye contact, slight movements forward, small pos- tural shifts, and eyebrow raising. Regulators carry no message content independent of the verbal conversation. Rather, they convey control messages necessary for pacing the exchange of information via the verbal band (Ekman & Friesen, 1969). Thus far, we have defined illustrators, affect displays and regulators in terms of their primary functions for the purpose of setting some parameters on the class of events termed informative nonverbal acts. These subcategories can, and likely do, overlap in actual conversation. Affect dis- plays, for example, can also serve as regulators. As Mahl (1968) has suggested, the fact that a particular nonverbal act may exert some influence on the verbal flow, does not mean that regulation is its sole or even primary purpose. Hence, the term regulator is reserved here for those 20 behaviors that do not conceptually fit into other categories such as illustrators or affect displays. The latter cate- gories were also defined in terms of their major functions. In short, regulators appear only to regulate and tend to occur at a relatively low level of awareness. In sum, preceding discussion describes the domain of events with which subsequent theoretic statements are con- cerned. Four types of action were subsumed under the rubric of information transfer: verbal communication, informative verbal acts, nonverbal communication and informative non- verbal acts. This choice of nomenclature deliberately reserves the term behay1g£ for reference to verbal/nonverbal actions that are uninformative by prior definition. Verbal behavior, as opposed to communicative/informa- tive verbal acts, will refer to unintentional linguistic innovations (Sturtevant, 1947), i.e., articulatory or per- formance errors that do not contribute to, and may actually inhibit, information exchange. Boomer and Laver provide a useful definition in this regard: "A slip of the tongue ... is an involuntary deviation in performance from the speaker's current phonological, grammatical or lexical intention" (l968:4). Behaviors of this type include, for example, non- fluencies such as meaningless combinations of phonemes or morphemes that occur during a lapse in the encoding process, whole words out of semantic context, repetition of syllables, spoonerisms, omissions and mispronunciations. While these 21 behaviors may occur within the encoder's awareness, they fail to meet the criterion of agreed-upon, decoded meaning. Similarly nonverbal behavior, as opposed to communica- tive/informative nonverbal acts, will refer to what appear (on present knowledge) to be more random, idiosyncratic movements which specific individuals develop over time. The conceptualization of adaptors provided by Ekman and Friesen serves well as a general description of the type of nonverbal activity intended here. Adaptors are believed to be movements learned in childhood as part of adaptive efforts to satisfy bodily needs, manage emotions, learn instrumental activities or develOp and maintain prototypic interpersonal contacts. While adaptors represent behaviors first learned "as part of a total adaptive pattern where the goal of the activity was obvious, when these actions are emitted by the adult particularly during social conversation, only a frag- ment of the original adaptive behavior is seen. These frag- ments or reductions of previously learned adaptive acts are maintained by habit ... By this reasoning, adaptors when emitted by the adult are habitual, not intended to transmit a message, and usually without awareness" (1969:84-85). Examples of this behavioral type are hand-to-face (or head) movements such as grooming the hair with one's hands in public, wiping the lips with the tongue or hands, picking or scratching the scalp, and wiping the brow or nose with the back of the hand. Restless movements such as bouncing 22 the lower leg and foot in a seated position also exemplify this category. While these movements can on occasion pro— vide a basis for attitudinal inferences, Ekman and Friesen argue that their interpretation "is difficult, often specu- lative and uncertain" (1969:87). Behavioral Determinants of Social Interaction This section reviews basic research in pursuit of heuristically promising independent variables. Social intercourse in general has been the subject of considerable study. Investigation of this area has had at least two emphases: (1) description and measurement of specific behaviors that determine social interaction, and (2) exam- ination of effects situational and individual differences may have on these behaviors (Mehrabian & Ksionzky, 1972). This essay relies on findings in the former area to develop theory of the latter type. Prior research repeatedly demonstrates that two or three basic factors can be used to describe and measure social interaction. Several studies using Bales' (1950) Interaction Process Analysis (e.g., Carter, 1954; Couch, 1960; Bales, 1950, 1968) consistently produced three under- lying dimensions identified as affection (Positive/negative evaluation), power, and contribution to group tasks. 23 Others have obtained two major factors (e.g., Borgatta, 1962): sociability, similar to affection, and assertive— ness relating to power. Mehrabian and Ksionzky (1972) used twenty-six reliably scored verbal/nonverbal cues with previously demonstrated social significance (of. Mehrabian, 1972a) to measure behaviors of participants in live interaction. Factor analysis of their observations yielded a six-factor solu- tion: (1) affiliative behavior, related to the affection factor above; (2) responsiveness, an index of activity or reaction to the counterpart; (3) relaxation, believed to imply dominance and thus related to the power factor ident- ified earlier; (4) ingratiation; (5) distress or active avoidance; and (6) intimate position, an index of physical proximity. The authors note that the first three factors are of major import in characterizing social interaction. The foregoing research confirmed original factor ana- lytic work by Williams and Sundene (1965) and Osgood (1966) which identified essentially the same three factors as major referents of nonverbal communication. Similar ana— lyses (Osgood, Suci & Tannenbaum, 1957; Snider & Osgood, 1969) on verbal cues using semantic differential scales likewise obtained factors labeled evaluation (liking or preference), dominance (power) and activity (responsiveness). Mehrabian (1972b) has used these three referent meanings to characterize nonverbal communication and demonstrated their 24 relevance across a wide range of social situations. An early review (Foa, 1961) of factor-analytic studies in this area indicated that two factors consistently emerge, love-hostility and dominance-submission. In a therapeutic context, Lorr and Suziedelis (1969) factor analyzed ratings of social behavior across patient/normal groups; the result- ing structure included essentially the same two factors, sociability (or affiliation) and control (or power). Cross- cultural factor-analytic studies on (1) group behavior of children (Longabaugh, 1966) and (2) role perceptions, be— havioral intentions and perceptions of social behavior (Triandis, Vassiliou & Nassiakou, 1968) have produced comparable results. Summarily, research attempting to characterize social interaction via factor-analytic techniques indicates that three factors are consistently identified: affiliation, potency (or status) and responsiveness. The first factor, typically identified as affiliation (sociability, love- hostility, or affection), appears more tentatively to be the dominant dimension across a variety of social settings. Effects of situational and personality correlates on the characteristic dimensions of social interaction report- ed in factor analytic research have been examined less extensively. Two experiments in this area (Mehrabian, 1971a; Mehrabian & Ksionzky, 1972) have had considerable influence on the present work. These studies, based on a 25 model of affiliative behavior by the same authors (1970), were primarily designed to explore effects of social rein- forcement and individual differences on nonverbal and implicit verbal behaviors exhibited by pairs of strangers interacting in a waiting situation. Their proposed model hypothesized that affiliation increases as a function of positive (and decreases as a function of negative) rein- forcement by an interacting counterpart. The main experi- mental factors introduced in both studies were: (1) degree of positive reinforcement and (2) degree of negative rein- forcement received prior to or during interaction. The second, more comprehensive study introduced four levels of reinforcement (positive, negative, mixed and neutral) and included anticipated cooperation versus competition with the target person as an additional independent variable. Five individual difference variables: affiliative tendency, sensitivity to rejection, achieving tendency, sex and birth order were measured by questionnaire. Dependent measures, taken on each subject during spon- taneous interactions with an experimental confederate, were twenty-six nonverbal/implicit verbal behaviors drawn from prior factor-analytic research. Scoring criteria for each dependent measure are reported in Mehrabian (1972b, Appendix A). The observations were factor analyzed and a principal component solution yielded the following six factors: affiliative behavior, responsiveness, relaxation, 26 ingratiation, distress and intimacy. Table 1 reports specific behaviors subsumed under the first factor which is most relevant here. Variables are listed in order according to the magnitude of their loadings (highest to lowest). Direction of the loading for each variable is also indicated in parentheses. Additional analyses related each of the six factors (via composite indices for each subject) as dependent measures to the eight independent variables listed above. Significant main effects were found, as expected, for affiliative behavior as a function of affiliative tendency and anticipated c00peration. The direction was positive in both cases. Several significant interaction effects, particularly when positive/negative reinforcements were included, provide refinements on these single-variable relationships. First, subjects who scored high on the affiliative tendency measure exhibited more affiliation gn1y when they did not receive negative reinforcement from the confederate prior to interaction, regardless of whether or not positive evaluations were also involved (i.e., via mixed reinforce- ments). Second, a significant third-order interaction effect was found in one of the four evaluation conditions. 27 TABLE 1 Nonverbal and Implicit Verbal Behaviors Defining Affiliation, a Primary Dimension of Social Interaction (Mehrabian & Ksionzky, 1972: 591) FACTOR I: Affiliative Behavior Total number of statements per minute (+) Number of declarative statements per minute (+) Percent duration of eye contact with confederate (+) Percent duration of subject's Speech (+) Percent duration of confederate's Speech* (+) Positive verbal content (+) Head nods per minute (+) Hand and arm gestures per minute (+) Pleasantness of facial expressions (+) *The only item that did not pertain to subject's behavior. 28 Subjects who received mixed evaluations (i.e., high posi- tive and high negative reinforcements) affiliated signifi- cantly less when anticipating competition rather than cooperation. In sum, these findings were largely consistent with the authors' original model of affiliative behavior. They also support the primacy of affiliation as a basic dimen- sion of social interaction and suggest the following con- clusions. First, the Mehrabian and Ksionzky (1970) model provides a useful framework for investigating the special type of interaction at hand. Second, results of the latter study above are intuitively consistent with early research on communication with handicapped persons (e.g., Kleck, Ono & Hastorf, 1966; Kleck, 1969; Worthington, 1974) and encourage an initial approach to this area focusing pri- marily on affiliation as a promising source of further insight. Third, the interaction effects reviewed above clearly indicate that mixed or discrepant evaluative cues (emitted by a counterpart via only a single channel) tend to reduce affiliative effort even in relatively "normal" social situations, most notably among persons with strong tendencies to affiliate. It is reasonable then to expect that this negative effect will increase if discrepancies in affiliative communication occur across the two major channels when handicapped/nonhandicapped persons interact. 29 Theoretic Moorings If the interpretation of empirical data depends on theory (Heise, 1975), it is also the case that sound theory must be grounded in empirical data (Glaser & Strauss, 1967). Typically, the researcher attempts to formulate the best theoretic statement possible in light of existing evidence (Blalock, 1969). It seems reasonable then to approach the problem of handicapped/nonhandicapped communication from the vantage of current theoretic developments and prior research on interpersonal relationships in general. Results of previous handicap research reviewed earlier (e.g., Kleck, 1966; Marinelli & Kelz, 1973; Marinelli, 1974) suggest that atypically high levels of uncertainty are present when nonhandicapped persons initially interact with handicappers. Hence, the general uncertainty reduction paradigm outlined below was adopted as the most promising theoretic point of departure. Human beings have a strong innate need to make sense out of their environment (Heider, 1958). We strive for certain- ty and have, perhaps, more tolerance for ignorance than ambiguity. The theoretic perspective to follow rests on the assumption that when unfamiliar persons initially inter- act, they are primarily concerned to reduce their own uncertainty in order to predict, explain and ultimately 30 control the effects of their respective communication behaviors (Jones, Kanouse, Kelley, Nisbett, Valins & Weiner, 1972; Kelley, 1973; Berger & Calabrese, 1975). Prior to choosing a course of action, interacting persons attempt to predict the most likely alternative among possible responses their counterpart might make (Miller & Steinberg, 1975). This predictive process is conceptually analogous to the strategy—planning forethought of a chess player who anticipates the counterpart's response alternatives before each move. In fact, probable outcomes for several reciprocal moves will likely be pre- dicted in advance. Such predictions are assumed to be more or less certain depending on the amount of relevant infor— mation available to the predictor. Information is defined as any observation, perception or cognition that reduces uncertainty. Generally, uncer— tainty arises when one is confronted with several alterna— tives and does not know which to choose. The level of uncertainty in a given situation is defined by the prob- ability of error inherent in the number of plausible and desirable options from which one must choose. In theory, information eliminates alternatives from the response field, reducing the probability of error proportionately. Conse- quently, the level of uncertainty associated with the choice is also reduced. In short, to the extent that one's choice is informed, the number of plausible options is 31 decreased and prediction is made with greater certainty. Information, so defined, is the complement of uncertainty (Krippendorff, 1975) and the stock-in-trade of uncertainty reduction. Following recent attribution theory, Berger and Calabrese (1975) provide a useful distinction between two types of uncertainty. First, the proactive type is associ- ated with prediction as described above; it occurs prior to the actual behavior in question. Proactive uncertainty is a function of the amount of information available to partic- ipants during interaction enabling them (1) to predict the most likely response their counterpart may make, and (2) to choose an appropriate strategy from their own repertoire so as to maximize likelihood of a desired response (Miller & Steinberg, 1975). Second, retroactive uncertainty is associated with the problem of explaining or interpreting a particular behavior after its occurrence. Several theorists (e.g., Jones e3 g1., 1972; Kelley, 1973) hold that in addition to predicting responsive actions, human interactants are equally con- cerned to explain or validate social behavior, their own as well as that of others. There are, typically, a number of plausible attributions that can be made from a particu- lar social act. In this regard, Schachter and Burdick (1955) demonstrated that recipients of ambiguous messages tend to engage in intensive social contact for the purpose 32 of reducing uncertainty of this type. Similarly, humans are often as uncertain about events within themselves as they are about events in their envi- ronment. Considerable research on the link between the need for self-evaluation and affiliative behavior has shown that human beings desire to identify or confirm their own reactions to specific situations (e.g., Fromm-Reichman, 1959; Schachter, 1959). This research is supportive of Festinger's (1954) social comparison theory suggesting that persons seek out similar others with whom to confer when they experience high levels of uncertainty regarding the appropriateness of their own behavior and/or their percep- tions of a particular situation. More specifically, retroactive uncertainty is a func- tion of the amount of information available to participants during interaction enabling them (1) to select the most likely explanation for a counterpart's behavior, and (2) to identify or confirm their own reactions to that behavior. In addition, information exchange (or transfer, in this context) is regarded as the primary means interacting per- sons use to minimize the number of alternatives from which they must choose for both predictive and explanatory purposes. According to Blalock (1969), a deductive theory con- tains both axioms and theorems. Axioms are those proposi- tions assumed to be true while theorems are logically 33 derived or deduced from the axioms. Moreover, axioms should be causal assertions that are, strictly speaking, "untestable because of the fact that it will never be pos- sible to control for all 'relevant' variables" (1969:11). This is tantamount to saying that axioms should be causal, "lawlike" probability statements for which there is sub— stantial evidence. By these criteria, the axioms to be proposed momentar- ily do not qualify this formulation for deductive theory status. It is my judgment that the following propositions should be considered associational, rather than lawlike causal statements, in lieu of additional support. They were formalized in light of the preceding rationale and additional evidence to be cited. Each axiom pertains to initial interaction between strangers and thus assumes a high level of uncertainty at the outset (Lalljee & Cook, 1973). Axiom 1: As level of information transfer generated by e1ther interactant increases, level of communica- tive uncertainty experienced by the counterpart will decrease. Conversely, increasing levels of communi- cative uncertainty experienced by the counterpart will accompany decreasing levels of information trans- fer generated by either interactant. Axiom 1 relates the two primitive terms with which our formal theory begins. It is logically true in terms of prior complementary definitions provided for information transfer and uncertainty. This proposition might simply have been stated as the pivotal a priori assumption 34 underlying the entire theory. It was formalized here in the interest of clarity. Subject to future refinements, Axiom 1 may also prove useful beyond present definitional constraints. In addition, Axiom l expresses a theoretical relation- ship between highly abstract concepts. Even before these concepts are operationalized via specific observables, a second important inferential leap is logically required. In the interest of experimental control (i.e., testability), information transfer must be reduced to a lower level of abstraction. The necessary reduction can be accomplished by assuming that the amount of information transferred in a given interaction is approximately coterminus with the number of communicative/informative acts (as defined above) transmitted from source to receiver. This assumption is formalized in the next two propositions. Axiom 2: As number of communicative/informative verbal acts emitted by either interactant increases, level of information transfer will increase. Con- versely, decreases in level of information transfer will accompany decreases in number of communicative/ informative verbal acts emitted by either interactant. Axiom 2 is a crucial step in the logical process at hand for at least two reasons. First, the informative value of any given verbal act will likely vary in actual performance depending on the context, information levels held prior to interaction, individual differences reflected in the encoding process, and so on. Second, it is highly 35 unlikely that discrete verbal acts (even of the same type), to be quantified as a means of measuring information trans— fer, will be equally informative. For example, several declarative sentences expressed by a given source may have less informative value than the smaller number expressed by a counterpart. Nor can it be taken for granted that dis- crete verbal acts regularly function in a linear, additive fashion, i.e., without interactive effects. Hence, the number of discrete verbal acts produced by interactants is assumed to be a gross index of the actual amount of information transferred in a given context. The preceding proposition makes this assumption explicit. The following rationale is offered in support of Axiom 2. Lalljee and Cook (1973), in a study allowing two-way communication, found that as interactions between strangers progressed, the number of words spoken per minute increased significantly over a nine minute period. The present form- ulation suggests that as the number of informative verbal acts exchanged in this study increased, the amount of infor- mation available to each person also increased, resulting in uncertainty reduction for both persons. In turn, reduced uncertainty encouraged the production of verbal acts by each participant at an accelerating rate. Conversely, Berger and Larimer (1974) found that when feedback was not allowed in a similar context, the number of words spoken per minute decreased significantly over a 36 four minute period. Our theory suggests that restricting feedback in this instance interrupted the flow of informa- tion by introducing a significant decrease in the number of informative verbal acts emitted by at least one person at a given point or over time. Intuitively, the absence of an appropriate verbal response can be expected to impede exchange of information by increasing retroactive 32335- tainty dramatically. That is, little or no feedback would seem to be the least informative, and may be the most ambig- uous, of plausible response alternatives when an uncertain interactant initially solicits or discloses information. 'rrflndeed, the absence of verbal feedback can be expected to increase the number of possible interpretations open to the receiver and to render the next response virtually un- predictable. If interacting strangers are primarily con- cerned to reduce uncertainty, then initial attempts to communicate that result in greater uncertainty might appear (to the source) as unrewarding, or even to elicit punishment. Moreover, if absence of feedback is perceived as negative reinforcement or punishment (via increased uncertainty, we would expect a reduction in the subsequent number of verbal acts generated by the unrequited source. Hence, the decrease in number of words per minute observed by Berger and Larimer. On the prior line of reasoning, it is assumed that level of information transfer and number of informative verbal acts will covary in the same direction (Axiom 2). 37 By extension of the preceding rationale, a similar positive association is posited between information trans- fer and informative nonverbal acts. Axiom 3: As number of communicative/informative nonverBal acts emitted by either interactant increases, level of information transfer will increase. Conversely, decreases in level of information transfer will accompany decreases in number of communicative/informative nonverbal acts emitted by either interactant. Introductory comments alluded to the common assumption that certain nonverbal cues are closely associated with, and auxiliary to, linguistic behavior (e.g., Harrison, 1974; Ekman & Friesen, 1969). Under normal circumstances, these nonverbal counterparts of language function to repeat, substitute for, complement, accentuate, contradict or reg- ulate verbal communication (Burgoon & Saine, 1978). Al- though the processes involved may differ considerably, the present claim is that verbal and nonverbal actions (within the domain defined above) serve essentially the same general purpose, exchange of information. Hence, the parallel relations posited in preceding propositions. The rationale offered in support of Axioms 2 and 3 attempts to provide some theoretic justification for reduc- ing the concept, information transfer, to an experimentally manageable level of abstraction. This move has consider- able bearing on available options from which relevant, observable variables can subsequently be selected. How- ever, the domain of eligible communicative/informative 38 events that might be used to index information transfer (across the two major channels) is still extremely large. Thus, an important question rises at this juncture. Given our theoretic concern with uncertainty reduction, what type of information is most relevant, i.e., most likely trans- ferred, during initial interactions? The following pair of propositions and supporting rationale address this issue. Both axioms express reciprocal relationships. Axiom 4: As level of verbal affiliation increases, level of information transfer will increase. As level of information transfer increases, level of verbal affiliation will further increase. Decreases in level of information transfer will accompany decreases in level of verbal affiliation. Axiom 5: As level of nonverbal affiliation increases, IeveI of information transfer will increase. As level of information transfer increases, level of non- verbal affiliation will further increase. Decreases in level of information transfer will accompany decreases in level of nonverbal affiliation. Axioms 4 and 5 are based on the following summary of previous discussion. Prior factor-analytic research on behavioral determin- ants of social interaction indicated that three factors consistently emerge. The first factor, typically identified as affiliation (sociability, love-hostility or affection), appears to be the dominant referent dimension of social interaction across a variety of settings (Mehrabian, 1971a). Two such studies in particular (Mehrabian, 1971a; Mehrabian & Ksionzky, 1972), exploring relationships among nonverbal/ 39 implicit verbal behaviors of strangers in a waiting situ- ation, suggest several associations of theoretic interest here. The model under study hypothesized that affiliation is encouraged by positive verbal and nonverbal reinforce- ment while it is discouraged by negative reinforcement of both types. Significant positive correlations were found in both studies between indicators of the amount of verbal communication and several measures of positive reinforce- ment. Both sets of indicators showed strong positive load- ings on the first factor labeled affiliative behavior./ Affiliation is a useful construct in this context for two reasons. First, it is akin to the concept of "approach- avoidance" as discussed by Mehrabian (1971b) and thus sug- gests the tendency for individuals to decrease their physi— cal distance in order to interact. In terms of the present formulation, affiliation is then a logical precondition for increased interaction and uncertainty reduction via infor- mation transfer. Second, affiliation is generally identi- fied with positive reinforcement which serves to encourage both physical and social proximity during interaction. Thus, affiliation may be considered a fundamental (positive), source-oriented stimulus which is expected to foster the exchange of similar (as well as other) types of information. Factor-analytic studies cited above indicate that posi- tive reinforcement is exchanged via nonverbal as well as verbal cues. Approximately half of the variables loading 40 on the affiliative factor in both studies (Mehrabian, 1971a; Mehrabian & Ksionzky, 1972) were nonverbal measures. The latter study in particular provides evidence recommend- ing two constructs of theoretic interest for purposes of examining verbal/nonverbal relationships in the context of handicapped/nonhandicapped communication. Division of the affiliative factor (cf. Table 1) into its verbal and non— verbal components yields a composite set of positive indi- cators for both verbal and nonverbal reinforcement. For reference purposes here, the composite set of verbal meas- ures is labeled verbal affiliation; their nonverbal counter- parts, nonverbal affiliation. Theoretically, both constructs represent the prerequi- site type of reinforcement by which social interaction is initiated and continues to develop. Defined in terms of specific behaviors identified by Mehrabian and Ksionzky (1972), affiliation is the logical first-step in our uncer- tainty reduction model presented in Figure 1. Based on prior discussion, both verbal and nonverbal affiliation are expected to occupy a direct, positive and reciprocal rela- tionship with the broader class of events defined previously as information transfer. These relations were formalized in Axioms 4 and 5. 41 nommcmuu comics-35 u ZOmflMnm .o>«fl&ou0n— coufiavoz >33uounb no Eouh 2.3-qu 153.3513: \vR—nduuvnn: 30033 23309-35 ‘33.: «co-9.33: 3 cocoa—cun— roosh. 05 no :32 7033 a ”Sauna-3a «...—.0 Do: .— Cunarn "-..... s . 30¢ 3.10232 “ 03.15.33 .3 ”03.332500 ....-'.. coaaluuuu< 7.: 02.32 .‘---... .- 30< _3uo> u 0323.335 no . 01733325550 r----- :33: on: 35 25... 09—35 533.33. :omuumzu“ .530”: um uoumcmuu 3031:: 35 N ZOWQHQ .3 < «I.» 0232 0335.33 .3 a 93 Humans—=00 r....-- 60331.3( ulna 02.02 J and» 0> . .. .t< it; . 0335.31: no .- o>3335==t00 — ZOmlflnm 42 Observed correlations among empirical indicators of verbal and nonverbal affiliation (Mehrabian, 1971a) suggest a strong association between these variables. Fifteen of sixteen relevant intercorrelations were significant (a = .01) in a positive direction; the single exception ap- proached significance at this level in the same direction. These data, collected exclusively from observation of target subjects, imply that interacting strangers ordinar- ily encode their affiliative messages with a high degree of consistency across the two channels. On this account, levels of verbal and nonverbal affiliation emitted simul- taneously by a single source are expected to covary in the same direction. Axiom 6: As level of verbal affiliation increases, level of nonverbal affiliation from the same source will increase. Decreases in level of nonverbal affiliation will accompany decreases in level of verbal affiliation from the same source. A major concern of this paper is the potentially serious problem the previous discussion implies for handicapped persons with normal verbal skills who may be unable to encode appropriate nonverbal cues in a manner consistent with Axiom 6. As noted earlier, physically impaired per- sons may frequently exhibit nonverbal signals that are per- ceived to be incongruous with their verbal communication. Axiom 6 provides a theoretic basis for the following con- ceptual distinction between congruent and incongruent verbal/nonverbal messages. Incongruent messages are 43 defined as those in which the level of a specific type of information conveyed via the verbal channel is perceived to be significantly different than the accompanying level of the same type of information conveyed via the nonverbal channel by the same source. Conversely, congruent messages are those in which levels of a given type of information emitted simultaneously across the two channels (by a single source) are not perceived to be significantly different. In this connection, it is postulated that when incongru- ency occurs between simultaneous verbal/nonverbal cues emitted by the same source, the nonverbal portion will dominate in determination of the total message impact. Direct empirical support for this prediction was found in two experimental studies (Mehrabian & Ferris, 1967; Mehrabian & Wiener, 1967). Their findings indicate that the combined effect of verbal, vocal and facial cues which convey liking attitudes is a weighted sum of their inde- pendent effects as represented in the following equation (Mehrabian, 1971b): Total liking = 7% verbal liking + 38% vocal liking + 55% facial liking More recently, two studies confirmed the former finding (Argyle, Salter, Nicholson, Williams, & Burgess, 1970; Argyle, Alkema & Gilmour, 1971). Subsequent research sug- gests that attitudes conveyed through various channels may interact to determine the total message impact (Bugental, 44 Kaswan, Love & Fox, 1970; Lampel & Anderson, 1968). Hence, while the equation above represents only a first-order ap— proximation (Mehrabian, 1972b), it does provide general support for the dominant effect of the nonverbal band. With regard to affiliative cues in this case, when incon- gruency occurs across the two channels, the nonverbal com- ponent is expected to exert a greater influence on the sum total of information conveyed. This expectation is formal- ly stated in the following proposition. Axiom 7: Given incongruent levels of verbal/non- verbal affiliation emitted simultaneously by a single source, level of nonverbal affiliation will exercise the dominant influence on level of information transfer. Since affiliation is only one of three basic determi- nants of social interaction consistently identified in prior research, it seems reasonable to assume that other factors involving different types of information (e.g., potency) will also influence uncertainty reduction and contribute to the receiver's ultimate perception of how he/ she is being evaluated. Hence, the present model (see Figure 1) suggests that level of information transfer will mediate the effects of verbal/nonverbal affiliation on communicative uncertainty. The latter construct, communi- cative uncertainty, is logically associated in this model with attitudes inferred by, and other responses of, the receiver as reasoned below. 45 In support of Axiom 6, it was argued that interacting strangers ordinarily encode their affiliative messages with a high degree of consistency across the two major channels. If this is the case, it is reasonable to expect that less congruent messages will introduce greater ambiguity into the decoding process than more congruent levels of the same verbal/nonverbal cues (Mehrabian & Reed, 1968). That is, incongruent messages are expected to be more difficult to decode since they tend to increase the number of plausi- ble, alternative interpretations from which a receiver must choose. Assuming that some degree of cancellation occurs when conflicting bits of information emanate from the two channels, the combined informative impact of incongruent verbal/nonverbal messages is expected to be lower and, con- sequently, to have less uncertainty reducing effect than congruent messages. They may, in fact, increase the receiver's uncertainty depending on the degree of discrep- ancy between the two channels. While some degree of ambi- guity may be characteristic of all human communication, as discrepancy across the two channels increases, recipients of incongruent messages are expected both to experience less uncertainty reduction and to attribute greater uncer- tainty to the source. More specifically, perceived verbal/ nonverbal incongruency in single source messages should be inversely related to information transfer (and, in turn, to the level of certainty experienced by the receiver). 46 Hence, the following formal proposition is advanced. Axiom 8: As level of verbal/nonverbal incongruency in Single source messages increases, level of infor- mation transfer will decrease. Increases in level of information transfer will accompany decreases in level of verbal/nonverbal incongruency perceived in single source messages. Several theorists have advanced the notion that "response matching" behavior (Argyle, 1969; Brooks & Emmert, 1976) or a reciprocity norm functions to control exchange of information in interpersonal transactions (Gouldner, 1960; Jourard, 1960; Berkowitz, 1968; Berger & Calabrese, 1975). According to Argyle, "During social interaction it is very common for an act by A to be fol- lowed by a similar act from B. This we call response matching ..." (1969:171). Numerous studies between 1950 and 1969 have verified this phenomenon with regard to length of utterances (Matarazzo, Wiens & Saslow, 1965; Argyle & Kendon, 1967); interruptions and silences (Argyle & Kendon, 1967); type of utterance (Bales, 1950); gesture and posture (Rosenfeld, 1967); and self-disclosure (e.g., Jourard, 1960). It is also likely that response matching occurs with regard to emotional states, i.e., an "emotional contagion" takes place (Argyle, 1969; Brooks & Emmert, 1976). The latter area may be particularly important in the context of communication with handicapped persons where high stress has been observed repeatedly. 47 It should be noted that the term response matching as used in this context implies only that increases or de- creases in a particular behavior by a given source tend to be reciprocated via changes in the same type of behavior moving in the same direction by the respondent. Response matching does not imply that the reciprocal changes are quantitatively equal or symmetrical (cf. Matarazzo, Wiens & Saslow, 1965). It has been argued that the reciprocity or response matching construct is of limited theoretic value since theorists have been unable to adequately explain the phe- nomenon (Altman & Taylor, 1973). In response to this claim, Berger and Calabrese (1975) advance a general rationale for reciprocity based on Goffman's (1959) notion that people prefer smooth running rather than stressful interactions. Utilizing an uncertainty reduction paradigm similar to the present formulation, these authors contend that the "easi— est way to reduce mutual uncertainty would be to ask for and give the same kinds of information at the same rate of exchange. In this way, no one interactant in the system would be able to gain information power over the other" (1975:105). While the former explanation moves in a heuristically promising direction, the writer finds it inadequate for two reasons. First, to assert a human preference for less stressful interactions and to assume that reciprocity is 48 the "easiest way" to reduce uncertainty does not advance our understanding of the phenomenon much more than the assertion of a reciprocity norm ipso facto explains its presence. Second, the explanation as stated would seem not to account for response matching behavior that involves stress itself. For example, in situations such as those under analysis here, stressful interactions frequently do occur and anxious reactions per se appear to be recipro- cated. The notion of "information power" is, in the writer's view, a more promising idea for development of useful theory to explain reciprocity within the uncertain— ty reduction perspective. If it is assumed, as several influential theorists argue (e.g., Homans, 1961; Thibaut & Karma; 1959; Miller & Steinberg, 1975), that the basic function of communication is to control one's environment so as to realize certain rewards, information which enables a communicator to predict the effect of his/her own behavior with greater certainty becomes a powerful means of such control. Pos- session of information, required by either interactant to reduce uncertainty, represents potential power to direct the interaction toward one's own desired end. Given the transactional nature of human communication (Miller & Steinberg, 1975), exchange of information can be concept- ualized as a kind of bartering process in which the inter- actants attempt to relinquish as little information as 49 possible in quest of maximum returns. On this account, unfamiliar persons are expected to exercise considerable caution in their release of and requests for relevant information. In effect, they may initially engage in a mutual testing exercise to determine "supply and demand" characteristics of the situation. That is, during early stages of the encounter,interactants may be willing to risk only minimal amounts of relatively superficial information while attempting to reduce their own uncertainty regarding the counterpart's willingness to reciprocate in kind. Indeed, several studies have shown that the amount of demographic (low intimacy) information requested and disclosed is highest during the earliest stages of interaction. Significant decreases in the amount of demographic information exchanged occur rapidly after the first minute of interaction (e.g., Berger, 1973; Cozby, 1972; Sermat & Smyth, 1973; Taylor, Altman & Wheeler, 1973). At the same time requests for, and disclosures of, more intimate information (e.g., personal attitudes and opin- ions) were observed to increase significantly. Altman and Taylor (1973) contend that intimacy level of communication content tends to increase over time as the relationship becomes more rewarding and less costly. The Miller/Steinberg (1975) distinction between psychological and sociological information suggests an explanation more consistent with the present formulation. They argue that 50 predictive (or proactive) uncertainty is reduced more effi- ciently by psychological level data, i.e., predictive error can be greatly reduced by basing one's forecast on informa- tion that distinguishes the counterpart as a unique person. Given the basic control function of communication and assuming that control depends on predictive accuracy, interacting strangers should prefer to reduce their uncer- tainty as rapidly and efficiently as possible. Hence, each will be inclined to pursue the more idiosyncratic informa- tion his/her counterpart may be willing to disclose. At the same time, while personal information represents increased potential for control and is therefore in greater demand, it also involves greater risk to the discloser. On the prior rationale, response matching is even more likely to occur and/or continue as the interaction moves toward psychological level data, i.e., as more intimate information is exchanged. Thus, the early "demographic stage" may serve to initiate the preliminary prediction- making process via exchange of low risk information; it may also function to establish the response matching rule by which exchange of more personal information will be regulated. Indeed, several investigators have observed the response matching phenomenon at high levels of self- disclosure (e.g., Jourard, 1960; Cozby, 1972; Sermat & Smyth, 1973). 51 The present formulation posits an inverse relationship between a source's level of uncertainty and his/her self- disclosure. As uncertainty is initially reduced, particu- larly with regard to the counterpart's willingness to respond in kind (which greatly facilitates predictive accu- racy), more intimate self-disclosure is expected to increase. As intimacy increases, response matching is expected to become the more definitive due to potentially negative con- sequences which accrue with intimate self-disclosure, e.g., risk of rejection, abuse, or loss of control in the inter- action. Indeed, as intimacy level increases, response matching may also increase within certain parameters appro- priate for a given stage of relational development. That is, a high intimacy threshold can be reached so as to violate social norms for initial encounters and/or place the receiver under an unacceptable obligation to respond in kind. At this extreme, response matching may begin to decrease as suggested by Miller and Steinberg (1975) and observed by Cozby (1972). Ordinarily, interacting strangers are not expected to violate such intimacy norms. This account attempts to explain response matching, as observed in prior research, at various levels of information exchange. It diverges considerably from the Berger and Calabrese (1975) formulation which postulates that high levels of uncertainty produce high rates of reciprocity while low levels of uncertainty produce low rates of 52 reciprocity. The authors do not clearly define reciprocity rate. However, their view does not appear to take account of the increased potential for uncertainty reduction, pre- dictive accuracy and control involved in the exchange of more salient, idiosyncratic information. Neither does it explain the response matching phenomenon observed repeated- ly when highly intimate information is being exchanged. The contention here is that response matching continues to influence the type and flow of information well beyond initial interactions and is particularly critical when high risk information is being disclosed. On present evi- dence, it is quite conceivable that the decline in reci- procity rate associated with decreasing uncertainty by Berger and Calabrese does not occur until well-established interpersonal relationships have developed over time. Three specific associations reflecting prior research on response matching can be meaningfully integrated with the theoretic framework proposed thus far. First, in a series of experimental studies by Matarazzo, Wiens and Saslow (1965) significant increases in length of interviewee speech duration were observed to follow increases in length of utterances by the interviewer. The associated increases were not quantitatively symmetrical. Observed ratios of interviewee to interviewer speech duration were, however, remarkably constant at five or six to one across several 53 studies. Admittedly, length of utterance alone is not likely a very reliable index of informative value as, for example, in the case of a congressional filibuster. It is intui- tively reasonable to assume for theoretic purposes that speech duration and number of communicative verbal acts (as defined above) will ordinarily covary in the same direc- tion. By extrapolation, the response matching pattern observed by Matarazzo EE,El° might also be expected to occur re: number of communicative verbal acts as well as other informative events within the domain of theoretic interest here. Hence, on the assumption expressed in Axiom 2, gross levels of information transfer generated by inter- actants in initial dyadic encounters are expected to covary in the same direction. Formally stated, the following proposition refers to the combined impact of verbal/non- verbal acts (emitted by a single source) previously defined as either communicative or informative events. Axiom 9: As level of information transfer generated by either interactant increases, level of information transfer generated by the counter- part will increase. Decreasing levels of infor- mation transfer will be reciprocated with deceasing levels of the same variable. Second, response matching has repeatedly been found to occur with regard to communication content or type of verbal message exchanged. Worthy, Gary and Kahn (1969), in addi- tion to several studies cited earlier, provide empirical 54 evidence in support of matching message types involving self-disclosure. Given the influence, for example, of psychological level data on uncertainty reduction as dis- cussed above, it is reasonable to assume that reciprocal content-matching messages will have a direct bearing on information transfer, mutual levels of uncertainty reduc- tion and the outcome of an interaction. For present pur- poses, this reciprocal relationship focuses exclusively on verbal components of the interaction. In general, levels of a given type of verbal content generated by individual interactants are expected to covary in the same direction. For example, increases in positive verbal reinforcement produced by a given interactant will accompany increases of the same verbal type from the counterpart. It is further assumed that specific types of verbal content are usually conveyed via the standardized use of a relatively specific symbol set. For example, self-disclosure may involve increased use of the first person, singular pro- noun. On these accounts, the following response matching association extends to all verbal acts previously included within the larger domain labeled information transfer. Axiom 10: As level of a specific type of verbal information transferred by either interactant in- creases, level of the same type of verbal infor- mation transferred by the counterpart will increase. Decreasing levels of a specific type of verbal in- formation will be reciprocated with decreasing levels of the same variable. 55 Third, primarily by extrapolation from research on related variables, interactants will tend to adopt and reflect what they perceive to be the counterpart's emo- tional state as inferred from affect displays and other relevant nonverbal cues. Bales (1950) provides indirect evidence for this hypothesis from his analysis of response sequences observed in small groups. For example, stimulus joking behaviors were frequently followed by joking responses. In dyadic contexts, Condon and Ogston (1966) produced evidence for "interactional synchrony" between source and receiver in terms of small movements of hands, head and eyes. Kendon (1968) observed listeners giving continuous kinesic commentary on a speaker's performance, consisting of headnods and changes in facial expression. Nonverbal response matching has also been observed in psy— chotherapeutic dyads with respect to smiling, gestures and posture (e.g., Rosenfeld, 1967; Scheflen, 1965). Nonverbal messages are of central theoretic import in this context due to the type of information they convey. Bettinghaus (1973) suggests that nonverbal communciation serves most effectively to reveal information about the communicator per se, rather than about ideas he/she may espouse. Of the five gestural categories classified by Ekman and Friesen (1969), affect displays are particularly important here because they provide access to private emo- tional states. Hence, they represent a potentially critical 56 form of self-disclosure. Theoretically then, we would expect affect displays to convey valuable information pri- marily at the psychological level and to be an integral part of the information transfer/uncertainty reduction par- adigm under consideration. By the same token, response matching patterns involving other informative nonverbal acts can be expected to play a crucial role in the manage- ment, control and transfer of salient interpersonal information. The argument here is that response matching patterns observed on several types of nonverbal cues (as reviewed above) warrant the theoretic expectation that reciprocity occurs with respect to all communicative/informative non- verbal acts included by definition in the larger domain, information transfer. The proposition to follow implies that the number of specific emblems, illustrators, affect displays and regulators emitted by individual interactants in initial dyadic encounters will covary in the same direction. Axiom 11: As level of a specific type of nonverbal information transferred by either interactant increases, level of the same type of nonverbal information transferred by the counterpart will increase. Decreasing levels of a specific type of nonverbal information will be reciprocated with decreasing levels of the same variable. At this juncture, focus of the present theoretic devel- opment turns more directly to handicapped/nonhandicapped interactions. Reference was made earlier to a series of 57 programmatic studies indicating the serious "strain in social interaction" present during initial face-to-face encounters between these individual types. This strain is believed to result from prior attitudes of participants toward each other which give rise to uncertainty about the most appropriate communicative strategy for maximizing the quality of their transactions. For instance, Siller 33 31. (1967) found that nonhandicapped persons participating in their research invariably reported: . . . uncertainty about how to comport them- selves, e.g., whether to take notice of the disability or not. Frequently and readily, people admit a compulsion to stare, fully realizing that the disabled person would be made uncomfortable by this, and that they, in turn, would be branded as cruel or social- ly inconsiderate for indulging their curiosity (p. 64). Goffman (1963) discusses interpersonal interactions with stigmatized persons such as the visibly handicapped in terms of the "pathology of interaction-uneasiness." He describes this condition as "accompanied by one or more of the familiar signs of discomfort and stickiness: the guarded references, the common everyday words suddenly made taboo, the fixed stare elsewhere, the artificial levity, the compulsive loquaciousness, the awkward solemnity" (p. 19). In a similar vein, Davis (1961) noted that this type of personal interaction gives rise to anxiety result- ing from an imputation of the handicapped person which expresses itself . . . "in the embarrassment of the normal 58 by which he conveys the all too obvious message that he is having difficulty in relating to the handicapped person as he would to 'just an ordinary man or woman'" (p. 121). Limited research to date actually examining handicapped/ nonhandicapped interactions generally comports with the previous citings (Fontes, Miller & Kaminski, 1978). As reported at the outset, nonhandicapped individuals tend to increase their physical distance (Kleck, 1969; Worthington, 1974), terminate the interaction sooner and exhibit less variance in their verbal behavior (Kleck, Ono & Hastorf, 1966), demonstrate greater motoric inhibition, i.e., restrict their nonverbal behavior (Kleck, 1968), and manifest more anxiety (Marinelli & Kelz, 1973; Marinelli, 1974) when interacting with handicapped persons. The non- handicapped also report more emotional discomfort following this type of interaction (Kleck, 1966). On present evidence, it is conceivable that the uneasi— ness, apparent ambivalence and strain observed repeatedly when handicapped/nonhandicapped persons interact are essen— tially akin to, and have their source in, communicative uncertainty as described theoretically above with reference to initial interactions in general. Consequently, it is assumed that the uncertainty reduction perspective proposed earlier can be applied to, and will facilitate basic research on, interactions involving handicapped persons. Differences observed between this type of interaction and 59 the general case are expected to be quantitative rather than qualitative. If communicative uncertainty is ordinarily high at the outset whenever strangers interact (Berger & Calabrese, 1975), this condition should be exacerbated when either person bears a visible disability. On that assumption plus prior evidence, the following proposition seems tenable. Axiom 12: Initial interactions between handicapped and nonhandicapped persons are characterized by atypically high levels of communicative uncertainty. Extending the preceding rationale, it is reasonable to further assume that high uncertainty gives rise to the high anxiety previously observed (e.g., Marinelli, 1974) in interactions of this type. Hence, the expectation that uncertainty and state anxiety (Marinelli & Kelz, 1973; Spielberger, 1966) will covary in the same direction. The following proposition refers specifically to initial inter- actions between handicapped and nonhandicapped persons. Axiom 13: As level of communicative uncertainty experienced by either interactant decreases, level of state anxiety experienced by the same interac- tant will decrease. Increasing levels of state anxiety will accompany increasing levels of com- municative uncertainty experienced by the same interactant. In support of Axiom ll, evidence was cited suggesting that the number of specific affect displays emitted by individuals in dyadic encounters will tend to covary in the 60 same direction. This presupposes, of course, that a given set of nonverbal cues are perceived and identified with a particular emotional state by the counterpart, perhaps at a low level of awareness. Thus, to the extent that anxiety is typically accompanied by observable and interpretable affect displays, one would expect the level of state anx- iety experienced by a given interactant to covary with the counterpart's perception of these anxiety-revealing cues. That is, changes in level of state anxiety experienced by either person should be accompanied by similar directional changes in the counterpart's attribution of this emotional state to the source. Further research in this area would likely enlighten our understanding of the anxiety syndrome that apparently occurs when handicapped/nonhandicapped dyads initially interact. In conjunction with Axiom 13, the prior associ- ation between state anxiety and level of that affect per- ceived by the counterpart (hereafter labeled perceived state anxiety) leads to the following prOposition. Axiom 14: As level of communicative uncertainty experienced by either interactant decreases, level of perceived state anxiety (by the counterpart) will decrease. Increasing levels of perceived state anx— iety will accompany increasing levels of communicative uncertainty experienced by either interactant. In support of Axioms 4 and 5, it was reasoned that af- filiation is a prerequisite type of positive reinforcement through which social interaction is initiated and ongoing relationships involving other types of information exchange 61 are encouraged. Intuitively, affiliative information might then be considered a particularly important source of uncer- tainty reduction in initial encounters, determining whether or not the interaction will continue. More specifically, as affiliative information from a given source increases, the receiver's uncertainty should be reduced, particularly with respect to how he/she is being evaluated by the source. Assuming that affiliative cues normally signal a favorable disposition, as they increase we would expect the receiver to perceive himself/herself as more interpersonally attractive to the source. Essentially, foregoing arguments amount to the assertion that affiliation is (1) a necessary precondition for uncer- tainty reduction, in general, involving other types of information and (2) a sufficient condition for reducing receiver uncertainty with specific reference to his/her evaluation by the source. On these accounts, an inverse relationship is posited between uncertainty and perceived interpersonal attraction for initial encounters in handi- capped/nonhandicapped dyads. The latter variable refers to the extent to which a given interactant perceives himself/ herself as interpersonally attractive to the counterpart. Axiom 15: As level of communicative uncertainty experienced by either interactant decreases, the same interactant's perceived interpersonal attrac- tion to the counterpart will increase. Decreasing levels of perceived interpersonal attraction will accompany increasing levels of communicative uncer- tanity experienced by the same interactant. 62 Counter the rationale advanced in support of Axiom 15, it may be argued that negative reinforcement or nonaffil— iative (as well as affiliative) cues can be equally effec- tive toward reducing receiver uncertainty in the opposite direction with regard to how he/she is being evaluated. That is, recipients of nonaffiliative cues could also be certain of negative evaluation. Thus, both high and low levels of perceived interpersonal attraction could accompany low levels of uncertainty, suggesting a curvilinear relation. The simpler, linear association was posited on two accounts. First, the rudimentary concept of approach - avoidance underlying present use of the term affiliation does not imply an overtly negative counterpart for affiliative behavior. That is, casual observation suggests that aygig- 3293 can usefully be defined as simply a failure to approach. Likewise, nonaffiliation can meaningfully refer to the absence of affiliative cues when unfamiliar persons meet. In which case, nonaffiliation would not ordinarily reduce uncertainty and could conceivably increase it. Second, the writer suspects that nonaffiliation (or the absence of affiliative cues), as opposed to overt negative communication, does frequently occur in initial encounters, particularly those involving handicapped/nonhandicapped persons. There is some evidence suggesting that overt neg- ative reactions such as straightforward verbal rejection of the handicapped are culturally inhibited (e.g., Siller et 21., 63 1967; Kleck, 1966) and do not ordinarily occur in interac- tions of this type. More likely, meaningful interactions fail to develop and termination occurs prematurely (Kleck, Ono & Hastorf, 1966) due to the absence of an apprOpriate level of affiliative information rather than the presence of distinctly negative information. Hence, Axiom 15 seems the more theoretically sound, heuristically promising approach. One additional proposition completes the prop- sitional framework from which a limited number of possible hypotheses will be generated in Part IV. On the common assumption that people tend to like those who like them (Brooks & Emmert, 1976), it is reasonable to expect that people are also initially attracted to others who appear to be attracted to them. More specifically, as a receiver's perception of his/her attractiveness to a mes- sage source increases, that source will become more inter- personally attractive to the receiver. Thus, perceived interpersonal attraction (i.e., the attractiveness of R to S as perceived by R) and interpersonal attraction (i.e., the attractiveness of §.t° R) are expected to covary in the same direction. Consequently, given the inverse associa— tion posited in Axiom 15, an inverse relationship between uncertainty and interpersonal attraction is also logically warranted. Axiom 16: As level of communicative uncertainty experienced by either interactant decreases, the same interactant's interpersonal attraction to the 64 counterpart will increase. Decreasing levels of interpersonal attraction will accompany increasing levels of communicative uncertainty experienced by the same interactant. édwxgg, Testable Theorems j Science does not aim, primarily,at high probabilities. It aims at high informa- tive content, well backed by experience. But a hypothesis may be very probable simply because it tells us nothing, or very little (Popper, 1955:146). Presumably, Popper does not mean that the theorist can or should avoid all concern with probability. Axio- matic foundations in the prior section rely heavily on existing theory and prior evidence. To that extent, the following hypotheses do aim at high probabilities. Given current literature in this area, they relate fundamental variables most likely to inform development of more effective commun- ication strategies for both handicapped and nonhandicapped persons. In other words, these basic relationships should be examined empirically first, then discarded if necessary, before search is made for additional and/or other relevant variables. In the latter sense, theorems to follow are primarily concerned with informative content rather than high probability. Since this theory focuses specifically on handi— capped/nonhandicapped dyads, at least three independent 65 variables are implied: (l) verbal/nonverbal affiliation, (2) verbal/nonverbal congruency, and (3) type of dyadic interaction. In addition, prior axioms express relation- ships involving five dependent variables: (1) communica- tive certainty/uncertainty, (2) state anxiety, (3) perceived state anxiety, (4) perceived interpersonal attrac- tion, and (5) interpersonal attraction. Minimally then, the theory will allow generation of fifteen first-order predictions relating three independent variables to each of five dependent variables. Fifteen second-order (A x B) and five third-order (A X B X C) interaction effects are also possible. In the interest of brevity, the sample of fifteen theorems below provides first and second—order examples from which additional hypotheses can easily be constructed. It also reflects that set of theorems with which the writer expects to begin testing the present theory. Theorems 1 through 5 specify main effects predicted from the primary independent variable, verbal/nonverbal affiliation (abbreviated as V/NV affiliation). Each theorem was deduced from two or more of the prior axioms identified in parentheses. All hypothesized relations to follow presuppose Axiom l and refer to initial dyadic encounters between handicapped/nonhandicapped persons. Theorem 1: Interactants exposed to increasing levels of V/NV affiliation will report increasing levels of communicative certainty (Axioms 4 and 5). 66 Theorem 2: Interactants exposed to increasing levels of V/NV affiliation will report decreasing levels of state anxiety (Axioms 4, 5 and 13). Theorem 3: Interactants exposed to increasing levels of V/NV affiliation will report decreasing levels of perceived state anxiety (Axioms 4, 5 l4). Theorem 4: Interactants exposed to increasing levels of V/NV affiliation will report increasing levels of perceived interpersonal attraction from the message source (Axioms 4, 5 and 15). Theorem 5: Interactants exposed to increasing levels of V/NV affiliation will report increasing levels of interpersonal attraction toward the message source (Axioms 4, 5 and 16). Similar hypotheses predicting main effects from the second independent variable can be generated by substitut- ing "V/NV congruency" for "V/NV affiliation" in Theorems 1 through 5. Relevant axioms, in this case, are l, 8, 13, 14, 15 and 16. Theorems 6 through 10 below identify main effects (holding affiliation constant) predicted as a func- tion of the specific type of dyadic interaction to which this theory applies (i.e., as a consequence of the fact that one interactant is handicapped). In essence, given similar amounts of information transfer, the theory pre- dicts proportionately less uncertainty reduction in handi- capped/nonhandicapped dyads since interactions of this type are characterized by atypically high levels of uncertainty at the outset (Axiom 12). Theoretically, other variables depending on uncertainty reduction are expected to follow suit in either a positive or negative direction according 67 to prior predictions. Again, all subsequent theorems pre- suppose Axiom 1. Theorem 6: Nonhandicapped persons will report more communicative certainty when exposed to the same level of V7NV affiliation generated by a nonhandicapped, as opposed to a handi- capped, source (Axiom 12). Theorem 7: Nonhandicapped persons will report less state anxiety when exposed to the same level of V/NV affiliation generated by a nonhandicapped, as Opposed to a handicapped, source (Axioms 12 and 13). Theorem 8: Nonhandicapped persons will report less perceived state anxiety when exposed to the same level of V7NV affiliation generated by a nonhandicapped, as opposed to a handicapped, source (Axioms 12 and 14). Theorem 9: Nonhandicapped persons will report more perceived interpersonal attraction (from the counterpart) when exposed to the same level of V/NV affiliation generated by a nonhandicapped, as opposed to a handicapped, source (Axioms 12 and 15). Theorem 10: Nonhandicapped persons will report more interpersonal attraction (toward the counter- part) when exposed to the same level of V/NV affiliation generated by a nonhandicapped, as opposed to a handicapped, source (Axioms 12 and 16). The foregoing five theorems focus on reactions of the nonhandicapped interactant. However, given Axiom 12, there is no reason to attribute these reactions to one interactant more so than the other. Hence, the theory allows generation of parallel predictions focusing on reactions of the handicapped participant by reversing the terms "handicapped" and "nonhandicapped" in Theorems 6 through 10. 68 Five additional theorems complete the present sample. Theorems 11 through 15 posit second-order inter- action effects predicted from the two independent variables for which main effects were formalized in prior theorems (i.e., level of V/NV affiliation and type of dyadic inter- action). In lieu of additional information, our theory leads to the expectation that these predictors will inter- act in an ordinal manner (see Keppel, 1973), producing conjoint effects on communicative certainty and other theo- retically related variables. This general interaction hy- pothesis is advanced on two counts. First, systematic differences between two types of dyads (Theorems 6 through 10) were predicted primarily on account of atypically high levels of uncertainty expected when handicapped/nonhandicapped persons initially interact. Holding V/NV affiliation constant, we reasoned that pro- portionately less uncertainty reduction would occur in con- texts where uncertainty is characteristically higher at the outset. Second, Axiom 4 posits a reciprocal relationship between verbal affiliation and information transfer. Axiom 5 posits a similar reciprocal association for non- verbal affiliation. In combination, Axioms 4 and 5 imply a reciprocal association between level of V/NV affiliation and information transfer. Thus, as information transfer initially increases, further concomitant increases in V/NV affiliation are also expected to occur. 69 If this is the case, information transfer should increase at an accelerating rate over time, particularly in initial interactions. As a result, given Axiom 1, un- certainty reduction should also occur at an accelerating rate. On this rationale, anticipated conjoint effects of the predictors at hand are twofold: (1) a given input of V/NV affiliation will initially result in proportionately less uncertainty reduction for handicapped/nonhandicapped interactants, and (2) given similar rates of acceleration and proportionately higher levels of uncertainty, overall uncertainty reduction should occur less rapidly in handi- capped/nonhandicapped dyads (as opposed to nonhandicapped dyads) during a comparable period of time. In turn, main effects predicted from V/NV affiliation on other dependent variables previously related to uncertainty should in- crease or decrease less rapidly in dyads involving the handicapped. Ordinal interactions predicted in Theorems 11 through 15 will be clarified by means of the matrix pre- sented in Figure 2. Suggestive of a factorial design, this matrix illustrates the simultaneous crossing of three levels of V/NV affiliation with two levels of the second independent variable, type of interaction. Level of V/NV_Affi1iation Type of * Interaction high medium low NH/NH l 2 3 H/NH 4 5 6 Note: NH/NH H/NH Figure 2. nonhandicapped/nonhandicapped dyads handicapped/nonhandicapped dyads Dyadic Interaction. Representation of a Factorial Design Crossing Three Levels of Verbal/Non- verbal Affiliation With Two Types of dicted where: 71 The pattern of mean differences in effect size pre- via each theorem will be identified symbolically xl u mean for nonhandicapped persons exposed to a high level of V/NV affiliation generated by a nonhandicapped source; xl u mean for nonhandicapped persons exposed to a medium level of V/NV affiliation gener- ated by a nonhandicapped source; xl u mean for nonhandicapped persons exposed to a low level of V/NV affiliation generated by a nonhandicapped source; xl u mean for nonhandicapped persons exposed to a high level of V/NV affiliation generated by a handicapped source; xl u mean for nonhandicapped persons exposed to a medium level of V/NV affiliation gener- ated by a handicapped source; Y = mean for nonhandicapped persons exposed to a low level of V/NV affiliation generated by a handicapped source. These theorems assume the same rank order of effects pre- dicted previously in Theorems 1 through 10. Each presup- poses Axiom 1. As nonhandicapped persons are exposed to increas- ing levels of V/NV affiliation generated by handicapped versus nonhandicapped sources . . . Theorem 11: Their reported levels of communica- five certainty will increase more rapidly when the source is a nonhandicapped person (Axioms 4, 5 and 12). (X1 - X2) (i - Y ) > (Y — Y ) N DJ 72 Theorem 12: Their reported levels of state anxiety will decrease more rapidly when the source is a nonhandicapped person (Axioms 4, 5, 12 and 13). Theorem 13: Their reported levels of perceived state anxiety will decrease more rapidly when the source is a nonhandicapped person (Axioms 4, 5, 12 and 14). Theorem 14: Their reported levels of perceived interpersonal attraction (from the counterpart) will increase more rapidly when the source is a nonhandicapped person (Axioms 4, 5, 12 and 15). Theorem 15: Their reported levels of inter- personal attraction (toward the counterpart) will increase more rapidly when the source is a nonhandicapped person (Axioms 4, 5, 12 and 16). 1 ' X2) > (X4 ' x5) (X - X ) > (X5 - X6) The rank order of effects and ordinal interactions predicted in Theorems 11 through 15 are graphically illus- trated as shown in Figures 3 and 4 below. 73 NH/NH Dyads ~ ~~.‘ ._ “~ X1 ~‘s‘ §‘~ .‘ .— ‘~§‘ X2 ‘~~~~ § 0“ H/NH Dyads — ~ 2.4 -. .--..~--- — -- .--- x5 _ X6 I ' j j high medium low Level of V/NV Affiliation Figure 3. Pattern of Means Showing Rank Order of Effects and Interactions Predicted in Theorems ll, 14 and 15. ‘-.o" H/ NH Dyads ”w . -——"' _ o‘.o"‘-‘ _ x6 .. X5 .4#' X4 "'o l”' X3 o" o". o ._ NH/NH Dyads ’ ,¢ X2 ” af.' ' — x1 1 .4 1 high medium low Level of V/NV Affiliation Figure 4. Pattern of Means Showing Rank Order of Effects and Interactions Predicted in Theorems 12 and 13. CHAPTER II THE METHOD Conceptual variables and corresponding operations required for the proposed study are the initial concern of this chapter. Attention is first given to experimental variables implied in prior hypotheses. The Independent Variables Definitions and Operations presented here anticipate a two-way factorial design to be discussed later in this paper. Effects of two independent factors are to be ob- served: (1) level of verbal/nonverbal affiliation and (2) type of dyadic interaction. The first factor is a composite index of several verbal and nonverbal variables. That is, level of verbal/nonverbal affiliation is a combination of eight affiliative acts emanating simultaneously across the two channels. First to be defined are four nonverbal indi— cators of affiliation drawn directly from the Mehrabian/ Ksionzky (1972) listing in Table 1 above. These four vari- ables comprise the nonverbal component of the first factor. A dichotomous manipulation will be used within each channel. Thus, operations to follow describe a relatively high positive level of each variable labeled the affiliative treatment; a relatively low positive (or neutral) level of each variable will constitute the nonaffiliative treatment. 74 75 With some modification due to our use of these variables as stimuli, their definitions are consistent with those listed by Mehrabian (1972b, Appendix A). Reliability estimates provided by Mehrabian are shown in parentheses. A. Nonverbal Indicators of Affiliation 1. Duration of Eye Contact (0.65) Portion of the interaction between specific dyads when one participant focuses gaze upon head area or eyes of the counterpart, regardless of the latter's gaze direction. (This measure is to be varied in the present study primarily as a function of which participant is speaking.) Operations by condition: a. Affiliative - The stimulus message source maintains continuous head area or eye contact with target subject (8) during all verbaliza- tions by S. He/she alters eye gaze direction away from S at natural locations during his/ her own verbalizations (e.g., while recalling information from past history or formulating a thoughtful response). Stimulus person diverts eye gaze approximately 50 percent of the time while he/she is speaking. Appropriate locations for changes in eye gaze direction are predetermined via the standardized inter- view format so as to establish the desired ratio. b. Nonaffiliative - The stimulus source person maintains head area or eye contact with S approx- imately 50 percent of the time during verbaliza- 'tions by S. During his/her own speech production, he/she diverts eye gaze away from S as described in the previous condition. — 2. Head-Nodding Movements (0.90) Cyclical up and down movements of the head commonly used to signal acknowledgment, agreement or appro- bation. Head nods must be clearly directed toward addressee either during or in response to verbaliza- tions by the latter. Operations by condition: a. Affiliative - Moderate use of natural head- nodding movements by the stimulus source person at appropriate locations as predetermined by content of the verbal exchange. He/she presents an average 76 of three (3) head nods (3 seconds or less duration) per minute throughout the inter- action. b. Nonaffiliative - Use of head-nodding move- ments by the stimulus source person is eliminated entirely from the interaction. 3. Hand and Arm Gestures (0.94) Side-to-side, forward-backward, up-and-down, or cyclical movements of hands and arms (including use of fingers) as in natural gesticulation commonly used to illustrate, clarify and/or support the verbal message. Operations by condition: a. Affiliative - Moderate use of natural hand/ arm gestures by the stimulus source person at apprOpriate locations as predetermined by the content of the verbal exchange. He/she presents an average of two discrete gestural units per minute throughout the interaction. b. Nonaffiliative - Use of hand/arm gestures by the stimulus source person is eliminated entirely from the interaction. 4. Facial Pleasantness (0.79) Positive facial expression including natural variations from pleasant affect display to more active smiling be- havior commonly used to signal enjoyment, approbation and/or positive attitude. Operations by condition: a. Affiliative - Stimulus source person maintains distinctly pleasant (slightly positive) facial expressions throughout interaction. Facial affect display is varied to more definite smiling behavior at apprOpriate locations as predetermined by con- tent of the verbal exchange. Stimulus source person displays an average of two (2) discrete smiling units (3 seconds or less duration) per minute during the entire interaction. b. Nonaffiliative - Stimulus source person dis- plays apprOpriate, culturally conditioned smiling behavior during initial greeting exchange with S. Subsequently, he/she maintains neutral, business? like facial expression avoiding positive/negative affect display throughout the interaction. Occasionally, as predetermined by content of the verbal exchange, the stimulus source flashes a polite smile in response to obviously lighthearted or humorous statements by S. An average of one 77 discrete smiling unit (2 seconds or less duration) per four minute period of inter- action is displayed by the source of the stimulus message. B. Verbal Indicators of Affiliation Four implicit verbal variables were selected to com- prise the verbal component of the first factor under experi- mental control in this study. Since several of the verbal measures identified with affiliation by Mehrabian and Ksionzky (1972) are redundant indices of the gross amount of conversation produced by their subjects, only two verbal variables selected here were drawn from Table 1 above: (1) number of declarative statements and (2) positive verbal content. Two additional verbal indicators were suggested by results of the previous Mehrabian (1971a) study: (1) verbal reinforcers and (2) number of questions asked. The latter variable, number of questions asked, is modified conceptually here to more consistently represent the theo- retical perspective in Section I. Our version of this vari- able, labeled number of person - oriented questions, is based on the Miller and Steinberg (1975) conceptual dis- tinction between sociological and psychological level data. The following descriptions attempt to define four relatively exclusive types of implicit verbal information selected to comprise the verbal component of the verbal/ nonverbal affiliative stimulus. It is not assumed that an equal number of verbal and nonverbal indicators is suffi- cient to balance the intensity of the manipulation across 78 the two channels. 1. Number of Declarative Statements (0.90) The average number of simple declarative sentences and/or independent clauses spoken per minute by one participant during a complete dyadic interaction. Elliptical statements in which at least one verb form is either expressed (e.g., ". . . Makes sense to me") or clearly implied (e.g., ". . . Quite certain" implying in context, "I am quite certain") are included in this category. Expressions of positive content (as defined below) that meet the foregoing criteria are also included in the total number of declarative statements used to establish this average. Operations by condition: a. Affiliative - Using a standardized interview format, the interaction is predetermined to include an average of six (6) declarative state- ments per minute by the source of the stimulus message. b. Nonaffiliative - The stimulus source simu- lates a business like demeanor by expressing a limited number of declarative statements that function as topic transitions and/or substantive responses to questions and comments by the inter- viewer. He/she conveys comparatively less will- ingness to divulge information than is the case in the affiliative condition described above. The standardized interview is predetermined to include an average of four (4) declarative state- ments per minute by the stimulus source person. 2. Positive Verbal Content (0.86) The average number of discrete instances per minute when the choice of words used by one participant conveys a positive attitude toward the addressee or toward the interaction in progress with the addressee. Discrete instances of positive verbal content are identified as single words (e.g., "Great!"), ellipti- cal expressions (e.g., ". . . Interesting question!), or complete statements (e.g., "This has been most informative") expressed in response to messages pro- duced by the counterpart. Operations by condition: a. Affiliative - Using a standardized interview format, the interaction is predetermined to include an average of one and one-half (1.5) expressions of positive verbal content per minute by the stimulus source. 79 b. Nonaffiliative - Expressions of positive verbal content by the stimulus source are limited primarily to the formal introduction and closing of the interview. The standardized interview is predetermined to include an average of one (1) positive verbal expression by the stimulus source per each two minute period of interaction. 3. Verbal Reinforcers Brief vocalizations that express acknowledgment or understanding of messages produced by the addressee. Verbal reinforcers are common vocal responses that may or may not represent words generally accepted as part of the standardized usage (e.g., "uh-huh," "hmmm," "Okay," or "Really!"). Verbal reinforcers usually occur during or immediately after messages produced by the counterpart to whom the reinforcement is addressed. Operations by condition: a. Affiliative - The standardized interview is predetermined to include an average of three (3) verbal reinforcers by the stimulus source person during each two minute period of interaction. b. Nonaffiliative - Use of verbal reinforcers by the stimulus source is negligible in this condition (i.e., their use is eliminated with the possible exception of instances in the verbal exchange where the complete absence of some verbal acknowledgment would normally be perceived as awkward or impolite). 4. Number of Person - Oriented Questions Based on the Miller and Steinberg (1975) conceptual- ization of interpersonal communication, person- oriented questions are designed to elicit idiosyncratic, psycholOgical level information for the purpose of distinguishing the addressee as a unique individual from other persons who hold membership in the same or similar social groups (e.g., age, sex, ethnic affilia- tion, socioeconomic status, culture, political affil- iation, etc.). For present purposes, this variable is defined as the average number of such questions raised by one participant during dyadic interaction over a specified period of time. In an employment interview context, a question of this type might solicit information regarding the addressee's personal perspective on a current, business- related issue e.g., "I was just wondering what you see as the number one problem or challenge in this business today?" Operations by condition: 80 a. Affiliative - The standardized interview is designed to present an average of one (1) person— oriented question by the stimulus source person per each two-minute period of interaction. b. Nonaffiliative - Person-oriented questions are eliminated entirely in this condition. Each personal question appearing in the affiliative verbal version of the standardized interview is de-personalized by shifting the focus of each request to company policies and/or perspectives, e.g., "I was just wondering what SRA sees as the number one problem or challenge in this business today?" To this point, four nonverbal and four verbal variables have been defined conceptually; two levels of control, labeled affiliative and nonaffiliative, were Operationalized for each of the eight variables. Within each channel the four variables will be combined at the same level of control to establish two composite indices of affiliative informa- tion. Hence, in combination, the four nonverbals controlled at the affiliative level will comprise the affiliative non- verbal component of the first factor under observation here. Likewise, the same variables controlled at the nonaffiliative level will comprise a nonaffiliative nonverbal component of the first factor. Similarly, the verbal variables are combined at each level of control to establish the affiliative verbal and nonaffiliative verbal components of the same experimental factor. The second experimental factor to be introduced, type of dyadic interaction, is designed to isolate differential effects of communication resulting specifically from 81 initial interactions between handicapped/nonhandicapped persons as opposed to similar transactions between non- handicapped individuals. Effects of varying levels of verbal/nonverbal affiliation will be compared across two types of initial dyadic interactions: (1) nonhandicapped with nonhandicapped persons (NH/NH) and (2) nonhandicapped with handicapped persons (NH/H). The distinguishing feature of these two types of interaction, handicapped persons, required further definition. Review of relevant literature indicates that a stand- ardized definition of the term "handicap" does not exist to date. Confusion over the meaning of the term becomes even greater when "impaired" and "disabled" are used as synonyms for "handicapped." The Urban Institutes Rgport on the Com- prehensive Service Needs Study (RCSNS, 1975) points out that relevant literature on problems encountered by the handi- capped contains ". . . a melange of uses of the terms 'im— pairment,’ 'disability,' and 'handicap'; and a melange of programs and purposes. It seems that the terms disability and handicap can mean whatever one wants them to mean" (pp. 21-22). In lieu of precise definition, one finds a plethora of categorical systems for classifying various handicaps, develoPed primarily by social agencies. These classifica- tion systems vary substantially depending on the objectives of a given organization (cf. Wilder, 1971, 1975; Allen & 82 Cinsky, 1972; Rehabilitation Service Manual, 1976). Relative to the focus of this research, the defini- tional problem is compounded by the fact that persons with the same or similar handicaps may vary significantly in terms of the degree of neurological damage, motoric restric- tion, and consequently, their ability to encode verbal/non- verbal messages. Conversely, persons with radically differ- ent disabilities (by various conceptual schemes) may experience similar communication problems. Hence, it is the writer's view that extant definitions and classification systems are of little value for defining the type of dyadic interaction presently under study. As a result, descriptive parameters based primarily on the functional criteria described below were established to identify a limited segment of the handicapped population. Two types of disability representing opposite extremes on the functional-cosmetic continuum developed by Siller, Ferguson, Vann and Holland (1967) will be combined to repre- sent the handicap stimulus in our nonhandicapped/handicapped interactions. First, facial distortion was selected because it is believed to rank highest among visible disabilities on anxiety arousal and aversion to interaction for nonhandi- capped persons (Siller, Chipman, Ferguson & Vann, 1967). Second, amputation was selected because amputees are typical- ly perceived (l) as less dependent on others, (2) as more "normal" in their personal development, and (3) as suffering 83 fewer social consequences than persons with congenital defects (Siller, Chipman, Ferguson & Vann, 1967). In moderate degree, this combination of visible dis- abilities (facial distortion with amputation of one lower limb) was selected to convey a general impression of dis— ability that might represent a variety of specific physical limitations (e.g., hemiplegia) without generating an ex- tremely aversive stimulus. For purposes of the present study, a handicapped person is defined as an individual with the following de- scriptive and functional characteristics: 1. Slight facial distortion potentially capable of interfering with the process of encoding verbal/ nonverbal messages 2. Distinct physical disability potentially capable of inhibiting the process of encoding nonverbal messages (e.g., as in the general case of ampu- tees, hemipelegics, parapelegics, etc.) 3. Normal psychological development and use of mental faculties 4. Normal sight and hearing ability 5. Normal speech production (i.e., without discern— ible speech impediment) Conversely, a nonhandicapped person is defined as an individual who shares the latter three functional normal- ities described above, without facial disfigurement or dis- cernible physical disability of any type. Operations by condition: 1. Nonhandicapped/Handicapped (NH/H) Handicapped person, a right leg amputee (below the kneecap), is the stimulus message source. He/she 84 is presented, and remains throughout the inter- action, in a hand-driven wheelchair. Slight muscular distortion of the face (right side) is discernible from the corner of the mouth to the cheekbone. The latter affect can be simulated by voluntary contraction of the cheek muscle or use of liquid collodion. Interacting counter- part is a nonhandicapped person. 2. Nonhandicapped/Nonhandicapped (NH/NH) Both the stimulus message source and the inter- acting counterpart are nonhandicapped persons (as defined above) in this condition. Preceding operations which define two levels of the second factor, complete the description of relevant inde— pendent variables. The Dependent Variables Extensive rationale supporting the selection of five dependent variables, implied in prior research hypotheses, was provided in the theoretical portions of this paper. First, level of uncertainty is a central intervening vari- able in the uncertainty reduction paradigm presented there. This variable is expected to provide a definitive link be- tween varying amounts of affiliative communication and the remaining four dependent measures presently under observa- tion (cf. Figure 1 above): (1) level of state-anxiety, (2) perceived level of state-anxiety re: the counterpart, (3) interpersonal attraction, and (4) perceived interpersonal attraction re: the counterpart. Second, the degree of covariation between the two anxiety measures is expected to shed light on the recipro- cal effects of anxiety-revealing cues and the escalation 85 syndrome theorized earlier. Third, since the selected stimulus indicators of affiliation have been shown repeatedly to be primary refer— ents of social interaction, the two measures of inter- personal attraction are expected to provide further insight into the anxiety-reducing effects of varying levels of verbal/nonverbal affiliative communication. The following conceptual and operational definitions delineate use of these variables in the proposed study. 1. Level of Uncertainty: Uncertainty is defined as the absence of relevant information used by a message source to: (l) predict the most likely alternative among plausible responses the addressee may give, (2) select the most likely alternative among plausible interpretations for a response that has been given by the addressee, and (3) ultimately control the effects of his/her own communicative efforts in favor of desired outcomes. Measurement: Level of uncertainty is measured by means of a 10 item Likertetype scale, an adaptation and extension of the 7 item CLUES instrument developed by Clatterbuck (1979) to measure attributional confidence. As modified here (cf. Appendix B, items 1-10), the instrument presents ten complete sentences in question form (e.g., How are you of your general ability to predict the way this person is likely to behave in different situations?). Each item offers a 6-point response field ranging from "very 86 uncertain." Items 1, 3, 4, 5 and 7 were adopted, with minor modification, directly from the Clatterbuck CL7 instrument. Five additional items were generated in an attempt to increase likelihood of valid/reliable measurement, given the relatively recent and untried instrument developed by Clatterbuck. Item 2 was designed to tap the retroactive mode of the uncertainty construct, i.e., it represents a general measure of confidence in interpreting responses pro- vided by the addressee. Item 6 focuses directly on the pro- active mode of this variable requesting an assessment of confidence in the predictability of the addressee's re- sponses to communication in general. Items 8, 9 and 10 are more specific questions related to the topic of the last item included in the CL7 instrument (How well do you know him/her?). For example, item 8 asks, "How certain are you about whether or not this person is highly motivated toward his/her work?" Individual responses to each item are coded on a scale from one to six (1 = very uncertain, 6 = very certain) and summed across the ten items to provide a gross measure of uncertainty for each respondent ranging from 10 to 60. —;fiy 2. Level of State-Anxiety: State—anxiety is defined .05). Male participants averaged 23.40 years of age (s.d. = .198). Mean age of the female segment was 118 119 21.66 years (s.d. = 2.99). This sex by age difference was statistically significant via a two-tailed T-test (T = 2.86, df = 142, p < .05). Random assignment of male participants by age was assessed using two-way ANOVA. Results were nonsignificant (F < 1.0, df = 3, p > .05). The distribution of females by age across treatment groups was examined in the same manner with a similar result (F < 1.0, df = 3, p > .05). The sample averaged 15.70 years of formal education (s.d. = 1.60). Again, two-way ANOVA was used to assess the assignment of all subjects by formal education. No signifi- cant differences occurred (F = 1.86, df = 3, p > .05). Male participants reported more formal education (§'= 16.05, s.d. = 1.68) than female participants (i = 15.36, s.d. = 1.45). This sex by education difference was statistically significant via results of a two-tailed T-test (T = 2.64, (df = 141, p < .05). Level of education for male participants did not differ significantly across treatment groups as indicated by two-way ANOVA (F < 1.0, df = 3, p > .05). Results Of the same test did reveal significant differences in level of education for female participants across treatment groups (F = 2.97, df = 3, p < .05). Marital status was initially observed via four cate- gories with the following results: 23 persons were married, 117 single, one widowed and three divorced. The latter two 120 categories were collapsed with the single type due to small frequencies. Chi-square for k-independent samples was used to assess the distribution of single persons across treat- ment groups. No significant differences emerged (X2 = 1.257, df = 2, p > .05). Small frequencies in several experimental groups constrained use of Fisher's exact probability test to assess the overall distribution of married persons. Results were significant (p < .05) indicating nonrandom assignment of married persons across treatment groups. No significant sex by marital status differences were found using chi-square for two independent samples (X2 = 3.31, df = l, p > .05). Fisher's exact probablility test was used to assess the distribution of married females. Results were nonsignificant (p > .05). The same test indi- cated significant differences in the distribution of married males across treatment groups (p < .05). Results of the chi-square for k-independent samples revealed no significant differences in the distribution of single males (x2 = 1.93, df = 2, p > .05) or single females (x2 < 1.0, df = 2, p > .05) across treatment groups. PriOr employment experience involving handicapped persons was considered critical as a sample characteristic potentially capable of confounding results of this study. Consequently, participants were asked to indicate whether or not they had previously worked in an organization providing 121 professional services for the handicapped (cf. Appendix B, item 75). Thirty-three persons included in the sample (22.9%) provided an affirmative response. Hence, this characteristic was scrutinized in several ways. First, the distribution of all respondents reporting some work eXperience with the handicapped was examined via chi-square for k-independent samples. Frequencies relative to this variable did not differ significantly across treat- ment groups (X2 = 2.60, df = 2, p > .05). Second, the sample was stratified by sex to ascertain differences occurring between male/female participants with regard to employment involving handicapped persons. Results of the chi—square test for one sample were nonsignificant (x2 < 1.0, df = 1, p > .05). Third, random assignment of females with employment experience of this type was assessed using chi-square for k-independent samples. Again, no significant differences across treatment groups emerged (X2 = 1.06, df = 2, p > .05). The identical chi-square produced similar nonsignificant results for the male segment of the sample (X2 = 2.55, df = 2, p > .05). Participants' priOr social experience with handi- capped persons was considered equally important as a poten- tial confounding factor. Questionnaire item 79 asked participants to indicate whether or not they had previously maintained a regular social relationship with handicapped 122 individual(s). Fifty persons (34.7%) provided an affirm- ative response. Random assignment of this variable for the overall sample was examined via chi-square for k- independent samples. No significant differences across treatment groups were found (X2 < 1.0, df = 2, p > .05). Results of the chi-square for one sample indicated no significant differences between male/female participants relative to prior social relationships with the handicapped (X2 < 1.0, df = 1, p > .05). Distribution of females reporting some social history of this type was assessed using chi-square for k-independent samples. No significant differences occurred (X2 < 1.0, df = 2, p > .05). The identical test for male participants was nonsignificant (x2 < 1.0, df = 2, p > .05). In sum, foregoing analyses indicate that all sample characteristics considered relevant were randomly distrib- uted with the following exceptions. First, male partici- pants were significantly older and had more formal education than the female segment of this sample. Second, level of education for female participants was unequally distributed across experimental groups. Third, married persons in general and married males in particular were nonrandomly assigned to treatment conditions. Implications of devia- ‘tions from random assignment will be examined and discussed post hoc. 123 Assessment of Experimental Manipulations Observations were made on two experimental factors: (A) level of verbal/nonverbal affiliation and (B) type of dyadic interaction. Given the attempt to manipulate eight variables simultaneously on Factor A, direct assessment of their success was considered particularly crucial for inter- preting results of this research. Thus, response patterns and the magnitude of effects attributable to Factor A are examined below in detail. Factor B manipulations were not assessed directly in an attempt to minimize response bias. The concern was to avoid direct reference to the physical condition of the stimulus source who appeared as an amputee occupying a wheel- chair in the handicapped/nonhandicapped treatment (H/NH) of Factor B. Given use of the videotape medium, hindsight suggests that some assessment of the extent to which this manipulation was effective would have been useful. Three levels of affiliative communication comprised Factor A. On theoretic grounds, treatments were ordered (left to right) according to predicted effects of high - medium - low magnitude. The following sequence of treatments was an integral part of the research design: (a) V-Aff/NV- Aff, (b) V-Nonaff/NV-Aff, and (c) V-Aff/NV-Nonaff (of. Figure 5, p. 31). Hence, determination of the success of these manipulations, as well as subsequent analyses, will depend in part on the extent to which the data approximate 124 this pattern. Three types of assessment were made relative to Factor A. First and foremost, prior treatments varied verbal/ nonverbal affiliation. As previously defined, affiliative communication is expected to effect perceptions of friendli- ness or sociability. Thus, first concern of the manipulation check was to measure perceptions of the stimulus source's sociability. A two-item scale was developed for this purpose: 1. How friendly did you perceive Mr. Holtrop to be during the interview? 2. How sociable did you perceive Mr. Holtrop to be during the interview? These parallel items were separated in the final question- naire; each provided a 4-point response field (cf. Appendix B, II, 71 and 72). Pearson's r produced an acceptable estimate of intern- ial consistency for the sociability scale (r = .70). Individ- ual responses were summed across scale items and analyzed via two-way ANOVA. Results indicated a significant main effect for Factor A (F = 18.02, df = 2, p < .01). Marginal means for Factor A in Table 2 clearly show the expected direction of effects. However, this outcome was complicated by an A x B disordinal interaction (F = 9.88, df = 2, p < .01). Hence, the main effect cannot be completely interpreted with- out reference to an accompanying interaction of this type (Keppel, 1973). 125 Table 2. Means Resulting From Manipulation Check Data- Ratings of Sociability Type of Level of V/NV Affiliation _ Interaction a b c X NH/NH (1) 6.50a 5.92ad 4.38 5.60 H/NH (2) 6.04ab 4.79c 5.33bcd 5.39 i 6.27 5.35 4.85 i=5-49 Note: Cell means subscripted with a common letter do not differ significantly. The disordinal interaction is reflected in an invert- ed order of means occurring at the low level of Factor A (condition c). All cell means in Table l comport with the expected pattern excepting that located at c, 2. Although nonsignificant, the difference between marginal means for Factor B is worth noting. On the average, the handicapped source was perceived as somewhat less sociable than the non- handicapped source during H/NH type interactions (condition 2). Using Scheffé's test, significant differences were found for 7 of 15 post-hoc comparisons between treatment means as identified by subscripts in Table 1. Alpha was set at .10 as recommended by Scheffé (1959). Tukey's test pro- duced exactly the same results (a = .05). These tests were chosen for their conservatism regarding experimentwise Type I error. Both tests were conducted in this case to compare results. Subsequent analyses will rely on the Tukey test to maintain a consistent alpha level at .05. The magnitude of 126 Factor A effects was estimated by means of eta squared (nz = .19), a moderately strong effect for social science data (Cohen, 1969). In addition to varying affiliative cues, Factor A manipulations reflect a second major concern of the research. In theory, handicapped persons of the type under study may be expected to exhibit fewer nonverbal cues in support of their verbal communication than physically able persons. At some level, this nonverbal deficit may lead to perceptions of incongruent V/NV communication. On prior research dis- cussed in Chapter I, perceptions of incongruency can be ex- pected to bear negative consequences, particularly in an employment interview. Hence, Factor A was also designed to vary the level of congruency between (affiliative) V/NV cues produced by the stimulus source. Condition 3 attempted to simulate natural communica- Vtion, i.e., an appropriate and spontaneous level of V/NV congruency in the given context. Condition 9 replicated a in the nonverbal band with a moderate reduction in the number of verbal cues exhibited. Again in theory, when incon- sistent V/NV cues occur simultaneously, nonverbal constitu- ents will dominate in determining the total message impact (e.g., Mehrabian & Ferris, 1967; Mehrabian & Wiener, 1967). Consequently, condition 2 was expected to effect only a moderate reduction in perceptions of congruency since the nonverbal constituent was comparable to that in condition a. 127 Condition 9 replicated a_in the verbal band with a moderate reduction in the number of nonverbal cues exhibited. As- suming a dominant nonverbal impact, condition 9 was expected to effect a low level of V/NV congruency. A two-item scale was constructed to assess the effec- tiveness of prior manipulations: 1. How consistent were Mr. Holtrop's facial expressions, gestures and body movements with the meaning of the words he spoke during the interview? 2. How dissimilar (different) were Mr. Holtrop's verbal messages from his non- verbal messages during the interview? These items were separated in the final questionnaire; each provided a 4-point response field (cf. Appendix B, II, 70 and 73). Pearson's r produced an acceptable, if not optimal, estimate of internal reliability for the congruency scale (r = .65). Individual responses were summed across scale items and analyzed via two-way ANOVA. A significant main effect in the expected direction was found for Factor A (F = 11.51, df = 2, p < .01). A x B interaction did not occur. However, Factor B also produced an unexpected main effect on perceptions of V/NV congruency (F = 4.66, df = l, p < .05). As shown in Table 3, all cell means, as well as column marginals representing Factor A, follow the pre— dicted high - medium - low pattern. However, the direction of effects generated by Factor B is of equal theoretic 128 import for purposes of the study. Table 3. Means Resulting From Manipulation Check Data - Ratings of V/NV Congruency. Type of Level of V/NV Affiliation _ Interaction a b c x NH/NH (1) 6.42a 5.75a 4.50b 5.56 H/NH (2) 6.67a 5.88a 5.71a 6.08 SE 6.54 5.81 5.10 i = 5.82 Note: Cell means subscripted with a common letter do not differ significantly. All cell means representing Factor B, condition 2 are higher than their counterparts in condition 1. Using the Tukey test, 5 of 15 post-hoc comparisons be- tween treatment means in Table 2 produced significant differ- ences (a = .05). Condition 3, l was the major source of variance, differing significantly from all five remaining Imeans. Magnitude of effects attributable to Factor A was moderate (112 = .14). The same index indicated a minor effect size for Factor B (n2 = .03). Assuming that incongruent V/NV messages can influence perceptions of the source's integrity (e.g., Ekman & Friesen, 1969; Mehrabian, 1971, 1972), a third assessment of Factor A was made involving the ethical dimension of source credibility. Again, a two-item scale was developed for this purpose: 1. How truthful do you think the responses provided by Mr. Holtrop were during the interview? 129 2. In general, how trustworthy do you think Mr. Holtrop is? These items were also separated in the complete questionnaire and included a 4-point response field (cf. Appendix B, II, 69 and 72). Results of the Pearson r, used as an estimate of internal reliability, were considered unacceptably low for valid measurement (r = .50). In the interest of additional information, individual responses were summed across scale items and subjected to two—way ANOVA. Neither interaction nor main effects were statistically significant. However, inspection of means reported in Table 4 suggests certain trends worth noting in these data. Table 4. Means Resulting From Manipulation Check Data - Ratings of Source Credibility Type of Level of V/NV Affiliation _ Interaction a b c X NH/NH (l) 6.50 6.04 5.71 6.08 H/NH (2) 6.38 6.38 6.46 6.40 i 6.44 6.21 6.08 i = 6.24 The column marginals indicate that Factor A treatments did generally influence credibility ratings in the expected direction although the effect was not strong. Row marginals (Factor B) and cell means in condition 2, again suggest generally higher credibility ratings for the handicapped source as in the previous analysis. 130 Several reasons for the weak effect found in these data are plausible. First, the scale in use was not ade- quately reliable. The unreliability was likely a function of providing only two items, each requesting a different type of information. Item 69 focused on the truthfulness of statements made during the interview; item 72 requested an evaluation of trustworthiness in general. Source credibility is a dynamic variable that may change rapidly depending on the amount and type of information available (Burgoon & Ruffner, 1978). Hence, given a relatively brief initial interaction, subjects may have had difficulty forming the more general judgment. Second, there are at least five known dimensions of source credibility, each of which tends to Operate inde- pendently (Burgoon & Ruffner, 1978). As defined here, affili- ation relates specifically to only one of these dimensions, Inamely sociability. The weak effects reflected in Table 3 may then be due to the fact that the scale in use focused exclusively on a different dimension of source credibility, oftenlabeled character. As measured in this case, source credibility was only remotely related to the substantive content of the original manipulations. In short, affiliative cues may typically have little effect on character evaluations in initial interactions. Third, differing levels of V/NV congruency were ex- pected to more directly influence credibility ratings. 131 However, lower levels of Factor A (b and 3) involved only a reduction in the number of verbal or nonverbal cues exhibited in condition a. Conceivably, the quantitative difference in either case was not large enough to signal a qualitative difference in messages emanating from the two channels. Stated differently, the absence of certain cues in support of the opposite channel may influence credibility to a lesser extent than the presence of clearly conflicting cues. In sum, prior analyses support the following conclusions: 1. The experimental manipulations of affilia- tive communication were relatively effec- tive in the intended direction. Magnitude of the effect was moderately strong. Comparable levels of V/NV affiliation were generally perceived as somewhat less soci- able when generated by a handicapped source although the difference was non- significant. One clear exception to this pattern occurred when the nonverbal message component was reduced (of. Table l,c). The experimental manipulations of V/NV congruency were relatively effective in the intended direction. Magnitude of the effect was moderate. Comparable levels of V/NV congruency were perceived as significantly more congruent ‘when generated by a handicapped source. Factor A manipulations produced a discern- ible, though nonsignificant, influence on the character dimension of source credibility in the intended direction. The handicapped source was perceived as somewhat more cred- ible although the difference was not statis- tically significant. 132 6. The major contribution of variance relative to Factor A came from a reduction in non- verbal affiliative communication. Reducing the number of verbal cues (condition 3 vs. 9) did not generally produce significant differences. One exception to this pattern is shown in Table 1, condition b, 2. 7. Factor B, type of interaction, had a signif- icant impact on perceptions of V/NV congruency and sociability, particularly when the non— verbal message component was reduced. Unexpected elements in these results regarding the influence of Factor B are of central theoretic import for primary objectives of this study. They are likely to shed considerable light on the interpretation of subsequent find- ings. It is intuitively reasonable to explain the present effects of Factor B in terms of different sets of expecta- tions operating when nonhandicapped persons engage the handi- capped, as Opposed to more typical interactions. Attention will be given to these critical issues in Chapter IV. Factor Analyses on Dependent Measures Effects of prior manipulations were observed on five dependent variables: communicative certainty, state anxiety, interpersonal attraction, perceived state anxiety and per- ceived interpersonal attraction. The latter two variables refer to subject perceptions of the stimulus person's affec- tive states. Definitions and operations for these variables are presented in Chapter II and will not be repeated here except as required for clarity. 133 Observations on all dependent measures (excepting two-item scales noted below) were factor analyzed to assess the unidimensional validity of each scale. The principal— component procedure (unities in the diagonals and eigenvalue default of 1.0) with varimax rotation was initially used in each case. Multi-factor solutions were then forced as necessary to discern appropriate factor structures. Four criteria were established a priori to determine optimal solutions: (1) items must load at a minimum of .60 and crossload at a maximum of .40 to be retained on a given factor, (2) a minimal alpha coefficient of .70 is acceptable within each factor, (3) items associated with each factor must clearly exhibit common meaning, and (4) a maximum number of items meeting the prior criteria should be retained to minimize loss of information. Certainty/uncertainty experienced by nonhandicapped 'persons interacting with the handicapped is a key link be- tween independent factors and remaining dependent variables in the theoretical model underlying this study (of. Figure l, p. 38). Moreover, the lO-item Likert-type instrument used to measure certainty was relatively untried (cf. Appendix B, I, 1-10). Consequently, the factor structure of observa- tions taken on this scale was scrutinized in several ways. The initial factor--ana1ytic procedure yielded a two- factor solution with three items failing to load adequately on either factor. In addition, the second factor did not 134 meet the specified reliability criterion. Controlling for one, two and three-factor structures, trivial items were serially deleted from subsequent analyses in an attempt to identify a satisfactory solution. In terms of number of items retained, optimal two and three-factor solutions consistently failed to meet the reliability criterion on one or more factors. The single- factor solution reported in Table 5 conformed more adequate- ly with all criteria specified above. Moreover, in terms of content, items loading on this factor appear to reflect the predictive element in interpersonal communication theoret- ically associated with proactive certainty/uncertainty by Berger and Calabrese (1975). Hence, the factor is labeled proactive certainty. As shown in Table 5, six items will ultimately comprise the certainty measure in subsequent analyses. Cronbach's alpha yielded an acceptable estimate of internal consistency among these items (a = .74). State anxiety was measured by means of a slightly modified version of the 20-item A-State scale (of. Appendix B, I, 11-30) developed by Spielberger, Gorsuch and Lushene (1970). The initial principal-component procedure (eigen- value default set at 1.0) with varimax rotation produced a highly unsatisfactory solution. Five factors were extracted with adequate loading/crossloadings for 12 of 20 original items. Only two items loaded adequately on Factor 3: one item loaded on Factor 5. 135 Table 5. Factor Loadings - Communicative Certainty Scale Factor 1 Item (Proactive Certainty) How certain are you . . . 1. of your general ability to pre- dict the way this person is likely to behave in different situations? .685* 2. that you understood or inter- preted this person correctly when s/he spoke to you? .521 3. that you know or can predict some of the moral and social values this person holds? .624* 4. that you know whether or not this person liked you? .593 5. that you know or can predict the basic attitudes that guide this person's behavior? .661* 6. that this person will generally respond to your conversation in the way you expect? .199 7. that you know or can predict how this person feels about him/ herself? .628* 8. about whether or not this person is highly motivated toward his/her work? .625* 9. that this person is genuinely concerned about whether or not others are at ease in his/her presence? .600* 10. of the way s/he would likely treat you as a person if you were working in the same office? .537 *Indicates item retained on the factor. 136 One to six-factor structures were then forced while systematically deleting trivial items in search of the optimal solution. The three-factor solution reported in Table 6 was ultimately accepted as conforming most clearly with prior criteria. This decision required a relaxation of the specified alpha level (.70) relative to Factor 3. Table 6. Factor Loadings - State Anxiety Scale Factor 1 Factor 2 Factor 3 Item (Positive) (Normal) (Negative) 2. I felt secure. .659* .358 .098 3. I was tense. .205 .633* .198 4. I was regretful. .259 -.125 .621* 5. I felt at ease. .608* .377 .351 6. I felt upset. .071 .098 .728* 11. I felt self-confident. .642* .333 -.101 ‘12. I felt nervous. .286 .792* -.035 13. I was jittery. .042 .805* -.039 14. I felt "high strung." -.043 .628* .397 16. I felt content. .740* -.006 .213 18. I felt over-excited and "rattled." .013 .283 .694* 19. I felt joyful. .718* .004 -.059 20. I felt pleasant. .715* .021 .344 *Indicates item retained on that factor. 137 Items 1, 7, 8, 9, 10, 15 and 17 were deleted from the analysis to reach this solution. Conceptually, all items loading on Factor 1 clearly reflect positive emotional states (e.g., secure, self—confi- dent, joyful). Those retained on Factor 3 clearly represent adverse reactions with potentially negative consequences, particularly in an evaluative interview (e.g., regretful, upset, rattled). Items associated with Factor 2 appear to occupy middle ground somewhere between the other two factors (e.g., tense, nervous, jittery). That is, some degree of tension might be considered "normal" under given circum- stances with the potential for positive or negative conse- quences depending on the level of reaction. Hence, in lieu of more definitive theoretical information, these three factors were interpreted as positive, normal and negative dimensions of state anxiety. Their alpha coefficients were .82, .74 and .64, respectively. Interpersonal attraction was measured by means of two scales developed independently and purported to operational- ize the conceptual variable of interest here. The first measure consisted of two items adopted directly from Byrne's (1971) Interpersonal Judgment Scale (cf. Appendix B, I, 31- 34). Internal consistency of the two items in this context, as estimated via Pearson's r, was somewhat less than optimal (r = .65). This result can be explained in terms of more recent factor-analytic studies confirming at least three 138 dimensions underlying the attraction construct: social, physical and task (e.g., McCroskey & McCain, 1974; Wakshlag, 1973; McCain & Repensky, 1972). Byrne's IJS items appear quite clearly to tap two different dimensions of attraction, social and task. The second measure consisted of a five-item social attraction scale (of. Appendix B, I, 35-39) developed by McCroskey and McCain (1974). The initial factor-analytic procedure identified this scale as unidimensional. As shown in Table 7, item two loaded slightly below the specified .60 level. Hence, items 1, 3, 4 and 5 ultimately comprised the interpersonal attraction measure. The alpha coefficient for these items was .85. Table 7. Factor Loadings - Interpersonal Attraction Scale Factor 1 Item (Social Attraction) l. I think s/he could be a friend of mine. .853* 2. It would be difficult to meet and talk with him/her again. .588 3. We could never establish a personal friendship with each other. .746* 4. I would enjoy having a friendly chat with him/her again. .845* 5. S/he just wouldn't fit into my circle of friends. .816* *Indicates item retained on the factor. 139 The fourth dependent variable, perceived state- anxiety, refers to a subject's estimate of the level of state anxiety experienced by the stimulus source during interaction. It was measured via a modified version of the 20-item state anxiety scale described above (cf. Appendix B, 1, 40-59). Initial principal-component procedures produced an unsatisfactory solution. Four factors were extracted with adequate loading/cross loadings for 12 of 20 original items. Only three items loaded adequately on the final two factors. One to five-factor solutions were then forced while system- atically deleting trivial items. The optimal three-factor solution, retaining 15 of 20 items, produced an unacceptable alpha coefficient for Factor 3 (a = .60). As reported in Table 8, a more acceptable two-factor solution was identified by eliminating items 3, 9, l3, 14, 18 and 19 from the analysis. All eight items retained on Factor 1 reflect positive affective states (e.g., joyful, self-confident, secure). Conversely, the six items loading on Factor 2 are clearly negative affects (e.g., worried, jittery, upset). Hence, the factors were labeled posie tive and negative dimensions of perceived state-anxiety. Alpha coefficients for these factors were .89 and .77, respectively. The final dependent variable, perceived interpersonal attraction, refers to the subject's estimate of how 140 Table 8. Factor Loadings - Perceived State Anxiety Scale Factor 1 Factor 2 Item (Positive) (Negative) 1. S/he felt pleasant. .760* -.004 2. S/he felt joyful. .689* -.012 4. S/he was worried. .320 .601* 5. S/he felt content. .710* .234 6. S/he was relaxed. .674* .339 7. S/he felt "high strung." .008 .726* 8. S/he was jittery. .147 .688* 10. S/he felt self—confident. .722* .259 ll. S/he felt comfortable. .736* .307 12. S/he felt anxious. .134 .735* 15. S/he was worried over possible misfortunes. .152 .628* 16. S/he felt upset. .346 .660* 17. S/he felt at ease. .779* .246 20. S/he felt secure. .737* .240 *Indicates item retained on that factor. interpersonally attractive s/he was to the stimulus source during the interview. Like its conceptual counterpart, interpersonal attraction, this variable was also measured via modified versions of Byrne's IJS and the McCroskey/ McCain (1974) social attraction scale (of. Appendix B, I, 61 and 62, 64-68). 141 Zero-order correlations on the two IJS items indicated an extremely low level of internal consistency in this case (r = .29). On that account, data collected with this instru- ment was not considered in subsequent analyses. Fortunately, the alternate perceived social attraction scale was found to be unidimensional and highly reliable. Table 9 reports results of the initial factor-analytic procedure for this scale. Item 4 alone failed to load adequately and was deleted. Table 9. Factor Loadings - Perceived Interpersonal Attraction Scale Factor 1 Item (Social Attraction) l. I believe this person feels that I just wouldn't fit into his/her circle of friends. .741* 2. I believe this person would enjoy having a friendly chat with me again. .800* 3. I believe s/he feels that we could never establish a personal friend- ship with each other. .869* 4. I believe s/he feels that it would be difficult to meet and talk with me again. .522 5. I believe this person feels that I could be a friend of his/hers. .871* Since factor loadings can change dramatically with the elimination of a single item, a second analysis was conducted using the identical procedure (minus item 4). 142 All four items retained loadings on the same factor equal to or higher than those reported in Table 8. The alpha coefficient for these items was .85. Tests of Theorems This section reports results of statistical tests con- ducted on fifteen theorems developed in Chapter I. The theorems are examined in order of their theoretic import according to the model presented in Figure 1. Each ANOVA reported here tested (or partially tested) three related theorems relative to a single dependent variable. Prior to results of each F-test, relevant theorems are presented in verbal and symbolic form for clarity. Individual responses were summed across each unidimensional scale (or scale factor) before conducting the two-way ANOVA. Tukey's test was used for all post-hoc comparisons. Statistical signif- icance was determined at the conventional a = .05 level. Effects on Communicative Certainty Communicative certainty, the dependent variable in theorems l, 6 and 11, is centrally related to all other variables in the uncertainty reduction model under study. Theorem 1 predicts a main effect from Factor A on communi- cative certainty in a high-medium—low direction. T1: Interactants eXposed to increasing levels of V/NV affiliation will report increasing levels of communicative certainty. 143 Theorem 6 predicts a main effect from Factor B such that interactions with the handicapped (H/NH) will generate lower levels of certainty than nonhandicapped (NH/NH) interactions. T6: Nonhandicapped persons will report more commun- icative certainty when exposed to the same level of V/NV affiliation generated by a nonhandicapped, as opposed to a handicapped, source. 171 >332 Theorem 11 predicts an A x B ordinal interaction assum- ing the same rank order of effects predicted in T6. That is, H/NH treatments will consistently produce lower levels of certainty than the NH/NH type. In addition, Tll pre- dicts that differences between these two types will increase toward the highest level of Factor A (i.e., the NH/NH curve will increase more rapidly). T11: As nonhandicapped persons are exposed to increasing levels of V/NV affiliation generated by handicapped versus nonhandicapped sources, their (NH) reported levels of communicative certainty will increase more rapidly when the source is nonhandicapped. (Eel - fez) > (i161 - i162) > (icl - 'icz) Results of the two-way ANOVA testing theorems l, 6 and 11 are summarized in Table 10. As modified by prior factor analyses, the certainty scale was comprised of six items with a scoring range from 6 to 36. A significant main effect was found for Factor A. Inspection of column marg- inals in Table Ulindicates that increasing levels of 144 Table 10. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affiliation and Interaction Type on Communicative Certainty. Level of V/NV Affiliation (A) Type of _ Interaction(B) a b c X NH/NH (1) 27.17d 23.42ef 22.58f 24.39 H/NH (2) 26.38de 24.7ldf 24.79def 25.29 x 26.77 24.06 23.69 i = 24.84 Source Sum of Squares df Mean Square F Affiliation(A) 271.72 2 135.86 6.65* .09 Type of Interaction(B) 29.34 1 29.34 1.44 - A x B 56.72 2 28.36 1.39 — Error 2821.54 138 20.45 Total 3179.33 143 22.23 Note: Cell means subscripted with a common letter do not differ significantly. *p < .05 communicative certainty accompanied increases in V/NV affiliation. Theorem 1 was clearly supported. Results of the Tukey test show that the major contri— bution of variance came from NH/NH interactions. Treat- ments b1 and c1 were both significantly different than a1. Although generally increasing in the predicted direction, H/NH interactions did not differ significantly from the NH/NH type with one exception, treatment a versus c . 2 l Magnitude of the effect on certainty attributable to 145 Factor A was moderate for social science data (n2 = .09). The data do not support theorems 6 and 11. Inspection of row marginals in Tablelk)suggests a trend quite opposite that predicted in theorem 6. Although differences were nonsignificant, participants generally reported higher levels of certainty when interacting with the handicapped, as opposed to the nonhandicapped, source. The single exception, conforming with the predicted pattern, occurred in the high condition of Factor A (i.e., a > a Reverse l 2)’ order of these effects at the two lower levels of Factor A produced a discernible, though nonsignificant, disordinal interaction. Certain trends in these data are worth noting. First, the general direction of certainty effects in both conditions of Factor B approximate the pattern predicted in T1. Second, the rank order of effects in four of six conditions (132 > bl; c2> Cl) is contrary to the pattern predicted in T6. Third, there is a trend toward disordinal interaction contrary to the ordinal type predicted in T11. The pattern of these results is graphically illustrated in Figure 6. Effects on State Anxietxti;_l-.-~v Theorem 2 predicts a main effect from Factor A on state anxiety. T2: Interactants exposed to increasing levels of V/NV affiliation will report decreasing levels of state anxiety. 146 28___ 27.__ CO 00 c 'H U m “3 26 __ >~. J.) C‘. 2H m U 8 C) 25 ___ H/NH Dyads 0) > '1-1 4.) m U .3 24 g .__ E o L) C‘. 8 z 23_ NH/NH Dyads 22fl__ l l A High Medium Low Level of V/NV Affiliation Figure 6. Graphic Representation of Communicative Certainty as a Function of Affiliation and Type of Dyadic Interaction. 147 Theorem 7 predicts a main effect from Factor B such that interactions with the handicapped (H/NH) will produce higher levels of state anxiety than nonhandicapped (NH/NH) interactions. T7: Nonhandicapped persons will report less state anxiety when exposed to the same level of V/NV affiliation generated by a nonhandicapped, as opposed to a handicapped, source. X1 (ic2 - 3Ec1) Prior factor analyses on the 20-item scale used to measure state anxiety produced a three-factor solution. Accordingly, each of three F-tests to follow represents a partial test of theorems 2, 7 and 12. Six items comprised the first factor (cf. Table 6) identified as a positive dimension of state anxiety with a scoring range from 6 to 24. 148 No significant differences were found via the two-way ANOVA summarized in Table 11. Table 11. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affiliation and Interaction Type on State Anxiety - Factor 1. Level of V/NV Affiliation (A) Type of _ Interaction (B) a b c X NH/NH (1) 12.63 12.92 15.08 13.54 H/NH (2) 13.38 15.04 14.00 14.14 3? 13.00 13.98 14.54 SE = 13.84 Source Sum of Squares df Mean Square F n2 Affiliation (A) 58.43 2 29.21 2.20 --- Type of Interaction (B) 12.84 1 12.84 .97 -—- A x B 62.18 2 31.09 2.35 --- Error 1829.87 138 13.26 Total 1963.32 143 13.73 Although statistically nonsignificant, certain trends in these data are noteworthy. First, inspection of column marginals in Table 10 suggests a pattern of decreasing state anxiety in the direction T2 predicts. Second, row marginals conform with the direction predicted in T7. Indeed, with the exception of oz, all cell means reflect the pattern of main effects predicted in both T2 and T7. 149 Third, foregoing comments must be qualified by the disturbance in these data occurring at the low level of Factor A. Reverse of the predicted order of effects at c1 and c2 (i.e., c > c2) introduces a trend toward disordinal 1 interaction contrary to T12. The A x B interaction, in this case, reached a .10 level of significance. Additional- ly, differences between the two types of dyad do not increase toward the high level of Factor A as predicted in T12. Four items comprised the second factor of the state anxiety scale (cf. Table 6) with a scoring range from 4 to 16. Results of the two-way ANOVA partially testing theo- rems 2, 7 and 12 via this factor are reported in Table 12. A significant main effect emerged for Factor A. While T2 is partially supported by these data, the column marginals in Table 12 do not clearly conform with the predicted pattern. Participants in general reported slightly more anxiety at the medium versus the low level of Factor A. With that exception, decreasing levels of state anxiety (normal type) accompanied increasing levels of affiliation. Post-hoc comparisons among cell means via the Tukey test were nonsignificant. Magnitude of effects attributed to Factor A on anxiety of this type was small (n2 = .04). Although the main effect predicted for Factor B was nonsignificant, row marginals in Table 12 indicate that NH participants generally reported more anxiety (of this type) 150 Table 12. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affiliation and Interaction Type on State Anxiety - Factor 2. Level of V/NV Affiliation (A) Type of _ Interaction(B) a b c X NH/NH (l) 5.50 6.79 6.46 6.25 H/NH (2) 6.13 6.71 6.71 6.51 i 5.81 6.75 6.58 3?: 6.38 Source Sum of Squares df Mean Square F n2 Affiliation(A) 24.01 2 12.01 2.97* .04 Type of Interaction(B) 2.51 l 2.51 .62 —-- A x B 3.01 2 1.51 .37 --- Error 558.46 138 4.05 Total 587.99 143 4.11 *p = .05 when the message source was handicapped. All cell means, excepting b1, conform with the pattern predicted in T7. Similarly, while T12 was not supported, the data show certain notable consistencies with this theorem. First, inspite of relatively low variance in these data, consider- able decreases in state anxiety occurred at the high level of Factor A for both types of interaction. Second, abso- lute differences between the two dyadic types are most dramatic at the high level of Factor A, i.e., (Ya 2 - Xal) > (sz - Xbl) and (Xa2 - Xal) > (Xc2 - Xcl). Exception to 151 the predicted pattern occurred as a result of small abso- lute differences at the medium level of Factor A, i.e., (53C2 - £61) > (ibz - §b1)- These trends may become more or less meaningful in light of subsequent analyses. Three items comprised the third factor emerging from prior factor analyses on the state anxiety scale. These items appeared consistently to represent a negative type of anxiety (e.g., I felt upset). Factor three had a scoring range from 3 to 12. Results of the ANOVA partially testing theorems 2, 7 and 12 via this dimension of state anxiety apprear in Table 13. The main effect for Factor A was highly significant (p = .019) in the direction predicted. Decreasing levels of state anxiety (negative type) accompanied increasing levels of V/NV affiliation as column marginals in Table 13 indicate. Cell means also follow the predicted pattern in both NH/NH and H/NH interactions. Again, T2 was partially supported. Magnitude of effects from Factor A was reason- able for social science data (Cohen, 1969), n2 = .06. Theorems 7 and 12 were not supported. Across all treat- ments, NH/NH dyads reported slightly more anxiety (negative type) than H/NH dyads as indicated by row marginals in Table 13. However, all cell means follow the order pre- dicted in T2 and T7 excepting c Anxiety of this type 2. also declined more rapidly for NH/NH dyads in accord with T12. 152 Table 13. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affiliation and Interaction Type on State Anxiety - Factor 3. Level of V/NV Affiliation (A) Type of _ Interaction(B) a b c X NH/NH (l) 3.13d 3.38 4.00d 3.50 H/NH (2) 3.33 3.42 3.58 3.44 3? 3.23 3.40 3.79 i = 3.47 Source Sum of Squares df Mean Square F n2 Affiliation(A) 8.01 2 4.01 4.09* .06 Type of Interaction(B) .11 1 .11 .ll --- A x B 2.51 2 1.26 1.28 Error 135.25 138 .98 Total 145.89 143 1.02 Note: Cell means subscripted with a common letter are significantly different. *p < .05 Effects on Interpersonal Attraction Theorem 5 posits a main effect from Factor A on inter— personal attraction. TS: Interactants exposed to increasing levels of V/NV affiliation will report increasing levels of interpersonal attraction toward the message source. Xa > Xb > Xc Theorem 10 predicts a main effect from Factor B such that NH/NH interactions will generate higher levels of 153 interpersonal attraction than the H/NH type. T10: Nonhandicapped persons will report more inter- personal attraction (toward the counterpart) when exposed to the same level of V/NV affiliation gener- ated by a nonhandicapped, as opposed to a handicapped, source. Xl>X2 Theorem 15 posits an A x B interaction assuming the same rank order of simple effects hypothesized in T10. Addition- ally, T15 predicts that differences between NH/NH and H/NH dyads will increase in the direction of Factor A, condition a, i.e., the NH/NH curve will increase more rapidly than the H/NH curve. T15: As nonhandicapped persons are exposed to increasing levels of V/NV affiliation generated by handicapped versus nonhandicapped sources, their (NH) reported levels of interpersonal attraction (toward the counterpart) will increase more rapidly when the source is a nonhandicapped person. (Xa1 - Xa2) > (Xbl - sz) > (Xc — Xcz) 1 For purposes of testing theorems 5, 10 and 15, data was collected on two different instruments, the McCroskey and McCain (1974) social attraction scale and Byrne's (1971) interpersonal judgment scale. Internal reliability of the latter 2-item instrument was less than optimal (a = .65). In the interest of additional insight, results obtained from Byrne's IJS will subsequently be compared with findings via the more reliable McCroskey/McCain social attraction scale (a = .85). 154 As modified by prior factor analyses, four items com- prised the McCroskey/McCain instrument with a scoring range from 4 to 28. Results of the two-way ANOVA assessing effects on interpersonal attraction via this scale are reported in Table 14. Table 14. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affiliation and Interaction Type on Interpersonal Attraction - Social Attraction Scale. Level of V/NV Affiliation (A) Type of _ Interaction(B) a b c X NH/NH (l) 20.04d 18.17 14.92def 17.71 H/NH (2) 19.71e 18.92 19.67f 19.43 SE 19.87 18.54 17.29 3? = 18.57 Source Sum of Squares df Mean Square F n2 Affiliation(A) 160.22 2 80.11 3.19* .04 Type of Interaction(B) 106.78 1 106.78 4.25* .03 A x B 172.06 2 86.03 3.42* .04 Error 3470.25 138 25.15 Total 3909.31 143 27.34 Note: Cell means subscripted with a common letter are significantly different. *p < .05 Both main effects and the A x B interaction were signif- icant. Column marginals in Table 14 indicate that over all participants, increasing levels of attraction accompanied 155 increases in V/NV affiliation as T5 predicts. However, this general finding does not take complete account of the data in presence of a significant disordinal interaction. In this case, the main effect cannot be interpreted inde- pendently of the interaction effect (Keppel, 1973). Inspec- tion of simple effects in Table 14 shows that the major contribution of variance came from NH/NH dyads. Condition a and c . The pattern 1 1' 2 2 of increasing attraction with increasing affiliation pre- c differed significantly from a dicted in T5 held consistently only at level 1 of Factor B (i.e., in NH/NH dyads). Moreover, magnitude of the main effect from Factor A was small (n2 = .04). Theorems 10 and 15 were not supported. Although the main effect for Factor B was significant, reverse of the order predicted in T10 occurred in two of three conditions (Factor A, b and g) as a result of the A x B interaction. Participants exposed to the handicapped source reported less interpersonal attraction (versus the nonhandicapped source) only at the highest level of Factor A. In addition, while attraction increased more rapidly in NH/NH versus H/NH dyads toward the high level of Factor A, rank order of the simple effects was not in accord with T15. These data are plotted graphically in Figure 7. Results of analyses on the alternate set of attraction data collected via Byrne's IJS were remarkably consistent with those obtained from the McCroskey/McCain instrument. 156 20,__ H/NH Dyads 19 ___ U) 60 .S u 18___ m a: c o «4 4.) 8 H 17 ___ 4.) 4.! .< c m m 2': 16___ NH/NH Dyads 15.__ I l I High Medium Low Level of V/NV Affiliation Figure 7. Graphic Representation of Interpersonal Attraction as a Function of Affiliation and Type of Interaction. 157 Two items comprised the Byrne's scale with a scoring range from 2 to 14. The two-way ANOVA used to analyze these data is summarized in Table 15. Table 15. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affiliation and Interaction Type on Interpersonal Attraction - IJS Scale. Level of V/NV Affiliation (A) Type of _ Interaction(B) a b c X NH/NH (l) 9.88d 8.71 7.29def 8.63 H/NH (2) 9.79e 9.17 10.08f 9.68 i 9.83 8.94 8.69 34': 9.15 Source Sum of Squares df Mean Square F n2 Affiliation(A) 34.85 2 17.42 2.45 -—— Type of Interaction(B) 40.11 1 40.11 5.64* .04 A x B 56.01 2 28.00 3.94* .05 Error 981.67 138 7.11 Total 1112.64 143 7.78 Note: Cell means subscripted with a common letter differ significantly. *p < .05 The main effect for Factor A approached, but did not reach, statistical significance (p < .10). With that excep- tion, the pattern of effects reported in Table 15 is highly consistent with prior findings in Table 14. Again, both main effects are qualified by the presence of a significant, disordinal A x B interaction. The pattern of means 158 predicted in T5 held consistently only at level 1 of Factor B. The main effect for Factor B was significant in the opposite direction T10 predicts. Attraction toward the non- handicapped source exceeded that for the handicapped type only at the high level of Factor A. In general, the handi- capped source was more interpersonally attractive. Magni- tude of this main effect was relatively small (n2 = .04). The curves representing NH/NH versus H/NH dyads in these data are very similar to those plotted in Figure 7. Although attraction increased more rapidly in NH/NH dyads, again the rank order of simple effects was contrary to prediction in T15. Confirming prior results, cell means c differed sig- 1 1' a2 and c2. Magnitude of the disordinal interaction was small (n2 = .05). nificantly in post-hoc comparisons with a Effects on Perceived Interpersonal Attraction Theorems 4, 9 and 14 predict effects on the fourth dependent variable in order of theoretic import, perceived interpersonal attraction. Theorem 4 predicts a main effect from Factor A in the high-medium-low direction. T4: Interactants exposed to increasing levels of V/NV affiliation will report increasing levels of perceived interpersonal attraction from the message source. Xa > Xb > Xe Theorem 9 predicts a main effect from Factor B such that interactions with the handicapped (H/NH will generate lower levels of perceived interpersonal attraction than the 159 nonhandicapped type (NH/NH). T9: Nonhandicapped persons will report more per- ceived interpersonal attraction (from the counter- part) when exposed to the same level of V/NV affiliation generated by a nonhandicapped, as opposed to a handicapped, source. X1>X2 Theorem 14 predicts an A x B ordinal interaction assuming the same rank order of effects predicted in T9. That is, H/NH treatments will consistently produce lower levels of perceived interpersonal attraction. In addition, T14 predicts that differences between the two dyadic types will increase toward the highest level of Factor A, i.e., the NH/NH curve will increase more rapidly. T14: As nonhandicapped persons are exposed to increasing levels of V/NV affiliation generated by handicapped versus nonhandicapped sources, their (NH) reported levels of perceived interpersonal attraction (from the counterpart) will increase more rapidly when the source is nonhandicapped. (Xa - Xa2) > (Xb1 - sz) > (Xc - Xc2) l 1 Results of the two-way ANOVA testing theorems 4, 9 and 14 are reported in Table 16. The perceived interpersonal attraction scale was unidimensional with a scoring range from 5 to 35. A significant main effect emerged for Factor A. Column marginals in Table 16 indicate that over all participants, increasing levels of perceived interpersonal attraction accompanied increasing affiliation. Theorem 4 was supported; however, its interpretation must be qualified in light of the significant disordinal interaction occurring 160 Table 16. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affiliation and Interaction Type on Perceived Interpersonal Attraction. Level of V/NV Affiliation (A) Type of _ Interaction(B) a b c X NH/NH (1) 27.50a 26.17 22.75a 25.47 H/NH (2) 26.71 24.42 26.42 25.85 E 27.10 25.29 24.58 i = 25.66 Source Sum of Squares df Mean Square F 02 Affiliation(A) 162.26 2 81.13 3.30* .04 Type of Interaction(B) 5.06 l 5.06 .21 --- A x B 200.54 2 100.27 4.08* .05 Error 3390.46 138 24.57 Total 3758.33 143 26.28 Note: Cell means subscripted with a common letter differ significantly. *p < .05 at the low level of Factor A. The pattern predicted in T4 held consistently across three levels of Factor A for NH/NH interactions. This was the case in only two of three con- ditions (a2 and b2) representing H/NH interactions. Although the difference was nonsignificant, participants perceived more interpersonal attraction from the handicapped source at the low level (c2), as compared to the medium level (b2), of Factor A. The major contribution of variance 161 came from NH/NH interactions. However, only treatments a1 and c were significantly different among cell means. 1 Magnitude of effects attributable to Factor A on perceived . . . 2 interpersonal attraction was relatively small (n = .04). These data do not support theorems 9 and 14. Row marginals in Table 16 suggest a trend contrary to the direc- tion of main effects predicted for Factor B in T9. This conclusion is somewhat misleading due to the strong A x B interaction occurring at condition 3 of Factor A. Less interpersonal attraction was perceived from the handicapped source in two of three Factor A treatments (a2 < a1; b2 b1). Magnitude of the A x B disordinal interaction effect was small (n2 = .05). These data are represented graph- ically in Figure 8. Effects on Perceived State Anxiety Theorems 3, 8 and 13 predict effects on the final dependent variable, perceived state anxiety. Theorem 3 predicts a main effect from Factor A in the familiar high- medium—low direction. T3: Interactants exposed to increasing levels of V/NV affiliation will report decreasing levels of perceived state anxiety. Xa < Xb < Xc Theorem 8 posits a main effect for Factor B such that H/NH interactions will generate higher levels of perceived state anxiety than the NH/NH type. 162 28 6% c 27 'H U m M H/NH Dyads a o I} U 26 m H 4.) 4.) ‘< 73 g 25 CO p 0) D. u 0) 4.) .3 24 vs CD > H m U 8 m 23 8 g NH/NH Dyads 22 L l I High Medium Low Level of V/NV Affiliation Figure 8. Graphic Representation of Perceived Interpersonal Attraction as a Function of Affiliation and Type of Interaction. 163 T8: Nonhandicapped persons will report less perceived state anxiety when exposed to the same level of V/NV affiliation generated by a nonhandi- capped, as Opposed to a handicapped, source. Xl (2152 - ibl) > (SECZ - Eel) Factor analyses on the instrument used to measure per- ceived state anxiety yielded a two-factor solution. Hence, results of each ANOVA to follow represent a partial test of theorems 3, 8 and 13. Eight items comprised the first factor (cf. Table 8) identified as a positive dimension of perceived state anxiety. This factor had a scoring range from 8 to 32. Results of the two-way ANOVA used to analyze data collected via the first scale factor are reported in Table 16. Neither T3 nor T8 was supported. However, the main effect for Factor A approached significance (p < .10) in 164 Table 17. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affiliation and Interaction Type on Perceived State Anxiety - Factor 1. Level of V/NV Affiliation (A) Type of _ Interaction(B) a b c x NH/NH (1) 17.54 16.52a 20.50 18.21 H/NH (2) 17.17 20.83a 18.71 18.90 X 17.35 18.72 19.60 i = 18.56 Source Sum of Squares df Mean Square F n2 Affiliation(A) 123.27 2 61.64 2.51** --- Type of Interaction(B) 16.98 1 16.98 .69 --- A x B 241.56 2 120.78 4.92* .06 Error 3365.32 137 24.56 Total 3747.25 142 26.39 Note: Cell means subscripted with a common letter differ significantly. *p < .05 **p < .10 the direction T3 predicts as column marginals in Table 17 indicate. Participants generally reported less perceived state anxiety as V/NV affiliation increased. More specif- ically, in the presence of a significant A x B interaction, simple effects in Table 17 follow the pattern T3 predicts only at the high and low extremes of Factor A. Row marginals in Table 17 show a trend in the direction T8 predicts. That is, the handicapped source was generally 165 perceived as somewhat more anxious than the nonhandicapped type. However, in this case the simple effects are clearly consistent with T8 only at the medium level of Factor A as a result of the A x B interaction. These data do not support T13. Excepting the order of simple effects in condition g of Factor A, both the rank order of effects and the disordinal interaction are con- trary to T13. Factor A, condition b, was the major source of variance; only b1 and b2 were significantly different as Figure 9 illustrates. Magnitude of the interaction effect was moderate (n2 = .06). In accord with prior factor-analytic results, six items comprised the second factor of the perceived anxiety scale (cf. Table 8). This factor, identified as the negative dimension of perceived state anxiety, had a scoring range from 6 to 24. Results of the two-way ANOVA used to analyze data obtained via this scale factor are reported in Table 18. Neither the main nor interaction effects were signifi- cant. Hence, theorems 3, 8 and 13 were not supported. Both column and row marginals are inconsistent with T3 and T8. In addition, a trend toward disordinal A x B interaction is evident due to the reverse order of simple effects in con- ldition b_of Factor A (i.e., b < bl)' In lieu of signifi- 2 cance, it is worth noting that both the rank order and directional pattern of simple effects in these data are quite similar to those obtained via Factor 1 of the 166 21 ___ NH/NH Dyads m 20 __ 60 c -H .L.) (U as >. 4.) 34’ 19__ x 5 m g H/NH Dyads a T, 18 ___ 0) > ‘H o U u m a. c 17 __. m m z 16 ___ 1 1 1 High [Medium Low Level of V/NV Affiliation Figure 9. Graphic Representation of Perceived State Anxiety as a Function of Affiliation and Type of Interaction - Factor 1. 167 Table 18. Means and Analysis of Variance Summary for Effects of Verbal/Nonverbal Affiliation and Interaction Type on Perceived State Anxiety Factor 2. Level of V/NV Affiliation Type of _ Interaction(B) a b c X NH/NH (l) 9.71 9.21 9.71 9.54 H/NH (2) 9.09 10.30 8.38 9.24 SE 9.40 9.74 9.04 if = 9.39 Source Sum of Squares df Mean Square F 02 Affiliation(A) 11.64 2 5.82 .66 --- Type of Interaction(B) 3.06 l 3.06 .35 ——- A x B 36.91 2 18.46 2.10 --- Error 1196.20 136 8.80 Total 1247.92 141 8.85 perceived state anxiety scale. These results are discussed in Chapter IV. CHAPTER IV DISCUSSION This chapter attempts a plausible interpretation of significant results emerging from the study. Since the research was largely exploratory, theoretic explanations are offered for other potentially informative trends in these data. Ensuing discussion utilizes as much informa- tion from the previous chapter as reason and relevancy dictate. First, significant results in accord with hypoth- esized relations are reviewed. Conclusions are drawn rela- tive to each dependent variable. Second, significant results contra hypothesized relations are examined. Third, nonsignificant trends of potential, heuristic value are considered. Fourth, attention is directed to salient limi- tations of the study. Finally, emergent directions for future research are reviewed. All results are discussed in the order of priority assigned each dependent variable in Chapter III. Eleven of twenty seven tests previously reported pro- duced statistically significant results. Four of fifteen theorems representing prior theory received full or partial support. Findings on two additional theorems approached, but failed to reach, the conventional (.05) level of sig— nificance. At face value, these results were disappointing. However, that assessment may be qualifiable on one account. Use of the videotape medium necessarily provided a 168 169 conservative test of the theory under study. That is, non— handicapped participants did not engage handicappers in live, face-to-face interaction. Consequently, some dilution of certain theoretically important nonverbal variables was inevitable and the magnitude of Observed effects in general was expected to be small. Using criteria provided by Cohen (1969), the a priori estimate of statistical power for the overall F-test on main effects was, in this case, above the .80 convention. Actual post hoc power achievements are discussed below relative to certain nonsignificant trends in the data that suggest the possibility of more definitive effects via live interaction. Hypothesized Relations As theorem 1 predicts, clear evidence emerged in support of a positive association between V/NV affiliation and (proactive) communicative certainty. Regardless of dyadic type, participants reported greater confidence in their ability to predict the interviewee's behavior when s/he exhibited higher (more congruent) levels of the rele- vant verbal/nonverbal cues. This effect was somewhat less pronounced in H/NH dyads where none of the simple effects was significantly different. Moreover, while certainty ratings tended to decrease in the direction predicted, no appreciable difference occurred in H/NH dyads as a function of reducing the nonverbal versus the verbal message compo- nent. In NH/NH interactions both the verbal and the 170 nonverbal reductions produced significantly different effects than the congruent manipulation (high level) of V/NV affiliation. In addition, the nonverbal reduction produced a discernibly larger (absolute) decrease in cer- tainty as expected. Dyadic differences of this type bear potential theoretic import and are discussed with other nonsignificant trends later. At face value, effects on (proactive) communicative certainty Observed in this study warrant the following conclusions with reference to dyadic interactions: 1. There is a positive, linear association between V/NV affiliation and communicative certainty regardless of interaction type. 2. Level of V/NV affiliation produced by a handi- capped source does not atypically affect the level of communicative certainty experienced by a NH counterpart. 3. As independent factors, V/NV affiliation and type of dyad, do not interact to produce differential, conjoint effects on communica— tive certainty in H/NH interactions. With reference to state anxiety, theorem 2 received partial support. Effects observed via the third factor of the tridimensional scale used to measure this variable clearly conform with the predicted pattern. Decreasing levels of state anxiety (negative type) consistently accom- panied increasing levels of V/NV affiliation in both types of interaction. No significant differences relative to negative state anxiety were observed between the two dyadic types. 171 While observations via the second scale factor indi- cated a significant main effect from V/NV affiliation on state anxiety (normal type), column marginals representing this effect did not conform with the linear pattern pre- dicted in theorem 2. Although NH participants reported higher absolute levels of state anxiety (positive type) when the message sOurce was handicapped, neither independ- ent variable produced a significant main effect on anxiety of this type. Effects on state anxiety observed in this study warrant the folloWing conclusions relative to dyadic interactions: 1. There is an inverse, linear association between V/NV affiliation and (negative) state anxiety regardless of interaction type. 2. Level of V/NV affiliation produced by a handi- capped source does not atypically affect the level of (positive, normal, negative) state anxiety experienced by a NH counterpart. 3. As independent factors, V/NV affiliation and type of dyad, do not interact to produce dif- ferential, conjoint effects on state anxiety in H/NH interactions. A relatively small main effect from Factor A on inter- personal attraction (via the McCroskey/McCain instrument) was found in support of theorem 5 (n2 = .04). Column marginals in Table 14 indicate that over all participants, increasing levels of attraction accompanied increases in V/NV affiliation according to the linear pattern predicted. However, this general finding does not take complete account of the data. In this case, the main effect cannot 172 be adequately interpreted without reference to the present A x B disordinal interaction (Keppel, 1973). Inspection of simple effects in Table 14 shows that the major contribution of variance attributable to Factor A came from NH/NH dyads. Treatment condition c1 differed 1, a2 and c2. NO significant differences were observed among simple effects representing H/NH dyads. significantly from a Moreover, the interaction effect in these data resulted primarily from an atypically large difference between the two dyadic types at the low extreme Of Factor A. Hence, the predicted pattern of increasing interpersonal attrac- tion held consistently only in NH/NH interactions. The interaction effect in these data, contra prediction, will be interpreted later. While relevant manipulation data may shed light on present findings (as discussed below), observed effects of affiliation on interpersonal attraction do not consistently follow theorem 5 in linear fashion with respect to H/NH dyads. Results of analyses on the alternate set of attrac- tion data collected via Byrne's IJS were remarkably similar to these data with one exception. The main effect for Factor A approached, but did not reach, statistical sig- nificance (p < .10). In terms of absolute values, once again the pattern of means predicted in theorem 5 held consistently only at level 1 of Factor B. Both data sets produced significant 173 main effects for Factor B and disordinal A x B interactions, contrary to prediction in theorems 10 and 15. These results are discussed in the subsequent section. In light of prior discussion, the following conclusions are drawn with reference to interpersonal attraction in dyadic contexts: 1. There is a positive, linear association between V/NV affiliation and interpersonal attraction in typical interactions engaging NH persons. 2. The foregoing linear association does not appear to hold in dyadic interactions engaging one handicapped person. 3. Level of V/NV affiliation produced by a handicapped source does not effect an atypically low level of interpersonal attraction from a NH counterpart. 4. As independent factors, V/NV affiliation and type of dyad, do not interact to pro- duce proportionately lower (linear) changes in interpersonal attraction relative to H/NH interactions. A main effect from Factor A on perceived interpersonal attraction was observed in support of theorem 4. Overall, participants reported increasing levels of perceived inter— personal attraction (from the counterpart) in the presence of increasing affiliation according to the linear pattern predicted. However, inspection of the simple means in Table 16 again shows that the major source of variance was NH/NH interaction. Only conditions a and c differed l l significantly via post hoc comparisons. 174 Moreover, the observed main effect must again be qualified at the low extreme of Factor A where an atypically large absolute difference (in reverse order) occurred be- tween the two dyadic types. This difference resulted in a significant, disordinal A x B interaction, contrary to pre- diction in theorem 14. While no significant differences were observed among simple effects representing H/NH dyads, their absolute values follow a curvilinear pattern. Hence, the predicted linear pattern and direction of effects held consistently only in NH/NH interactions. Optimally, these data suggest that the observed main effect can be expected to hold in H/NH dyads only at high/ medium levels of V/NV affiliation. Again, relevant manipu- lation data may shed light on these findings as discussed later. Theorem 9 predicting a main effect from Factor B was not supported. In sum, the following conclusions are drawn with reference to hypothesized relations involving perceived interpersonal attraction in dyad contexts: 1. There is a positive, linear association between V/NV affiliation and perceived interpersonal attraction in typical interactions engaging NH persons. 2. The foregoing linear association does not appear to hold in dyadic interactions en- gaging one handicapped person. 3. Level of V/NV affiliation produced by a handicapped source does not effect an atypically low level of perceived inter- personal attraction (as perceived) by a NH counterpart. 175 4. As independent factors, V/NV affiliation and type of dyad do not interact to pro- duce proportionately lower (linear) changes in perceived interpersonal attrac- tion relative to H/NH, versus NH/NH, interactions. Of the five dependent variables observed in this study, evidence relative to effects on perceived state anxiety was least conclusive and interpretable on present theoretic grounds. Across all treatments, absolute decreases in perceived state anxiety (positive type) generally accompanied increases in V/NV affiliation as column marginals in Table 18 indicate. This pattern representing the predicted main effect from Factor A ap- proached, but did not reach, statistical significance (p < .10). In addition, simple effects for both dyadic types were inconsistent, tending to reflect a curvilinear pattern. Complicating interpretation, the NH source was per- ceived as least anxious, the H source most anxious, in the medium condition (b) of Factor A, under a reduction of the verbal message component (cf. Figure 9). Magnitude of this difference, reversing the order of observations in condi— tions a1 versus a2 and c1 versus c2, was sufficient to pro- duce a significant, disordinal A x B interaction. Con- ditions b1 and b2 also produced the only significant differ— ence among simple means in these data via post hoc analyses. Hence, relevant theorems 3, 8 and 13 were not confirmed. \l 176 Evidently, to the extent that affiliation and type of interaction may affect perceptions of state anxiety experi- enced by a counterpart, the relationships among these vari— ables are more complex than the linear patterns present theory predicts. At face value, these data warrant the following conclusions with reference to hypothesized rela- tions involving perceived state anxiety in dyadic contexts: l. A positive, linear association does not obtain between V/NV affiliation and per- ceived state anxiety in either NH/NH or H/NH interactions. 2. Level of V/NV affiliation produced by a handicapped source does not atypically affect the level of state anxiety per- ceived by a NH counterpart--in consistent linear patterns. 3. As independent factors, V/NV affiliation and type Of dyad do not interact to pro- duce prOportionately lower (linear) changes in perceived state anxiety relative to H/NH, versus NH/NH, interactions. Significant Results Contra Prediction Analyses reported in Chapter III revealed several statistically significant findings that Oppose hypothesized relations deduced from prior theory. Surprisingly, both sets of data on interpersonal attraction produced a signif- icant main effect for Factor B in the opposite direction theorem 10 predicts. Results obtained via two independent scales were remarkably consistent relative to this effect. Although magnitude of the effect was small (n2 = .03 and .04), the H source was generally rated as more 177 interpersonally attractive than the NH source. However, inspection Of Tables 14 and 15 indicates that a signifi- cant difference among simple effects representing the two dyadic types occurred only at the low extreme of Factor A. Magnitude of this difference was sufficiently large to produce the concomitant disordinal A x B interaction. In effect, the major source of variance accounting for both the significant main effect (Factor B) and the interaction was treatment A, c representing incongruent V/NV affilia- tive cues via a reduction in the nonverbal message component. Ceteris paribus, the foregoing dyadic differences sug- gest two potentially important, theoretic possibilities to account for this finding. First, incongruent V/NV cues may have less negative impact on interpersonal attraction when exhibited by a handicapped source. Second, NV cues of this type may play a less dominant role in H/NH, as opposed to NH/NH, interactions. Theoretically, these differences may be explained in terms of differing sets of a priori expecta- tions. That is, nonhandicapped persons may typically be conditioned, via past experience, to expect fewer nonverbal cues and, consequently, to tolerate lower levels of V/NV congruency from handicapped counterparts. In turn, these expectations may encourage the nonhandicapped to rely more consistently on verbal cues in H/NH versus NH/NH interactions. 178 This line of reasoning would account for dramatically different effects on attraction observed here (re: H/NH dyads) when affiliative cues were reduced in the nonverbal channel. One less interesting explanation of these results deserves notice. Manipulation check data (Table 2) focus- ing on perceptions of the source's sociability are quite comparable to the attraction data in Tables 14 and 15. Plotted curves representing each dyadic type in these data sets are remarkably similar. Most conspicuous are significant differences among simple effects with the same rank order at the low extreme of Factor A (condition c) in all three cases. Hence, it is conceivable that the dis- turbance in these attraction data is a function of an ineffective manipulation at Factor A, condition c. Judg- ment on this issue is reserved in consideration of additional evidence. Thus far, the discussion has not attempted a theo- retic explanation to account for the significant, dis- ordinal A x B interaction effect emerging from analyses on the perceived interpersonal attraction data. Although their post hoc comparison was nonsignificant, the absolute and c 1 2 (reversing the rank order of effects observed at A, a and difference between simple means in conditions c b) accounted for the relevant interaction effect (cf. Figure 8). 179 In terms of absolute values, participants perceived more interpersonal attraction from the H, versus the NH, source primarily as a result of the Factor A manipulation at condition c. That is, under a reduction in nonverbal, as Opposed to a reduction in verbal, affiliative cues. Again, strength of this difference resulted in a dis- ordinal interaction quite comparable to that previously encountered (at the same location) in the manipulation data assessing sociability. In fact, plotted curves representing both dyadic types in these attraction data are virtually identical to corresponding curves obtained via manipulation check one. The rank orders of simple effects in these two data sets are also identical. On this account as previously suggested, the manipulation of Factor A at condition c is suspect. In addition, it is also conceivable that the absence of supporting nonverbal cues exerts a differential effect on perceptions of interpersonal attraction in H/NH, as opposed to NH/NH, interactions. Indeed, both foregoing factors may have contributed to the present unexpected result. A fourth significant interaction counter predicted effects on perceived state anxiety (positive type) remains to be explained. As previously observed, the NH source was perceived to be least anxious, the H source most anxious, in the moderate condition (b) of Factor A, i.e., under a 180 reduction of the verbal message component (cf. Figure 9). Magnitude of this difference produced the relevant, dis- ordinal A x B interaction. Conditions b1 and b also pro- 2 duced the only significant post hoc difference among simple effects in these data. Hence, the radical difference observed between dyadic types at the moderate level of V/NV affiliation has potential theoretic import and cannot be entirely ignored. By design, the moderate affiliation treatment combined a relatively normal number of affiliative nonverbal cues with an atypically low number of affiliative verbal cues. The H source was perceived to be least sociable (via mani— pulation check one) and most anxious in this condition versus other H/NH treatments. Perceptions of the HH source under a similar manipulation were significantly different in both instances. Ostensibly, these data indicate that messages composed of high nonverbal/low verbal, affiliative content are perceived and decoded quite differently by HH receivers depending on whether or not the source is phys- ically handicapped. As suggested previously, HH persons may expect H counterparts to be relatively inactive nonverbally and to compensate for minimal use of nonverbal cues by means of the verbal channel. If so, this expectation represents a reverse of the high nonverbal/low verbal combination intend- ed via the manipulation under discussion. Theoretically, 181 it is also conceivable that handicapped use of the non— verbal channel which exceeds HH expectations may be per- ceived as a manifestation of tension and/or anxiety. Intuitively, the same phenomenon commonly occurs at a higher threshold of nonverbal activity in more typical NH/ NH interactions. In sum, the foregoing discussion is highly speculative, suggesting the need for further research in this area. On present data, it does provide on plausible explanation for the atypically high assessments of state anxiety awarded the H source under a moderate manipulation of V/NV affili- ative cues, i.e., involving a reduction in the verbal channel. Nonsignificant Trends Technically, probability theory does not allow one to "bend" the meaning of statistical significance; nonsignif- icant findings are nonsignificant. However, given the exploratory nature and methodology of this study, certain nonsignificant trends in these data appear to inform pre- sent interpretation and may enlighten future research. Following examples are considered most salient. Communicative certainty ratings in H/NH dyads tended to decrease in the direction predicted, yet no significant differences occurred as a function of reducing the NV versus the V message component. In NH/NH dyads, both the V and the NV reductions produced significantly different 182 effects than the congruent (high) level of affiliation. Moreover, higher absolute values of certainty were report- ed by participants exposed to the handicapped source in both incongruent treatments of Factor A. Stated different- ly, in the presence of either type of incongruency partici- pants tended to be more certain of their ability to predict the H source's behavior. The largest difference (between the two sources) in favor of the H source occurred with a reduction in the number of NV cues exhibited (condition A, c). As in the clearer case of interpersonal attraction, incongruency appears to have less impact on certainty and the NV channel (affiliative cues in particular or NV cues in general) may be less salient in H/NH interactions. Interestingly, the manipulation check (two) data focusing on perceptions of V/NV congruency are quite comparable to the certainty data in Table 10. Plotted curves represent- ing each type of dyad in these data sets are remarkably similar, as are their rank orders of effects. Assuming that the manipulations (re: the two sources) were relative— ly consistent, V/NV cues exhibited by the H source were perceived as more congruent (than those exhibited by the NH source) at all three levels of V/NV affiliation. Once again, the largest difference in favor of the H source occurred at Factor A, condition c. That is, the absence of supporting NV cues had a considerably larger effect toward 183 reducing perceptions of V/NV congruency in NH/NH inter- actions. At the same time, differences in the congruency data attributed to the H source as a function of reducing the V versus NV component were negligible. (It should be noted that the manipulation check on V/NV congruency pro- duced a significant main effect for both independent variables.) Similarities between these two data sets lend addi- tional support to the notion that perceptions of V/NV con- gruency and the role of nonverbal cues may differ in H/NH interactions, particularly with regard to the relationship between affiliation and communicative certainty. It could, of course, be argued that the manipulations of V/NV congru- ency per se were unsuccessful, i.e., that the V/NV messages exhibited by the H source were in fact more congruent than those generated by the NH source. Clear consistency in the congruency data militates against this possibility. The issue will be discussed in greater detail later under limi- tations of the study. Present findings relative to state anxiety generally conflict with priOr research on H/NH communication.y If in- formative in no other sense, results obtained here provide evidence suggesting that state anxiety may not be a unidi- mensional variable as typically considered in previous research. Observed effects on three types Of state anxiety were reported in Chapter III as a function of the 184 tridimensional scale used to measure this variable. The first dimension reflected a positive affective state indexed by statements such as: "I felt secure"; "I felt at ease." According to scale design, low ratings on items of this nature were taken to represent high levels of state anxiety. Hence, this scale factor constitutes an indirect measure of state anxiety requiring an inferential "leap, i.e., from low security to high anxiety. On that account alone, results obtained via this scale factor were not surprising. Certain nonsignificant trends deserve notice. First, in terms of absolute values, decreasing state anxiety (positive) accompanied increasing affiliation in five of six experimental conditions. Exception to this pattern occurred in H/NH dyads at Factor A, condition c, i.e., less positive anxiety was reported under the NV versus V reduction. The main effect in this case reached a .11 probability level. Statistical power of the test was above the .90 level. Second, participants generally reported higher abso- lute levels of positive state anxiety when the message source was handicapped according to prediction. Two of three H/NH conditions conform with this pattern. The exception occurred at the low level of V/NV affiliation. There are at least two plausible explanations for this apparent disruption in the data. First, nonhandicapped 185 persons may operate under different nonverbal norms and expectations when interacting with the handicapped as sug- gested earlier. Second, the sole exception to the pattern in this case may be a function of an ineffective manipulation. Comparison of these data with results of the first manipulation check suggests likelihood of the latter possi- bility. A significant disordinal, A x B interaction oc— curred at the low level of Factor A in the manipulation data, i.e., the handicapped source was rated significantly more sociable (affiliative) than the nonhandicapped source only in this condition. Assuming an inverse relation be- tween sociability Of the source and state anxiety of the receiver as the present theory predicts, the positive anxiety data conforms quite consistently with results of this manipulation check in an inverted pattern of effects. On that account, it seems most reasonable to attribute the exceptional simple effect in this case to the manipulation. The trend toward an A x B disordinal interaction (p = .10) in these anxiety data (at the same location) tends to con- firm this conclusion. The second dimension of state anxiety (normal type) represented a more direct measure, indexed by statements such as: "I was tense." While the data indicated a sig- nificant relationship between V/NV affiliation and normal state anxiety, the pattern of column means was curvilinear. 186 None of the simple effects was significantly different. The lowest levels of normal state anxiety clearly accompa- nied the highest level of affiliation in both H/NH and NH/NH dyads. Although more anxiety was reported with lower levels of affiliation, no discernible difference occurred in H/NH interactions as a function of reducing the NV versus the V message component. Again, although the trend was nonsignificant, participants generally reported more anxiety of this type when the message source was handi- capped. In light of prior research, all effects on state anxiety observed in this study were generally less decisive than expected. Several factors which may have contributed to this outcome, in addition to the possible manipulation problem mentioned earlier, are discussed under limitations of the study. One nonsignificant trend in the interpersonal attrac- tion data is noteworthy. The predicted main effect from Factor A on attraction according to data collected via Byrne's IJS reached a probability below the .10 level. Actual power of the test, in this instance, exceeded the .80 convention. This scale was least reliable of instru- ments used in the study. Yet the magnitude, pattern and direction of effects obtained were highly consistent with data collected via the McCroskey/McCain instrument, except- ing the prior nonsignificant influence. On this account, the IJS data offers additional evidence in support of 187 theorem 5 and the general interpretation of effects on interpersonal attraction provided earlier. Limitations The present study was as much an experiment in method- ology as an investigation of real communication problems. Consequently, in retrospect, the list of limitations to which it was subject is more than typically legion. Some have already been implied. Many are more easily identified than eliminated. Subsequent review will focus on what appear to be major concerns. First, the study presupposed a significant interde- pendent relationship between verbal/nonverbal elements in ordinary human communication. Hence, the attempt was made to control and observe effects of eight verbal/nonverbal independent variables simultaneously. The videotape medium was used in lieu of live interactions to standardize these manipualtions. Some dilution of key nonverbal variables such as eye contact and facial affect was inevitable. Moreover, procedures used to incorporate the videotaped stimulus into a bona fide employment interview were tanta- mount to one-way communication between the confederate interviewee and participants role-playing the position of interviewer. Under these conditions, several patterns, if not the magnitude, of effects found in this study were remarkably clear and heuristically useful, encouraging future research via two-way, live interactions. 188 Second, hindsight suggests two important limitations relative to experimental manipulations used in the study. Evidence cited previosuly indicated likelihood of a defec- tive manipulation at the low extreme of Factor A. In addition, results of the study repeatedly show consistent patterns, yet low magnitude, of effects from Factor B. The fact that participants did not encounter the handicapped source directly in live interaction may account for this outcome to some extent. It is also possible that the H/NH manipulation was not sufficiently strong to project an effective image of the handicapped source via the videotape medium. Given their complexity, manipulations of both inde- pendent factors should have been pretested extensively to avoid the data interpretation problems which eventually developed in this case. A third critical problem involved the appropriate number of verbal/nonverbal cues to be eliminated and/or retained in treatments representing the incongruent, medium and low levels of V/NV affiliation. In lieu of quantita- tive norms representing the typical affiliative behavior of an interviewee, a standardized employment interview was constructed to represent the high (normal) treatment con- dition with the assistance of a professional actor. The actor role-played an enthusiastic NH interviewee, generat- ing what he perceived to be a "normal" number of the rele- vant verbal/nonverbal cues. The "normal" number of either 189 verbal or nonverbal cues was then reduced, somewhat arbi- trarily, by 40-50% to construct the medium and low level treatments. In lieu of additional information, it is not certain that the verbal versus nonverbal reductions were either qualitatively comparable or quantitatively realistic relative to what typically occurs in the "real" world. This problem suggests the need for further descriptive research to establish quantitative guidelines in this area. Fourth, these data were collected during a bona fide employment interview. The strategy was to increase respond- ents' motivation and establish the ecological validity of required tasks. Participants were being interviewed for summer work at a time when employment of this type was extremely difficult to obtain. Hence, the data was col- lected in a highly competitive context. Some evidence emerged to indicate that this feature may have introduced a response bias relative to certain dependent variables, particularly where the handicapped source was involved. For example, variance in the state anxiety data decreased systematically across the positive-normal-negative dimen- sions of this variable. In addition, on the average participants consistently utilized decreasing proportions (.44, .20 and .05) of the available scoring range for each scale factor frOm the positive to negative dimension Of state anxiety. This pattern suggests that respondents may have been reluctant 190 to report state anxiety (increasingly negative types, in particular) assuming that this information could jeOpardize their opportunity for employment. A similar response bias may have contributed to consistently low variance in these data relative to the handicapped stimulus source. Fifth, other noteworthy limitations involve the sample, design and measurement of dependent variables. The sample was relatively young (X = 22.5 years) and well educated (X = 15.7 years of formal education). Most participants were advanced university students. Theoretically, given the recent emphasis on rights and needs of handicappers, particularly in university contexts, a sample of this type may be more tolerant of H versus HH differences than an older sample with less education. Hence, generalizability of these results to the typical selection interviewer is questionable. In addition, level of education for female participants, married persons in general, and married males in particular were nonrandomly distributed across treatment conditions. Production costs of videotaped stimuli and available resources dictated an abbreviation of the original 2 x 4 factorial design. Informative value of this study would likely have been enhanced with inclusion of an additional treatment at the lower extremity of Factor A, i.e., a simultaneous reduction of both verbal and nonverbal affil- iative cues. Present results suggest the likelihood of 191 more definitive evidence had the V/NV affiliation continu- um been extended in this manner. Factor analyses on several dependent variables indi- cated the need for more precise, reliable measurement. The original communicative certainty scale was designed to include a retroactive dimension, related to decoding pro- cesses. These scale items failed to load adequately on one or more trial factors and were eventually eliminated. Thus, certainty data used in the study was exclusively of the proactive type. Given the central role of this vari- able, a fair test of the underlying theory will require more comprehensive, valid measures of communicative certainty. Measurement of state anxiety and perceived state anxiety was less than optimal. Seven items were deleted from the 20-item state anxiety scale in order to reach an acceptable three-factor solution. This reduction amounted to a 35% loss of costly information. In turn, the remain- ing data were dispersed across the final three-factor solu- tion. Not surprising, the magnitude of observed effects on state anxiety was relatively weak; their interpretation somewhat problematic. Similar limitations apply to the perceived state anxiety data obtained via a modified version of the same scale. 192 Directions For Future Research Several general ideas for additional research were suggested in conjunction with prior limitations. Three specific queries appear to hold promise for extending our present knowledge of H/NH communication. First, focal point of the original research question was the employment interview. However, the underlying theory and subsequent research were necessarily general, applying more realisti- cally to any initial encounter between handicapped/non— handicapped persons. I am convinced from present results that both independent variables and the five dependent variables observed here (with possible exception of per— ceived state anxiety) are relevant factors warranting additional study. The extent to which these variables alone (or in combination with relevant others) directly affect the all important decision-making process in an employment interview context involving the handicapped is not clear from this study. For example, it cannot be assumed automatically that variables such as interpersonal or social, as opposed to task, attraction significantly affect the ultimate selection process. Additional research focused more directly on this question would likely prove useful. Second, certain trends in results of this study raised a seemingly important question regarding a priori expecta- tions that may operate when NH persons engage the 193 handicapped in initial interaction. If in fact NH expectations differ, for example, relative to incongruent verbal/nonverbal messages generated by handicapped persons, strategies for developing more effective communication between these individual types will eventually require additional information in this area. Research examining the manner in which expectations regarding V/NV congruency alter NH assessments of the handicapper's source credi- bility should be particularly useful re: the employment interview. Finally, evidence from this study relative to the role of nonverbal communication in H/NH interactions raised, perhaps, more questions than it answered. Similarities and differences regarding the function of nonverbal cues in H/NH versus NH/NH interactions are by no means clear. Knowledge of the relative import of variables such as eye contact, facial affects, head-nods and larger body move- ments should be extremely helpful to the handicapper. More specifically, the extent to which these variables (and/or their verbal counterparts) can be used as substi- tutes to compensate for a particular physical disability would be invaluable information. Advancements toward more effective communication between handicapped/nonhandicapped persons await in depth research, particularly on the non- verbal channel and intricacies of its relationship with concomitant verbal cues. APPENDIX A STANDARDIZED EMPLOYMENT INTERVIEW APPENDIX A STANDARDIZED EMPLOYMENT INTERVIEW Treatment Condition : Verbal Affiliation ( V -Aff) Nonverbal Affiliation (NV - Aff ) Symbol Key" Nonverbal cues: O O = direct eye contact 0 O = diverted eye gaze = head nodding = hand/arm gesture : © = definite smile Verbal cues: I) = declarative statement PC = positive verbal content VR = positive verbal reinforcer 3X2 = person-oriented question * A11 nonverbal symbols inserted in the following text refer to be- haviors being exhibited by the confederate (C). Nonverbals held constant during a given speech by the surrogate subject (S) are inserted at the be— ginning of each utterance. Intermittent behaviors are given their approx- imate location during an utterance by either interactant. A pleasant facial expression is maintained by §_throughout the interaction representing this experimental condition. 194 195 INTRODUCTION H} Spontaneous greeting E; How'd you do! I'm Tom Holtrop. H} Doug Milton, here. Glad to meet you, Tom. __ My pleasure. ‘H: Have a seat . . . make yourself comfortable. 9; Thank you. STRUCTURING THE INTERVIEW H: ( O O ) We appreciate your interest in our firm, Mr. Holtrop, and the opportunity to talk with you about employment. ( r , © ) I'm the coordinator for a team of about 8 persons doing applied research--most1y in business and industrial settings. Basically, we're looking for someone with a behavioral bent--to help collect and code data. (VR, r ) But, before we get too far down stream, I'd like to give you some idea of what to expect here, what we hope to ac- complish, and so on. “1 (PC) Great! (D) I'd appreciate that. ( C) C) ) Let me say, first off, this isn't 3 "Salt Talk" (: <:> ) IU) so I hope we can be somewhat informal. (VR, r ) The basic philosophy of our firm, Tom, is that "people don't live in order to work" as our middle class work ethic suggests. We C. 6. I O |UJ 196 think they "work in order to live." ( r , © ) So, we try to approach personnel with less emphasis on the o'1e "how much can you do for us" routine. ( r ) (PC) (D)That's refreshing( Q 0 );(PC) (D)( 3 )makes sense to me. ( C) C) ) Obviously, we'll be looking at your training and experience profile, SE cetera ( r) ) But, quite frankly, Tom, at this point we're interested in a little broader picture, e.g., your personal perspective, reactions to other jobs, your ideas, ( r) ) some assessment of your own capabilities and how to maximize their use. In a nutshell, we'd like to know "where you're at" as well as where you've been. Uh . . . I see you majored in Sociology. How'd you happen to decide on that area? ( r ) BACKGROUND/ACADEMIC INTERESTS E; (PC)(D) Good question . . . (D) I've asked that ( Q ) myself a couple 0' times. ( © ) (D) As I look back ( O O ), there were several reasons. (0) Number one, ( E ) I had an un- usually stimulating instructor in my first Hgg_course at Penn State. (D) We all have our role models, you know. (0) Worked like ( Q ) Hades for him ( O O ), (D) That helped "turn me on"! (D) But probably the turbulent times had more to do with it. (D) I was an undergrad when King an' Kennedy were assassinated , (D) th' ( g ) riots near my home in L.A., peacemakers and protests . . . (D) th' campus was a caldron, of course. (D) SO, studying social stratification an' race relations seemed 10. 11. H1 IV) In IUD 197 ( Q , © ) like the only relevant thing in the world. ( C) C) ) Know what you mean--had our problems in Detroit too, you recall. (VR, r ) By the way, I'm here to field your questions tOO--peop1e are sometimes better known for their questions than for their answers. ( © ) This is a two-way street, so don't hesitate to check us out. (VR, r ) Anything in particular on your mind at this point? (PC) (D) Appreciate that . . .(D) uh ( Q 0 ) yes, . . . couple 0' things. (D) I'm fairly clear on the job description, (D) but I ( Q ) do think its important to know as much as possible before- hand about the people you'll work with . . . ( © ) ( O O ) Right! ( O O ) (PQ) I'd be interested in ( E ) your professional background, (PQ) your own expectations on the job, style of management, an' so on? (O 0) Turn about's fair play, huh? ( r , © ) I spent four years in the service after high sch001--working mostly in naval hospitals as a pharmacist's mate. ( r ) At the ripe o'le age of 26, I went back to school full time--Wayne State in Detroit-—majored in Psychology. ( r ) Did my Masters at Ohio State in Industrial Psych. Then, GM hired me fresh out of school. But, that was essentially a consulting job, I'm more into research on managerial potential and performance. (VR, r ) So, I came to SRA about six months ago. 198 D . SUBSTANTIVE I SSUES 12. £2 (PC)(D)Interesting . . . sounds like you've been around. ( Q 0 ) Uh . . . as a matter Of personal curiosity . . . (0) this ( © ) is gonna sound like a Carter summit conference . 13. |cn ( O O ) That's okay--go ahead! 14. (PQ) With your ( Q ) background, I was just wondering what( 0 O ) IO you see as the number one problem or challenge in business an' industry today? 15. I (I) ( O O ) That's a "biggee";course it depends on what part '0 the jungle you're coming from. (VR, r ) We're all preoccu- pied with energy and inflation right now. But, a major problem, as I see it, is the misuse or waste of human energy and re- sources. ( r ) There's a tremendous amount of underutilized talent and skill in most organizations I've studied. (VR, r ) 16. In (PC) (D)That's perceptive ...(PC)(D)couldn't agree more.( 0 Q ) (PQ) What's the cause, ( © ) training, supervision or . . . ? 17. | (I) ( C) C) ) Well, causes are hard to come by, as you know. (VR, r ) I don't think its generally true to say that managers just don't care. It's more the system as a whole. Corporate structures are so powerful and rigid they simply don't allow the individual to participate in the production process-- in innovative ways. (VR, r ) For example . . . better yet maybe I can illustrate the point from your own experience-~if you'll elaborate on two or three of the jobs you've listed, what you're responsibilities were (VR, r ), how you really felt about the job, et cete ra . E. 18. 19. 20. 21. 22. 23. 24. 199 PREVIOUS EXPERI ENCE/PERSPECTIVES H: Icn In I (I) In [(1) In (PC)(D)Sure,be glad to . . . ( .0 ) (D) Fisher Body was a summer job--(D) drove a fork lift carting ( Q ) quarter panels to assembly. ( C) O ) (D) It was a fairly responsible position because of safety regulations ( Q Q ) (D) but like driving a car, it becomes habitual. (D) You do it without ( Q ) thinking, (0) so it gets pretty monotonous! ( O O ) Exactly-~if you can do it without thinking, it's a bore. How 'about the office job? ( r ) ( O O ) (D) That was part-time during school ( Q ) in the registrar's Office here at MSU . . . (D) filing student tran- scripts an' records. (D) Somebody has t' do it, (D) but it doesn't take a lot 0' "smarts" ( Q , © ) to put a piece 0' paper in the right slot. ( O 0) Got it! ( O O ) (D) The best job I've had was with the city 0' Saginaw. (D) . . . a pilot project placing juveniles in foster homes. (D) The idea was to improve ( Q ) long term results by screen- ing potential parents psychologically. ( C) O ) You say "your best job." Why did that particular job "turn you on"? ( r ) ( O O ) (D) Well, for one thing it tied right ( Q ) in with my academic training. (D) Number two, . . . felt like I was doing something worthwhile. (D) The final outcome was important because ( g ) (D) I had something to do with it. 25. 26. 27. 28. 29. 30. H1 IUJ |O IU) H1 200 ( C) C) ) Precisely, that's the point! We did an efficiency study recently for a GM plant in Georgia. And, the only major problem was 400 hourly workers--dissatisfied because they had no responsibility ( r, ) for planning their work, quality con- trol, costs, absenteeism (VR). They had to restructure . so that people with the capability could function in a wider range of activity. What type of supervision did you have in Saginaw? ( r ) (PC)(D) Was hoping you'd ask( © , O O ) (D) It was quite loose ( Q ) 'an unstructured, but not laissez-faire. (D) We were trained on procedures,(D)what had to be done, (D) an' given a long range goal, ( C) C) ) (D) but we ( Q ) had to organize the work ourselves (D) an' get it done.( 0 O ) ( C) C) ) What were the advantages of that setup? ( r ) (D) Made the work alot more challenging. ( O O ) (D) We were gathering several different types of "info" on each family, (0) so we ( Q ) never ( O O ) knew how a given case would turn out until all the data was analyzed. (D) Once I came back Friday evening to "wrap up" an interesting case. (D) An' I was so( 0 O ) anxious to get the results, (D) th' sun ( Q , © ) came up Saturday morning before I realized it. ( C) C) ) That says it all! How 'bout drawbacks? Any disad- vantages to loose supervision ( r ) as you see it? Oh, sure! (D) "Nobody's perfect" ( © ) as V. W. says. ( O O ) (D) Think the research indicates that some people are more 31. 32. 33. |m IV! 201 productive--(D) apparently they feel ( Q ) more secure with closer'surveillance. (0) Others need regular benchmarks from the brass. ( C) O ) (D) It is difficult to know how you're ( Q ) doing if you don't get regular feedback. (0) I'd be interested in your view on a related issue. (PQ) To what, extent ( O O ) do you feel subordinates should use their own judgment to ( Q ) solve problems? ( C) C) ) I think they should be free to function with maximum autonomy--as adults. (VR, r ) Not that everybody can do his own thing--that's chaos. ( r ) But, in private life, the average individual is capable of exercising a good deal of dis- cretion. They make responsible judgments on financial matters, education investments _e_t_ _ce_te_r_a. (VR, r ) Then they come to work an' some manager makes all the critical decisions for them. They're frustrated because they're expected to behave like im- beciles on the job (VR, r , © ) (PC)(D)anrapproach is quite employee—oriented, (PC)(D) which I appre- ciate! (PQ) But, ( O O ) what about performance? (D) Employees 93 have to produce ( © ) for a company to stay in business. ( C) C) ) Sure do! . . . Productions the "name of th' game" if we want to survive. (VR, r ) But, you see, I don't be- lieve people are basically lazy by nature. They work damn hard for things they really want (VR, r ). As McGregor says, leadership in an organization has to create conditions so that individuals can achieve their own goals H§§£_by directing effort toward the success of the enterprise, e.g., if SRA can give you 34. 35. 36. 37. I ('1 IV) I O IU) 202 the opportunity to maximize your own long-range objectives (VR, r ), I believe you'll produce more than enough to satisfy us! (PC)(D)Thinkyou're right! ( © ) But ( O O ) . . . (D) suppose an employee wasn't producing up to your expectations. ( O C) ) (PQ) How would ( Q ) ya try to motivate him?( I O ) ( C) C) ) I'd try to find out what's missing for that person in that particular job. (VR, r ) Could be more money, but my theory is that "man doesn't live by bread alone." ( © ) Usually there's some intrinsic reward missing-—a sense of accom- plishment, control over one's own destiny, or self-realization. (VR, r ) If the company can provide that, the work will come naturally in most cases. Why did 1’29. work so hard in Saginaw? (PC)(D)Know exactly ( r ) what ya' mean. ( O O ) (D) Wasn't it Herzberg who said (D) we're motivated more by things like personal satisfaction, achievement, a sense of worth an' respons- ibility. ( O O) (D) In my ( Q ) case, it was doing some- thing worthwhile an' ( © ) (0) doing it fairly well.( 0 O ) ( O O ) Let me ask a similar question from a slightly different perspective. (VR, r ) This may be "off the wall," . . . but if you were director of a group like mine-~where some subordinates have expertise you may not have ( r )--would you focus on keeping up with technical decisions? Or would you be inclined to concentrate on just coordinating their activity? ( © , r ) 203 38. _G: ( O O ) (PC) (D) Interesting question! (D) Let's put it this way--(D) I don't think ( O O , Q ) you could manage a professional football team if you've never played th' game, ( © ) okay! (D) Ya' need some experience n' expertise ( Q )--even to coordinate. (D) But a manager can't play th' game ( Q ) for his players. (0) I'd say a good coach relies as much ( O O ) as possible on the know—how of those (D) who are out where the "rubber hits the road.” (D) Think I'd focus on the larger game plan. ( © ) ( C) 0) That's well said. Like your analogy. ( © , VR) 39. H: You sound like pretty fair manager material. There is one other question I meant to ask. ( r , VR) How's your health—-any problems we should be aware of . . . ? ( Q . ) 40. H: (D) Nothing ya' ( Q ) can't see! (D) I don't jog, (D) but I get ( Q ) around pretty well. (D) No, ah'm healthy as a bear in abeehive. ( © ) F. CLOSURE 41. _S_: ( O 0) Good, Tom, I'm satisfied at this point. Anything else you'd like to discuss?( O O ) 42. 9: (PC) (0) No,this has been most ( © ) informative. (0) That pretty well covers it (D) as far as I'm concerned. 43. _S_: ( O 0) We still have our homework to do, of course, ( r ) but we're impressed at this point. I hope you'll give us serious consideration. ( r ) 42. 43. 44. If) I (I) I O 204 (PC)(D)Certain1ywill! ( © ) (PC)(D) Appreciate your time.( Q , handshake) ( C) C) ) Thanks for coming in . . . we'll be in touch.( r , VR) (PC)(D)Thank you ! ( © ) APPENDIX B SRA Employment Application SRA Employment Interview Questionnaire Part I Part II APPENDIX B SRA Employment Application NAME: ADDRESS: (Street) (City) (State) (Zip Code) LOCAL PHONE NUMBER: Area Code_____ Number *Marital Status: ____Married ____Single ____Widowed .____Divorced Veteran of the Military Service: ____yes - no Education: Please indicate the number of years of formal education you have received including elementary school, high school, and college. years * * Age: years of age Sex: Male Female Social Security Number: Employment: Please provide the following information concerning your most recent or current employment. (Name of Firm) (Name of Immediate Supervisor) (Address of Firm) Have you ever been convicted of a felony? yes no Please provide two references, their addresses, and current telephone numbers if possible: Name: Address: Telephone Number: Name: Address: Telephone Number: (*This information is being solicited fer EOE purposes.) 205 SRA EMPLOYMENT INTERVIEW QUESTIONNAIRE PART I SRA EMPLOYMENT INTERVIEW QUESTIONNAIRE PART I Name: Social Security No. GENERAL INSTRUCTIONS Some individuals are more aware of their own feelings, attitudes, and internal reactions than are others. Some people are able to read and interpret the reactions of another person more accurately than are others. The following questionnaire has been carefully designed to tell us how well you are able to identify and accurately report your own initial reactions to unfamiliar persons. It will also measure your ability to tell how unfamiliar persons have reacted to you. There are no right or wrong, good or bad answers to these questions. There are no hidden meaHIngs. We are not interested in what speciflc reactions you had. Rather, our work requires persons who are well "in touch" with their own evaluations of others and are willing to report them honestly and accurately. You are most likely to qualify for one of our positions if you g9 29E attempt to provide answers that seem more socially acceptable or "expected" by an employer than your actual reactions. If additional information is required, physical measures will be taken during a live interview with Mr. Holtrop. These measures will determine more clearly what internal reactions you actually experienced. Please be assured that your responses will be held in strictest confidence. ' 206 .207 In an interview situation, some persons disclose important infOrmation about themselves quickly and easily. Their personal characteristics and typical behavior patterns can Often be predicted quite confidently by the interviewer from first impressions. On other occasions, the person interviewed may be more reserved so that further time and effort is necessary to obtain an adequate evaluation. Relying on information from your interview, the fellowing questions ask you to express how much confidence you would have in making certain evaluations and pre- dictions about the person you interviewed. You may use any evidence Obtained during the interview. Place a check (/) in the blank beside the best response from your point of view. Please read each question carefully and note that the response alternatives sometimes appear in reverse order. 1. How certain are you of your general ability to predict the way this person is likely to behave in different situations? very certain certain somewhat certain somewhat uncertain uncertain very uncertain 2. How certain are you that you understood or interpreted this person correctly when he/she spoke to you? very uncertain uncertain somewhat uncertain somewhat certain certain very certain 3. How certain are you that you know or can predict some of the moral and social values this person holds? very certain certain somewhat certain somewhat uncertain uncertain |||||| very uncertain 208 How certain are you that you know whether or not this person liked you? very uncertain uncertain somewhat uncertain somewhat certain certain very certain How certain are you that you know or can predict the basic attitudes that guide this person's behavior? very certain certain somewhat certain somewhat uncertain uncertain very uncertain During an initial interview, some people respond the way we expect them to respond; others are quite unpredictable. How certain are you very uncertain uncertain somewhat uncertain somewhat certain certain very certain How certain are you that you know or can predict how this person about himself/herself? very certain certain somewhat certain somewhat uncertain uncertain very uncertain 209 8. How certain are you about whether or not this person is highly motivated toward his/her work? very uncertain uncertain somewhat uncertain somewhat certain certain very certain 9. How certain are you that this person is genuinely concerned about whether or not others are at ease in his/her presence? very certain certain somewhat certain somewhat uncertain uncertain very uncertain 10. How certain are you Of the way he/she would likely treat you as a person if you were working in the same Office? very uncertain uncertain somewhat uncertain somewhat certain certain very certain A number of statements which people have used to describe themselves are given below. Read each statement. Then check (/) the response below which indicates how you felt during the interview. Do not spend too much time on any one item. Give the answer which seems to best describe your feelings while you were with the other person. Please note that the resppnse alternatives sometimes appear in reverse order. 11- I felt calm- 12. I felt secure. _____ not at all _____ very much so ______somewhat' _____ moderately so _____ moderately so _____ somewhat ______very much so ______not at all 13. 19. 15. 16. 17. 18. 19. I was tense. not at all somewhat moderately so very much so I was regretful. very much so moderately so somewhat not at all I felt at ease. not at all somewhat moderately so very much so I felt upset. very much so moderately so somewhat not at all I was worried over possible misfortunes. not at all somewhat moderately so very much so I felt rested. very much so moderately so somewhat not at all I felt anxious. not at all somewhat moderately so very much so 210 20. 21. 22. 23. 24. 25. 26. I felt comfortable. very much so moderately so somewhat not at all I felt self-confident. not at all somewhat moderately so very much so I felt nervous. I was very much so moderately so somewhat not at all jittery. not at all somewhat moderately so very much so I felt "high strung." I was very much so moderately so somewhat not at all relaxed. very much so moderately so somewhat not at all I felt content. ||| not at all somewhat moderately so very much so 211 27. I was worried. 29. I felt joyful. very much so very much so moderately so moderately so somewhat somewhat not at all not at all 28. I felt over-excited and "rattled." 30. I felt pleasant. not at all not at all somewhat somewhat moderately so moderately so very much so very much so Below are several items asking you to evaluate the person you inter- viewed. Please read each item carefully. Then place a check (/) in the blank beside one response that best describes your assessment of this person. Nora-That the response alternatives sometimes appear in reverse order. 31. How intelligent do you think he/she is? I believe that this person is very much above average in intelligence. I believe that this person is above average in intelligence. I believe that this person is slightly above averagg in intelligence. I believe that this person is average in intelligence. I believe that this person is slightly below averag§_in intelligence. I believe that this person is below averag§_in intelligence. I believe that this person is very much below averagg_in intelligence. 32. How well do you think you would like this person if you became better acquainted? I feel that I would probably dislike this person vegy much. I feel that I would probably dislike this person. I feel that I would probably dislike this person to a slight degpee. I feel that I would prObably neither particularly like nor particularly dislike this person. I feel that I would probably like this person to a sligHt degree. I feel that I would probably like this person. I feel that I would probably like this person vepy much. 33. 34. 35. 36. 212 How difficult do you think it would be to work with him/her? I believe it would be extremely easy to work with this person. I believe it would be easy to work with this person. I believe it would be easy to work with this person to a slight degree. I believe it would be neither particularly difficult nor particularly easy to work with this person. I believe it would be difficult to work with this person to a slight degree. I believe it would be difficult to work with this person. I believe it would be extremely difficult to work with this person. How well adjusted do you think he/she is? I believe that this person is extremely well adjusted. I believe that this person is well adjusted. I believe that this person is well adjusted to a slight degree. I believe that this person is neither particularly maladjusted nor particularly well adjusted. I believe that this person is maladjusted to a slight degree. I believe that this person is maladjusted. ..1 believe that this person is extremely maladjusted. I think he/she could be a friend of mine. strongly agree agree somewhat agree undecided somewhat disagree disagree strongly disagree It would be difficult to meet and talk with him/her again. strongly disagree disagree somewhat disagree undecided somewhat agree agree strongly agree 213 37. We could never establish a personal friendship with each other. strongly agree agree somewhat agree undecided somewhat disagree disagree strongly disagree 38. I would enjoy having a friendly chat with him/her again. strongly disagree disagree somewhat disagree undecided somewhat agree agree strongly agree 39. He/she just wouldn't fit into my circle of friends. strongly agree agree somewhat agree undecided somewhat disagree disagree strongly disagree A number of statements which people have used to describe others are given below. Read each statement. Then check (/) the response below which in- dicates how you think the other person felt during the interview. Do not spend too much time on any one item. Give the answer which seems to best describe his/her feelings while you were with the other person. Please note that the response alternatives sometimes appear in reverse order. 40. He/she felt pleasant. 41. He/she felt joyful. ______not at all ______ very much so _____ somewhat moderately so _____ moderately so —_—__.somewhat __ very much so ___: not at all 42. #3. 41+. 95. 46. 47. 48. He/she felt over-excited and "rattled." not at all somewhat moderately so very much so He/she was worried. very much so moderately so somewhat not at all He/she felt content. very much so moderately so somewhat not at all He/she was relaxed. not at all somewhat moderately so very much so He/she felt ”high strung." very much so moderately so somewhat not at all He/she was jittery. not at all somewhat moderately so very much so He/she felt nervous. very much so moderately so somewhat not at all 50. 51. 52. 53. SH. 55. He/she felt self-confident. not at all somewhat moderately so very much so He/she felt comfortable. very much so moderately so Ill somewhat not at all He/she felt anxious. not at all somewhat Ill moderately so very much so He/she felt rested. very much so moderately so somewhat not at all He/she felt calm. not at all somewhat moderately so very much so He/she was worried over possible misfortunes. not at all somewhat moderately so very much so He/she felt upset. very much so moderately so somewhat not at all 213 56. He/she felt at ease. 58. He/she was tense. not at all not at all sanewhat somewhat moderately so moderately so very much so very much so 57. He/she was regretful. 59. He/she felt secure. very much so very much so moderately so moderately so somewhat somewhat not at all not at all The following items are designed to tell us how well you were able to project yourself into the interviewer's role. In addition to an accurate report of your own reactions to the other person, it is extremely important during an interview to know how he/she reacted to you. Again, assume that you actually conducted the videotaped interview so that Mr. Holtrop's verbal and nonverbal responses were actually addressed directly to you. Then check (V ) SE answer that best describes what you Helimg his assessment was 9: ygu in the role of the interviewer. Note that the response alternatives sometimes appear in reverse order. 60. How well adjusted does he believe you are? He believes that I am extremely well adjusted. He believes that I am well adjusted. He believes that I am well adjusted to a slight degree. He believes that I am neither particularly maladjusted nor particularly well adjusted. _ He believes that I am maladjusted to a slight degree. He believes that I am maladjusted. He believes that I am extremely maladjusted. 61. How well do you think this person would like you if you became better acquainted? I feel that he would probably dislike me very much. I feel that he would probably dislike me. I feel that he would probably dislike me to a slight degree. I feel that he would probably neither particularly like nor particularly dislike me. I feel that he would probably like me to a slight degree. I feel that he would probably like me. I feel that he would probably like me very much. .— 216 How difficult do you think this person believes it would be to work with you? He believes it would be extremely easy to work with me. He believes it.would be easy to work with me. He believes it.would be easy to work with me to a slight degree. He believes it would be neither particularly difficult nor particularly easy to work with me. He believes it.would be difficglp_to work with me to a slight degree. He believes it would be difficult to work with me. He believes it would be extremely difficult to work with me. How intelligent do you think this person believes you are? He believes that.I am.very much below average in intelligence. He believes that I anlbelow average in intelligence. _————-— He believes that I am slightly below average in intelligence. He believes that.I anlaveragg in intelligence. —————o——-—— He believes that I am.slightly#above average in intelligence. He believes that I am above averag§_in intelligence. He believes that I am.very much above average in intelligence. ___—— I believe this person feels that I just wouldn't fit into his circle of friends. strongly agree agree somewhat agree undecided somewhat disagree disagree strongly disagree I believe this person would enjoy having a friendly chat with me again. strongly disagree disagree somewhat disagree undecided somewhat agree agree strongly agree 217 66. I believe he feels that we could never establish a personal friendship with each other. strongly agree agree somewhat agree undecided somewhat disagree disagree strongly disagree 67. I believe he feels that it would be difficult to meet and talk with me again. strongly disagree disagree somewhat disagree undecided somewhat agree agree strongly agree 68. I believe this person feels that I could be a friend of his. strongly agree agree somewhat agree undecided somewhat disagree disagree strongly disagree SRA EMPLOYMENT INTERVIEW QUESTIONNAIRE PART II Name: Social Security No.: The following questions are designed to double check your general impressions of.Mr. Holtrop. Please check (v/) one response to each question and note that the alternatives sometimes appear in reverse order. 69. How truthful do you think the responses provided by Mr. HoltrOp were during the interview: Not truthful at all Somewhat less than truthful Somewhat truthful very truthful _— 70. How dissimilar (different) were Mr. HoltrOp's verbal messages from his nonverbal messages during the interview: Not dissimilar at all Somewhat dissindlar Dissimilar very dissimilar 71. How friendly did you perceive Mr. Holtrop to be during the interview? ___m*_ very friendly _______'Friendly Somewhat unfriendly Not friendly at all 72. In general, how trustworthy do you think Mr. Holtrop is? very untrustworthy Eknewhat untrustworthy Somewhat trustworthy very trustworthy 73. How consistent were Mr. Holtrop's facial expressions, gestures, and body movements with the meanings of the words he spoke during the interview? very consistent Somewhat consistent Somewhat inconsistent very inconsistent 213 74. 219 How sociable did you perceive Mr. Holtrop to be during the interview? __ “_Not sociable at all -Somewhat unsociable Sociable very sociable The following information will not affect your evaluation for the present jOb. It may help us to determine which of our current projects and in what capacity you could most comfortably serve. 75. 76. 77. 78. 79. Have you ever worked in an organization providing professional services to: a. Minority persons? ._ Yes ___ No b. Handicapped persons? Yes __ No c. Elderly persons? Yes No (If you checked the "No" response to each of the above, go to question #79) If your response was "Yes" to any of the options in question #75, how long have you. worked with: a. .Minority persons? Yrs. Months b . Handicapped persons? Yrs . Months c. Elderly persons? Yrs. .Months Briefly describe the nature of the duties you performed. Where were you employed in this capacity? Name of firm or agency Address of firm or agency Have you ever maintained a regular social relationship with: a. A minority person Yes No ___—___ _- b. .A handicapped person Yes NO ___-I- c. An elderly person Yes No ___—_— 80. 220 If your response was "Yes" to any of the options in question #79, briefly describe the nature of this relationship(s). For example, what type of relationship(s) did you have and how frequently did you interact with this individual(s)? BIBLIOGRAPHY BIBLIOGRAPHY Allen, K.H., & Cinsky, M.E. General characteristics of the disabled population. Social Security Survey of the Disabled: 1966, Report Number 19, 1-46. Altman, I., & Taylor, D.A. SOOialypenetration: The development of ipterpersonal relationships. New York: Holt, Rinehart and Winston, 1973. Argyle, M. Social interaction. New York: Atherton Press, 1969. ' Argyle, M., Alkema, F., & Gilmour, R. The communication of friendly and hostile attitudes by verbal and non- verbal signals. European Journal of Social Psy- chology, 1971, i, 385-402. Argyle, M., & Kendon, A. The experimental analysis of social performance. Advances in Experimental Social Psychology, 1967, 2, 55-98. Argyle, M., Salter, V., Nicholson, H., Williams, M., & Burgess, P. The communication of inferior and superior attitudes by verbal and nonverbal signals. British Journal of Social and Clinical Psychology, 1970, 2, 222-231. Bales, R.F. Interaction process analysis. In D.L. Sills (Ed.), International encyclopedia of the social Sciences. New York: Crowell-COIlier and'Macmillan, 1968, 465—471. Bales, R.F. Interaction process analysis. Reading, Mass.: Addison-Wesley, 1950. Berger, C.R. The acquaintance process revisited: Explorations in initial interaction. Paper presented to the Annual Convention of the Speech Communication ‘ Association, New York, 1973. Berger, C.R., & Calabrese, R.J. Some explorations in initial interaction and beyond: Toward a develOp- mental theory in interpersonal communication. Human Communication Research, 1975, i, 99-112. 221 222 Berger, C.R., & Larimer, M.W. When beauty is only skin deep: The effects of physical attraEtiveness, sex and time on initial interaction. Paper presented to the annual convenEIon of the International Communica- tion Association, New Orleans, 1974. Berkowitz, L. Re3ponsibility, reciprocity, and social distance in help-giving: An experimental investiga- tion of English social class differences. Journal of Experimental Social Psychology, 1968, 3, 46-63. Bettinghaus, E.P. Persuasive communication. New York: Holt, Rinehart and Winston, 1973. Blalock, H.M. Theory_construction: From verbal to mathematical formulations. Englewood Cliffs, N.J.: Prentice-Hall, 1969. Borgatta, E.P. A systematic study of interaction process scores, peer and self-assessments, personality and other variables. Genetic Psychologngonographs, 1962, Hi, 219-291. Brooks, W.D., & Emmert, P. Interpersonal communication. Dubuque, Iowa: Wm. C. Brown Company, 1976. Bugental, D.E., Kaswan, J.W., Love, L.R., & Fox, M.N. Child versus adult perception of evaluative messages in verbal, vocal, and visual channels. Developmental Psychology, 1970, 3, 367-375. Burgoon, J.K., & Saine, T. The unspoken dialogue: An introduction to nonverbal communication. Boston: Houghton Mifflin Company, 1978. Byrne, D. The attraction paradigm, New York: Academic Press, 1971. Byrne, D., & Nelson, D. Attraction as a linear function of proportion of positive reinforcements. Journal of Personality and Social ngchology, 1965, i, 659-663. Carter, L.F. Evaluating the performance of individuals as members of small groups. Personnel Psychology, 1954, 1, 477-484. Clatterbuck, G.W. Attributional confidence and uncertain- ty in initial interaction. Human Communication Research, 1979, 5:2, 147-157. Cohen, J. Statistical power analysis for the behavioral sciences. 223 Comer, R.J., & Piliavin, J.A. As others see us: Atti- tudes of physically handicapped and normals toward own and other groups. Rehabilitation Literature, 1975, HQ, 206-221, 225. Condon, W.S., & Ogston, W.D. Sound film analysis of normal and pathological behavior patterns. Journal of Nervous and Mental Disease, 1966, 143, 338-347. Couch, A.S. Psychological determinants of interpsrsonal behavior. Unpublished doctoral dissertation, Harvard UniverSity, 1960. Cited in A. Mehrabian, & S. Ksionzky, Some determiners of social interaction, Sociometry, 1972a, Hg, 588-609. Cozby, P.C. Self-disclosure, reciprocity and liking. Sociometty, 1972, 3;, 151-160. Cronbach, L.J. Coefficient alpha and the internal struc- ture of tests. Psychometrika, 1951, i9, 297-334. Davis, F. Deviance disavowal: The management of strained interaction by the visibly handicapped. Social Problems, 1961, 2, 121-132. Dibner, A.S. Cue-counting: A measure of anxiety in interviews. Journal of Consulting Psychology, 1956, 22, 475-478. Ekman, P. & Friesen, W.V. The repertoire of nonverbal behavior: Categories, origins, usage, and coding. Semiotica, 1969, i, 49-98. Ekman, P., & Friesen, W.V. Head and body cues in the judgment of emotion: A reformulation. Perceptual and Motor Skills, 1967, 33' 711-724. Festinger, L. A theory of social comparison processes. Human Relations, 1954, 1, 117-140. Foa, U.G. Convergences in the analysis of the structure of interpersonal behavior. Psychological Review, 1961, fig, 341-353. Fontes, N.E., Miller, G.R., & Kaminski, E.P. Communication barriers between handicapped and nonhandicapped inter- actiOns: Identification and removaI. Unpublished manuscript, Department of Communication, Michigan State University, 1978. 224 Frujda, N.H., & Philipszoon, E. Dimensions of recognition of expression. Journal of Abnormal and Social Psy- chology, 1963, HE; 45-51. Fromm-Reichmann, F. Psychoanalysis and psychotherapy: Selected papers of Frieda Fromm-Reichmann. D.M. Bullard (Ed.). Chicago: Ufiiversity of Chicago Press, 1959. Fujimoto, E.K. The comparative power of verbal and non- verbal symbols. Unpublished doctoral dissertation, Ohio State University, 1971. Cited in J.K. Burgoon, & T. Saine, The unspoken dialOgue: An introduction to nonverbal communication. Boston: Houghton Mifflin Company, 1978. Glaser, B., & Strauss, A. THe discovery of gpounded theory. Chicago: Aldine, 1967. Goffman, E. Stigma: Notes on the managspent of spoiled identity. Englewood Cliffs, N.J.: Prentice-Hall, 1963. Goffman, E. The presentation of self in everyday life. New York: Doubleday Press, 1959. Gouldner, A.W. The norm of reciprocity: A preliminary statement. American Sociological Review, 1960, 32, 161-178. Harrison, R.P. Beyond words: An introduction to nonverbal Communication. Englewood Cliffs, N.J.: Prentice- Hall, 1974. Hays, W.L. Statistics for_psychologists. New York: Holt, Rinehart & Winston, 1963. Heider, F. The psychology of interpersonal relations. New York: John Wiley, 1958. Heise, D.R. Causal analysis. New York: John Wiley & Sons, 1975. ' Homans, G.C. Socialybehavior: Its elementary forms. New York: Harcourt, Brace and World, Inc., 1961. Izard, C.E. The emotions and emotion constructs in person- ality and culture research. In R.B. Cattell (Ed.), Handbook of modern personality theoty. Chicago: Aldine, 1968. 225 Jones, E.E., Kanouse, D.E., Kelley, H.H., Nisbett, R.B., Valins, S., & Weiner, B. Attribution: Perceiving the causes of behavior. Morristown: General Learning Press, 1972. Jourard, S. Knowing, liking, and the 'dyadic effect' in men's self-disclosure. Merrill Palmer Quarteriy of Behavior and Development, 1960, H, 178—186. Kelley, H.H. The processes of causal attribution. American Psychologist, 1973, EH, 107-128. Kendon, A. Some observations on interactional synchrony. A preliminary report, Cornell University, 1968. Cited in M. Argyle, Social interaction. New York: Atherton Press, 1969. Keppel, G. Design and analysis: A researcher's handbook. Englewood Cliffs, N.J.: Prentice-Hall, 1973. Kleck, R. Physical stigma and task oriented interactions. Human Relations, 1969, 3a, 53-60. Kleck, R. Physical stigma and nonverbal cues emitted in face-to-face interactions. Human Relations, 1968, Hi, 19-28. Kleck, R. Emotional arousal in interactions with stigma- tized persons. Psychological Reports, 1966, $2, 1226. Kleck, R., Ono, H., & Hastorf, A. The effects of physical deviation upon face-to-face interaction. Human Relations, 1966, i2, 425-436. Krippendorff, K. Information theory. In G.J. Hanneman, & W.J. McEwen (Eds.), Communication and behavior. Reading, Massachusetts: Addison-Wesley, 1975, 351-389. Lalljee, M., & Cook, M. Uncertainty in first encounters. Journal of Personality and Social Psychology, 1973, gs, 137-141. Lampel, A.K., & Anderson, N.H. Combining visual and verbal information in an impression-formation task. Journal of Personality and Social Psychology, 1968, 2, 1-6. Longabaugh, R. The structure of interpersonal behavior. Lopez, F.M. Personnel interviewing: Theory and practice. New York: McGraw-Hill, 1975. 226 Lorr, M., & Suziedelis, A. Modes of interpersonal behavior. British Journal of Social and Clinical Psychology, 1969, H, 124-132. Mahl, G.F. Gestures and body movements in interviews. Paper presented at the Third Research in Psycho- therapy Conference, Chicago, 1966. Cited in P. Ekman, & W.V. Friesen, The repertoire of nonverbal behavior: Categories, origins, usage, and coding. Semiotica, 1969, 1, 49-98. Marinelli, R.P. State anxiety in interactions with visibly disabled persons. Rehabilitation Counseling Bulletin, 1974, lg, 72-77. Marinelli, R.P., & Kelz, J.W. Anxiety and attitudes toward visibly disabled persons. Rehabilitation Counseling Bulletin, 1973, i1, 198-205. Matarazzo, J.D., Wiens, A.N., & Saslow, G. Studies of interview speech behavior. In L. Krasner, & L.P. Ullman (Eds.), Research in behavior modification: New developments and implications. New York: Holt, Rinehart and Winston, 1965: 179—210. McCain, T.A., Repensky, G.R. The effect of camera shot on interpersonal attraotion for comedy performers. Paper presented to the Annual Convention of the Speech Com- munication Association, Chicago, 1972. Cited in J.C. McCroskey, & T.A. McCain, The measurement of inter- personal attraction. Speech Monographs, 1974, 41:3, 261—266. McCroskey, J.C., & McCain, T.A. The measurement of inter- personal attraction. Speech Monographs, 1974, 41:3, 261-266. ‘ McMahan, E.M. Nonverbal communication as a function of attribution in impression formation. Communication Monographs, 1976, 12, 287-294. Mehrabian, A. Nonverbal communication. In J.K. Cole (ED.), Nebraska symposium on motivation. Lincoln, Nebraska: University oerebraska Press, 1972a, 107-161. Mehrabian, A. Nonverbal communication. Chicago: Aldine- Atherton, 1972b. Mehrabian, A. Verbal and nonverbal interaction of strangers in a waiting situation. Journal of Experimental Research in Personality, 1971a, H, 127-138. 227 Mehrabian, A. Silent messages. Belmont, Calif.: Wads- worth, 1971b. Mehrabian, A. Methods and designs: Some referents and measures of nonverbal behavior. Behavioral Research Methods and Instrumentation, 1969, i, 203-207. Mehrabian, A., & Ferris, S.E. Inference of attitudes from nonverbal communication in two channels. Journal of Consulting Psychology, 1967, Hi, 248-252. Mehrabian, A., & Ksionzky, S. Some determiners of social interaction. Sociometry, 1972, gs, 588-609. Mehrabian, A., & Ksionzky, 8. Models for affiliative and conformity behavior. Psychological Bulletin, 1970, 13, 110-126 Mehrabian, A., & Reed, H. Some determinants of communica- tion accuracy. Psychological Bulletin, 1968, 19, 365-381. Mehrabian, A., & Wiener, M. Decoding of inconsistent com- munications. Journal of Personality and Social Psy- chology, 1967, H, 109-114. Miller, G.R., & Steinberg, M. Between people: A new analysis of interpersonal communication. Chicago: Science Research Associates, 1975. Moffatt, T.L. Selection interviewing for managers. New York: Harper & Row Publishers, 1979. Morgan, H.H., & Cogger, J.W. The interviewer's manual. New York: Drake-Beam & Associates, Inc.,—I980. Osgood, C.E. Dimensionality of the semantic space for communication via facial expressions. Scandinavian Journal of Psychology, 1966, Z, 1-30. Osgood, C.E., Suci, G.J., & Tannenbaum, P.H. The measure- ment of meaning. Urbana, Illinois: University of Illinois Press, 1957. Park, L.D. Barriers to normality for the handicapped adult in the United States. Rehabilitation Liter- ature, 1975, HQ, 108—111. Rehabilitation service manual. Department of H.E.W., Office of Human Development, Rehabilitation Services Administration, 1976. 228 Report of the comprehensive service needs study. Washing- ton, D.C.: The Urban Institute, 1975. Richardson, S.A. Attitudes and behavior toward the physically handicapped. In D. Bergsma, & A. Pulver (Eds.), Developmental disabilities: Psychologica and social implications. New York: Alan R. Liss, Inc., 1976, 15-34. Rosenfeld, H.M. Nonverbal reciprocation of approval: An experimental analysis. Journal of Experimental Social Psychology, 1967, H, 102-111. Schachter, S. Theypsychology of affiliation: Experimental studies of the sources of greggriousness. Stanford: Stanford University Press, 1959. Schachter, S., & Burdick, H. A field experiment on rumor transmission and distortion. Journal of Abnormal Social Psychology, 1955, HQ, 363-371. Scheffé, H. The analysis of variance. New York: Wiley, 1959. Scheflen, A.E. Stream and structure of communicational behavior. Commonwealth of Pennsylvania: Eastern Pennsylvania Psychiatric Institute, 1965. Cited in M. Argyle, Social interaction. New York: Atherton Press, 1969. Sermat, V., & Smyth, M. Content analysis of verbal com- munication in the development of a relationship: Conditions influencing self-disclosure. Journal of Personality and Social Psychology, 1973, gs, 332-346. Shapiro, J.G. Responsivity to facial and linguistic cues. Journal of Communication, 1968, lg, ll-l7. Siegel, S. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill, 1956. Siller, J., Chipman, A., Ferguson, L., & Vann, D.H. Attitudes of the nondisabled toward the physically disabled. New York: New York University School of Education, 1967. Siller, J., Ferguson, L., Vann, D.H., & Holland, B. Structure of attitudes toward the physically disabled: Disability factor scales - amputationy_blindL cosmetic conditions. New York: New York University School of Education, 1967. 229 Snider, J.G., & Osgood, C.E. (Eds.). Semantic differential technique. Chicago: Aldine, 1969. Spielberger, C.D. (Ed.). Anxiety and behavior. New York: Academic Press, 1966. Spielberger, C.D., Gorsuch, R.L., & Lushene, R.E. Manual for the state - trait anxiety inventory. Palo Alto, Calif.: Consulting Psychologists Press, Inc., 1970. Taylor, D.A., Altman, I., & Wheeler, L. Self-disclosure in isolated groups. Journal of Personality and Social Psychology, 1973, HQ, 39-47. Thibaut, J.W., & Kelley, H.H. The social_psychology of groups. New York: Wiley, 1959. Tomkins, S.S. Affect, imagery, consciousness (vol. 2). New York: Springer, 1963. Tomkins, S.S. Affect, imagsry, consciousness (vol. 1). New York: Springer, 1962. Tomkins, S.S., & McCarter, R. What and where are the primary affects? Some evidence for a theory. Perceptual and Motor Skills, 1964, iH, 119-159. Triandis, H.C., Vassiliou, V., & Nassiakou, M. Three cross- cultural studies of subjective culture. Journal of Personality and Social PsychologY: 1968, H, Monograph Supplement, No. 4. Vande Creek, L., & Watkins, J.T. Responses to incongruent verbal and nonverbal emotional cues. Journal of Communication, 1972, 33, 311-316. Wakshlag, J. The effect of camera angle and image size on source credibility and interpersonal attraction. Unpublished M.S. thesis, Illinois State University, 1973. Cited in J.C.-McCroskey, 7 McCain, T.A. The measurement of interpersonal attraction. Speech Monographs, 1974, 41:3, 261-266. Wiener, M., Devoe, S., Rubinow, S., & Geller, J. Nonverbal behavior and nonverbal communication. Psychological Review, 1972'.12' 185-214. Wiener, M., & Mehrabian, A. Language within language: Immediacy, a channel in verbal communication. New York: Appleton-Century-Crofts,I19681 230 Wilder, C.S. Prevalence of selected impairments, United States, 1971. U.S. National Center for Health Statistics, Vital and Health Statistics, 1975, 22, Series 10, 1-64. Wilder, C.S. Chronic conditions and limitations of activity and mobility. U.S. National Center for Health Statistics, Vital and Health Statistics, 1971, Hi, Series 10, 1-26. Williams, F., & Sundene, B. Dimensions of recognition: Visual vs. vocal expression of emotion. Audio Visual Communications Review, 1965, iH, 44-52. Worthington, M.E. Personal space as a function of the stigma effect. Environment and Behavior, 1974, H, 289-294. Worthy, M., Gary, A.L., & Kahn, G.M. Self-disclosure as an exchange process. Journal of Personality and Social Psychology, 1969, iH, 59-64. "I3111111441111!!!)1'11!!!“