‘1 lll’lll \ \lllllllll llllflll \flltl \ 1‘ l 3 1293 00826 0904 LIE-RARY Michigan State University This is to certify that the dissertation entitled Effects of Preinteraction Expectancies on Subsequent Impressions of the Evaluative, Potency, and Activity Dimensions presented by Richard Marshall Garlick has been accepted towards fulfillment of the requirements for Ph.D. degree in Communication onm/M E 92%! Major professor Date November 11, 1987 MSUi: an Affirman'vt Action/Equal Opportunity Institution 0- 12771 MSU LlBRARlES ._3_. RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. I51“ .ilgN 'l ‘1 b a W EFFECTS OF PREINTERACTION EXPECTANCIES ON SUBSEQUENT IMPRESSIONS OF THE EVALUATIVE, POTENCY, AND ACTIVITY DIMENSIONS BY Richard Marshall Garlick A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication 1987 ABSTRACT EFFECTS OF PREINTERACTION EXPECTANCIES ON SUBSEQUENT IMPRESSIONS OF THE EVALUATIVE, POTENCY, AND ACTIVITY DIMENSIONS BY Richard Marshall Garlick This research examines the role of pre-interaction expectancies in three areas of person assessment: evaluation, activity, and potency. Two competing theories of person assessment are tested through these studies. The first is an averaging type of information integration model which states that initial expectations are integrated into later impressions, maintaining their initial relative weights. A second model is one which states that individuals find information which contradicts an initial expectation to be particularly informative. In this situation, there would be a significant interaction effect between the initial information and subsequent behavior. In order to test these models, participants were given initial information about a confederate describing him or her one of six ways: pleasant, unpleasant, strong, weak, active, or passive. In some conditions, participants received no initial information about the confederate. The participant then met the confederate and was told that he Richard Marshall Garlick or she would be interviewing the confederate through video monitors. Actually, the participant viewed prerecorded answers to a series of scripted questions. The responses were designed to confirm or disconfirm participants' initial expectations. Results showed that a weighted averaging model of information integration best fit the data for the evaluative and potency dimensions, while participants essentially disregarded initial information and based judgments on behavioral information for the activity dimension. ACKNOWLEDGEMENTS While there have been many individuals who have played an important role in my education and the completion of this dissertation project, there are a number of people who deserve special recognition. The few words that I can express toward each of them here cannot begin to scratch the surface in expressing my deepest gratitude for the help and support which they have provided in helping me achieve what, at one time, I had only dreamed of accomplishing. First of all, I would like to thank the members of my committee for their guidance. It has been a rare privilege to study under such a distinguished scholar as Dr. Gerald R. Miller. I am particularly grateful for his help and support as my advisor. I am also grateful to Dr. James Stiff for his helpful suggestions and support of my project and to Dr. Kathy Kellerman and Dr. Norbert Kerr for challenging me to pursue scholarship at such a high level. I am especially appreciative to Dr. Gordon Dahnke, without whose influence I probably never would have been here in the first place. Dr. Dahnke has served as a fine mentor to me in studying the field of communication. This project could also not have been completed without the help of a number of undergraduate research assistants iv who aided me throughout the course of the project. While they are too numerous to mention, I would like to especially thank Tom Hammang, Susan Gundry and Beth Henningsen for serving as confederates in the study. This role required a great deal of time and commitment over the course of several terms. Special thanks also go to Curtis Myers, whose technical assistance was invaluable. Curtis' help in this project went far above the normal call of duty. Another individual who served as a confederate in the study was my good friend, Roger Klahn, Roger received no reward for his work other than the reward of helping out a friend in need. He is one of the greatest friends an individual ever could hOpe to have. Special thanks also go to my many other friends who have supported me through this intense period of my life. I would like to mention every one of them by name if I had the space in which to do it. Each of them holds a special place in my heart. I consider myself to be the luckiest man in the world to be blessed with so many supportive friends. Finally, I dedicate this dissertation project to some very special people. Words can never express the gratitude that I have toward my parents, Kenneth and Bernice Garlick. I am proud to have them as my Mom and Dad and hope that I can always reflect the values and standards they have taught me in all I do. I also dedicate this dissertation to the memory of my grandma, Mrs. Edith Knight. I had really wanted her to see me get this degree and have missed her greatly. I thank God for the opportunities He has blessed me with. I also thank Him for the strength and endurance which I have so greatly needed (Phillippians 4:13). vi TABLE OF CONTENTS LIST OF TABLES . . . . LIST OF FIGURES . . . CHAPTER 1 REVIEW OF LITERATURE AND RATIONALE The Role of Person Assessment , , , , , _ Approaching the Issues from a COgnitive Perspective Trait Relationships in Dimensions of Person Perception , , . The Differential Impact of Negative Information in Forming Impressions , The Role of Expectations and Expectation Violation in Impression Formation , , The Impact of Inconsistent or Contradictory Information Rationale of the Current Study , 2 METHOD . . . . Overview , Subjects , Design . . on Impression Formation , Dependent Measures . . . Independent Procedure , 3 RESULTS . . . . Results for Results for Results for Summary . . 4 DISCUSSION . . APPENDIX A: APPENDIX B: Dependent Questions Variable Manipulations , the Evaluative Dimension the Activity Dimension , the Potency Dimension , , Measure Scales . . . . . and Scripts . . . . . . . vii ix 20 23 26 33 39 39 40 4O 41 42 51 55 55 62 67 73 75 91 94 APPENDIX C: Expectancy Manipulations . . . . . . . . 101 APPENDIX D: Scale Measuring Amount of Attention Given to Information . . . . . . . . . . 107 LIST OF REFERENCES 0 O O O O O O O O O O O C O O O O O 108 viii 10. ll. 12. 13. 14. 15. LIST OF TABLES Predicted Cell Means Using Various Weights for Informational Integration . . . . . . . . . The t- tests for Equidistance of Absolute Values of Ratings for Response Dimensions . . . . . . . The t- tests for the Equidistance of Absolute Values of Expectancy Manipulations . . . . . . The t- tests for Equidistance Between Expectancy Manipulations and Behavioral Manipulations . . . The t- -tests Comparing Means Measuring Amount of Attention Paid to Manipulations Against a Value of 0 . . . . . . . . . . . . . . . . . . . . . . The t- test Comparing Means for One' 5 Own Initial Impression to a Value of O . . . . . . . . . . . Analysis of Variance for the Evaluative Dimension . . . . . . . . . . . . . . . . . . . Ratings for the Evaluative Dimension . . . . . . Results for the Least Significant Difference (LSD) Contrasts for the Evaluative Dimension . . Results for Scheffe and Newman-Kuels Contrasts for the Evaluative Dimension . . . . . . . . . . A priori Contrasts for the Evaluative Dimension Analysis of Variance for the Activity Dimension. . . . . . . . . . . . . . . . . . . . Ratings for the Activity Dimension . . . . . . . A priori Contrasts for the Activity Dimension . Analysis of Variance for the Potency Dimension. . . . . . . . . . . . . . . . . . . . ix 34 45 46 47 49 51 56 57 59 60 62 64 65 66 68 16. 17. 18. Results for Newman-Kuels Contrasts for the Potency Dimension . . . . . . . . . . . . . . . 69 Ratings for the Potency Dimension . . . . . . . 71 A priori Contrasts for the Potency Dimension . . . . . . . . . . . . . . . . . . . 73 LIST OF FIGURES 1. Graphic representation of information integration model . . . . . . . . . . . . . . . . . 36 2. Graphic representation of interaction model . . . . 37 xi Chapter 1 Review of Literature and Rationale Information about individuals gained through friendship and social networks often has a great impact on subsequent interactions. If a person is described in a particular way, it can create expectations that may or may not be accurate. If a person is described positively, a "halo" effect can be created-—a negative description can cause more critical subsequent evaluation and even avoidance of interaction. Furthermore, the simple description of an individual as "friendly" or "warm" can activate a whole series of impressions concerning related personality traits. Others expect individuals to behave in ways that are consistent with qualities that a "friendly" or "warm" person should have. When actual behavior is incongruent with previous information, some kind of cognitive reappraisal is often needed to resolve the inconsistency. This dissertation will examine the various aspects of person assessment when prior expectancies are created by the communication of trait descriptive information. 2 The Role of Trait Relationships in Person Assessment Individuals often use trait terms to describe others. A trait term such as "warm," "cold," "strong" or "weak" carries a great deal of meaning for the individual perceiver. These trait terms create a series of related mental images which are organized along several dimensions. Three dimensions of person assessment that have been identified by Osgood, Suci, and Tannenbaum (1957) are the evaluative, potency and activity dimensions. The evaluative dimension refers to an assessment of liking or disliking. Persons make an early determination as to whether or not they wish to establish a relationship with that person. The potency dimension refers to a perception of strength or weakness. People make assessments as to the power that other people display. The third dimension refers to the assessments that individuals make concerning another's relative activity or passivity. As traits are arrayed along the various dimensions, they are clustered together in relationships with one another. Much research has examined how trait descriptions and behaviors form prototypes. Cantor and Mischel (1977a) have suggested that there are certain personality categories called prototypes that serve to organize impressions. A prototype such as extraversion consists of a number of related traits and behaviors. The prototype category of extraversion contains traits related to the evaluative dimension such as friendly; traits related to the activity 3 dimension such as talkative and outgoing; and traits related to the potency dimension such as loud. Behaviors related to this category might include liking to go to parties, having lots of friends and being involved in a number of activities. These traits and behaviors are all held together as part of the stereotype or prototype of the category of extraversion. Cantor and Mischel (1977a) studied the implications of this approach. One of the implications is that prototypes ought to aid in the memory of more specific information. In one study, subjects were shown a number of traits characteristic of a person. Some of these traits fit a prototype, while others were irrelevant to it. Later, when subjects were asked if they had seen certain traits in the list, they reported with a great deal of certainty that they had seen the traits to which they had been exposed. However, subjects also reported that they had seen a number of traits to which they had not been exposed, particularly when the traits fit the prototype. Other research has demonstrated that people are likely to report they have seen instances of behavior that are consistent with material previously presented (e.g., W011 and YOpp, 1978). In a follow—up study, Cantor and Mischel (1977b) showed subjects given behaviors representing a person. Some stimulus persons were described by several behaviors consistently related to a particular prototype while the other stimulus persons were described in an inconsistent 4 way. When subjects were asked to recall behaviors later, they remembered more behaviors when the behaviors fit the prototype than when they did not. There was also a tendency for written impressions to contain more information for the consistent than for the inconsistent stimulus person. This research suggests that individuals do not simply store concrete pieces of information about another, but rather, store a prototypic representation. Traits are not merely verbal labels, but are parts of a highly integrated set of other traits and behaviors that are organized at various levels of abstractness. Schneider and Blankmeyer (1978) contend that traits and behaviors related to a particular prototype are seen as being more closely related when the prototype is salient than when it is not. In their research, when the stimulus person was described as being extraverted, extraverted traits and behaviors implied each other more than when the stimulus person was described as introverted or not described at all. Introverted traits were seen as more closely related when the person was described as being introverted. The research on prototypes provides strong support for the idea that individuals cluster various traits together and see these traits as being related to one another. The early work of Asch (1946) was among the first to examine trait relationships. Asch tested two competing models of evaluative impression formation: (1) an elemental or 5 additive model versus (2) a configural model. The configural model was based upon holistic Gestalt principles. In this instance, the entire configuration of a person's attributes would determine a "root of personality." There would be a perceived unity of general impression based upon perceived relationships of traits to one another. The competing elemental model assumes that evaluations represent the sum of isolated attributes. In this case, the model posits that each attribute is individually and sequentially evaluated and then combined into a summary judgment. Asch developed the elemental model as a foil to the configural model, clearly preferring the configural model. The configural model implies that each trait affects other traits such that the final impression is dynamic and cannot be successfully predicted from each of the individual traits taken separately. Asch's experimental findings supported the notion of a unified impression of personality. Asch also suggested that not all character qualities of an individual carry the same importance in establishing the impression of a person. Some characteristics can be thought of as primary or central qualities, while others are more peripheral or secondary to impression formation. Asch believed that context affected the degree to which a trait was central or peripheral. For example, Asch originally placed the warm-cold variable in a context with the stimulus traits: intelligent-skillful-industrious- determined-practical-cautious. Upon further investigation, 6 Asch found that warm-cold was less central when placed in a list with obedient-weak-shallow-unambitious-vain: subjects responded in a way that indicated the inclusion of warm or cold in the list had little effect on their final impressions. In order to examine the perceived relationships among personality traits further, Wishner (1960) performed a reanalysis of the original Asch data. It was his reasoning that if certain stimulus traits were central, a strong correlation should exist between these stimulus traits and other response traits. From his results, Wishner concluded that if correlational relationships are known among traits, traits that people infer can be predicted. Traits are central to the extent that they are highly correlated with response traits. The most significant feature in this type of analysis is the notion that people have a clear idea of what traits should be expected if other traits are present, e.g., warm persons are generally expected to be kind, considerate, and generous. Thus, knowing one trait or a set of traits provides information about a person more than simply the set of traits; individuals relate given traits to clusters of relevant (nonprovided) traits. The literature examining this area demonstrates that trait terms carry a great deal of meaning for individuals. However, it is reasonable to assume that perceptions of other peOple are at least partially determined by perceiver 7 variables. If this is the case, then it can be assumed that perceptions of the "way someone is" are often shaped by the perceiver's personality. Kelly (1955) warned of potential methodological problems in studies of trait perception, simply because concepts related to trait terms are hard to pin down precisely and tend to be somewhat individualistic. Rosenberg (Rosenberg & Sedlak, 1972; Rosenberg, 1977) has done intensive analysis of individual perceivers. Rosenberg came to believe that a preprogrammed list of traits may yield less satisfactory structures than letting subjects think of their own traits. Rosenberg allowed his subjects to think of 100 names of people they knew and asked them to generate a list of traits and feelings elicited by each. Rosenberg's results demonstrated that subjects typically provided around 100 traits and 50 feelings. Each trait or feeling was then rated for applicability to each stimulus person used in the study. Rosenberg's work, employing both multidimensional scaling and clustering techniques, has allowed researchers to identify interesting clusters which provide insights into individual perception processes. Approaching the Issues From a Cognitive Perspective The processes individuals use to cluster traits and store person information is an important issue to examine. Individuals organize information in a particular manner. An information processing paradigm may be employed to help understand the perceived relationships among traits. The 8 information processing paradigm centers around the notion that the mind works similarly to a computer. An information processing view states that people collect, store, modify and interpret environmental information which has already been stored internally. Symbols which are processed may already contain meaning, based upon previously stored and processed cognitions. Collections of these symbols are processed and categorized in order to form schemas. The schema theory of Sir Fredric Bartlett (1932) represents one of the earliest efforts to explicate the relations between an abstract structure and the specific recall of facts. Bartlett introduced the idea that categorization of input information tends to introduce bias into the reconstruction process through the assimilation of input information into an activated category. Bartlett also suggested that the delayed influence of categorization on subsequent judgments may be greater than the immediate influence, the stored details of the input information are more rapidly forgotten than the categorization. A schema is a cognitive structure which contains knowledge about the attibutes of a category and the links among those attributes (Rumelhart & Ortony, 1977). A schema contains the features typical of any category of person, object or event. Fiske and Pavelchak (in press) comment that recent cognitively oriented social research has demonstrated that perception, memory and inference are guided by categorization and by social schemata in ways 9 that simplify information processing. Fiske and Pavelchak continue to support the notion that cognitive functions of categorization and schemata are consistent with the concept of a configural model. .Finally, Fiske and Pavelchak argue that while most studies performed in this area have ignored the role of affect, affect is the reason that social schemata matter so critically. If for example, someone studies the reasons why categorized individuals are perceived and remembered in biased ways, it is also important to consider the evaluative aspects of those biases. In Fiske and Pavelchak's view, people's reactions to a new person consist of an initial categorization stage in which people categorize according to a previously held schema and a second generation affect stage. As the schemata which is stored in memory is invoked, the social categorization process takes place. The schemata may include a set of cognitions about certain groups, personality types, trait relationships, etc. Fiske and Pavelchak (in press) hypothesize the schema to consist of a category label at the top level and expected traits or attributes at a lower level. The label and attributes can be thought of as nodes in a memory network. According to this View, memory retrieval operates by activation, spreading along the links from one node to another. Within this schema structure, the attributes are presumed to have strong links to the category label and to have weaker links to each other. In this sense, a category 10 label will have more impact than any single trait, because the label has more and stronger links to the attributes than they do to each other. This model assumes that recall of category consistent attributes is facilitated by the presence of a category label. Social learning plays an important part in establishing categories and category labels. Individuals learn to respond to certain types of persons in specific ways. People learn to react negatively to individuals who are dark, surly and shifty-eyed. At the same time, individuals respond positively to males who are tall, blond and blue eyed. People are taught early that large, round people are happy and jolly (e.g. Santa Claus). All of these descriptions fit certain schemata which contain category- based affective tags. It is no wonder that people like or dislike a certain person before they get to know them. These schemata may be triggered by various types of stimulus information. For example, a person's appearance, vocal qualities, behavioral manner or trait descriptions may activate a schema category. Each attribute a person possesses has an associated evaluative tag. A critical issue, however, is the type and nature of cognitive processing which takes place. It is not always a sequential process, but rather, may be a global process in which cognitions are categorized rapidly and associated with previously learned affective tags. Under this theory, an individual might have some pre-existing basis for making ‘ 11 an evaluative response. This affective reaction may take place at a very low level of cognitive awareness. Often a person will reflect later about why he or she experienced a certain evaluative response toward a person. Many times an individual is able to gain a conscious awareness later of cognitions previously processed at a low level of awareness. An individual's initial reaction will be based upon global coqnitive processing of a schema or category of characteristics. Individual qualities or characteristics of a person or object will often be analyzed later. However, the strength of the link between character traits, as well as the link to the overall category label will be significant in the process of impression formation. Bruner (1958) suggested that a fundamental process of person perception is to connect the input with some stored category. The readiness with which a person classifies information into a particular category is an indication of the accessibility of that category. If a category has been established for a collective group of traits, exposure to an experimenter's stimulus (i.e. the name of a trait category) should increase the likelihood that subjects will categorize the stimulus person in terms of that activated category. The act of categorization will, in turn, affect how the stimulus information is processed. Bruner cites an example of an average sized Black who is sitting on a park bench during his lunch hour. The Black is categorized as "lazy" by an observer and is later 12 remembered as being a big, healthy Black sprawling idly in the park doing nothing all day. This example provides evidence for the fact that once a category is activated, the category becomes an integral part of the memory process itself. Information processing models would suggest that a subject's judgment of a stimulus person will depend on the sample of information about the person that is most available to the subject for retrieval at the time his or her judgment is made (Wyer, 1974). If a subject has previously categorized a stimulus person, this categorization could affect judgments of this person through its effect on the construction and reconstruction of the stimulus information and through the category's own evaluative implications. Higgins, Rholes and Jones (1977) state that one can interpret the effects of an experimenter's description of a stimulus on subjects' judgments and recollections of the stimulus as being mediated by the types of information processing mechanisms previously described. Higgins et a1. tested the relative accessibility of different trait categories by manipulating exposure of subjects to trait terms which were unobtrusively embedded in a previously "unrelated" task. The idea was that exposure to a trait term should activate its trait category meaning, and this meaning will further activate closely related trait categories. Higgins ea: al.'s first hypothesis was that subjects would categorize an ambiguous stimulus person using 13 whichever categories had been activated or primed. Higgins et al.'s thinking was that effects of categorizations would be reflected in subjects' later characterizations and evaluations of the stimulus person, as well as their reproduction of information. Higgins et a1. predicted that the subjects would evaluate the stimulus person more favorably when the trait terms to which they were exposed had favorable referents as Opposed to unfavorable referents. The first assumption was that the effects of prior exposure to trait terms on subjects' subsequent evaluations of the stimulus person would be mediated by the categorization process. To test this assumption, the experimenters looked comparatively at the effects of exposing subjects to trait terms which were applicable for characterizing the stimulus person, along with the effects of exposing subjects to equally favorable or unfavorable trait terms which were not applicable for characterizing the stimulus person. The researchers also decided to test several of Bartlett's (1932) ideas concerning the delayed influence of categorization on subsequent judgments. In Higgins an: al.'s study, subjects were first administered a color naming task that required them to remember four trait names and six names of inanimate objects. In each of two conditions, considered to be applicable priming conditions, the trait terms could be used to describe the same set of behaviors. However, the terms in one condition were all favorable (adventurous, 14 self-confident, independent and persistent), while in the other condition they were all unfavorable (reckless, conceited, aloof and stubborn). In two other conditions, considered inapplicable priming conditions, the four trait terms also had favorable or unfavorable implications, but were inappropriate for describing the set of behaviors. Following exposure to one set of trait terms, subjects were asked to take part in an "unrelated" study of reading comprehension. They were asked to read a paragraph about a boy named Donald, whose behavior could be readily encoded using the trait terms in either of the applicable priming sets. After reading a paragraph describing Donald using the various trait terms, half of the subjects in each priming condition characterized Donald's behavior in their own words, while the remaining judges did not. Results of the Higgins et a1. study demonstrated that subjects who wrote descriptions of Donald typically described him using trait names that were similar to the priming words they were originally given. When trait categories were applicable to the stimulus person, a priming effect took place which produced an evaluative bias in an expected direction. When the priming words were not applicable, the evaluative implications were, if anything, opposite in direction to those of the priming words. The experimenters found that the overall evaluation of Donald was more positive when primed with applicable priming words, 15 but was minimally affected by the priming words when they were inapplicable. The findings by Higgins, Rholes and Jones (1977) suggest that the effect of information on one's attitudes and beliefs may be influenced substantially by prior experiences which increase the accessibility of a set of concepts for use in interpreting subsequent information. Wyer and Srull (1981) explore these concepts further and form, what they term, a "storage bin" model of category activation. Wyer and Srull (1981) claim that the priming task does more than make particular words accessible. Wyer and Srull argue that, in addition, it increases the accessibility of more general concepts or schemata to which the words refer. Thus, when the term "honest" is primed, the category of "trustworthy" is also primed since it is closely related and denotes the same general schema. Wyer and Srull cite evidence of this from the Higgins «at al. (1977) study which demonstrated that when subjects were asked to describe the target person in their own words, they used trait terms that were identical to the primed terms in less than half of the cases. The majority of descriptive adjectives that were used were either synonyms or phrases with similar semantic implications to the priming words. Wyer and Srull (1981) go on to state that in addition to trait terms, schemata contain certain prototypic behaviors which exemplify the trait. The similarity of 16 behaviors to those described in the new information would determine the extent to which the information is encoded in terms of that trait concept. Also, if the behaviors are part of the schema, they should have priming capabilities as well. As trait schemas are activated in which certain behaviors are contained, the increased accessibility of the schema may lead other more ambiguous behaviors to be interpreted in light of that trait schema. It can be concluded that the effect of priming relevant behaviors on the interpretation of other traits may increase with the different number of behaviors that are primed, as well as the likelihood that the schema associated with the trait is activated. There may be substantial differences with the different traits and schema in relation to the amount of priming required to produce a given effect. That is, some information may be seen as more salient than other information. Some studies (Wyer & Hinkle, 1976; Birnbaum, 1974) have demonstrated that unfavorable information about persons has more influence on evaluations than does favorable information. To a certain extent, a greater number of behavioral instances may be required to activate schema associated with favorable traits than those associated with unfavorable traits. Dimensions of Person Perception One point of criticism which has been raised in studies of trait relationships is concerned with the 17 semantic organization involved in such studies. It can be argued that the nature of trait dimensions themselves introduces some bias in perceived trait relationships. The dimensional structures which are considered in most studies of this nature are systems cu? interrelationships among words (i.e. the names of traits). It can be argued that when people make personality ratings of others, perhaps they are not thinking so much about people at all, but rather, make judgments on which words seem to go together. Some of the dimensions that emerge in subjects' ratings of people are very similar to the dimensions that emerge in research studies of the general meanings of words. Almost all studies find at least an evaluative dimension at the beginning of the assessment process. Several studies have demonstrated (Levy & Dugan, 1960; Mulaik, 1964; D'Andrade, 1965) that one of the first judgments people make is of an evaluative nature. They make a determination of whether or not they like a person or whether they are good or bad. Empirically, an evaluative dimension nearly always accounts for a large share of the rating variance of both objects and peOple (Osgood, Suci & Tannenbaum, 1957; Warr & Knapper, 1968; Frijda, 1969). It can be demonstrated that the evaluative dimension has great relevance for our behavior toward a stimulus person. For example, a study by Kelley (1950) demonstrated that students were more likely to interact with an instructor they perceived to be "warm" as opposed to when they perceived the instructor to be "cold." 18 Nisbett and Wilson (1977) performed a study in which subjects saw an interview with a professor who was either warm and friendly or cold and hostile. Not surprisingly, the subjects rated the warm professor more favorably on related personality characteristics such as friendliness and approachability. Interestingly, they also rated the warm professor more highly on pleasantness of appearance. Since warmth or coldness had nothing to do with appearance, the conclusion was the overall evaluative reactions also had an impact on perceptions of appearance. While the evaluative dimension is important, Osgood, Suci & Tannenbaum (1957) also identified two other dimensions of person assessment: activity and potency. These trait dimensions have been found in ratings of personality traits (Rosenberg & Olshan, 1970; Rosenberg, Nelson, & Vivekananthan, 1968). One area where activity has been demonstrated to be an extremely important dimension in the person perception process is in the area of emergent leadership in small group contexts. The first factor in eliminating leadership contenders appears to be participation. People who do not contribute to the discussion are quickly excluded as leadership possibilities. This participation factor was first discovered by Riecken (1958). Other studies (Morris & Hackman, 1969; Willard & Strodtbeck, 1972; Lucas & Jaffee, 1969; Knutson & Holdridge, 1975) found that while group leaders were not necessarily the most active group members, they were always 19 among the most frequent participants. These results suggest that if someone were aspiring to the leadership position, then they would be well advised to participate actively as soon as the discussion begins. The potency dimension has significance in the area of leadership perception, as well. An early study by Goodenough (1930) demonstrated that leaders tended to be individuals generally perceived as being more dominant. Perceptions of potency have been demonstrated to play an important role in the attributions made by others. Thibaut and Riecken (1955) ran two experiments 1J1 which the subject asked both a high-power and a low-power confederate to comply with a reasonable request. While both complied, the locus of causality was seen as internal for the high-power confederates and external for the low-power compliers. There was more correspondence between behavior and attributed intent for the high power confederates; they were perceived as acting more from feelings of good will rather than from external pressures. While most studies of impression formation have dealt primarily with the processes in making evaluative judgments, a study by Hamilton and Huffman (1971) examined the generality of impression formation judgments for all three response dimensions. The data analysis demonstrated that the results for the evaluative and potency dimensions were highly similar, suggesting that the same processes underlie these two kinds of judgments, whereas the results for the 20 activity dimension were less consistent. This would suggest that perhaps we employ different processes in making activity assessments. Hamilton and Huffman suggest that people more typically make evaluative and potency judgments in their initial encounters with others, but that the activity dimension is only of secondary importance in forming impressions. Hamilton and Huffman further suggest that when we first meet a person, we quickly gain a sense of whether or not we like him or her and whether or not the person holds power over us. Thus, initial judgments of evaluation and potency in the first impression may be of considerable importance if our future interactions with him or her are to be successful. Hamilton and Huffman conclude by stating that since activity judgments are considered to be of secondary importance, there is much greater variance in the processes which we typically use to make them. The Differential Impact of Negative Information in Forming Impressions As mentioned earlier, not all information is given equal weight in forming impressions of an individual. A large body of literature has focused on the prepotency of negative information in impression formation (Anderson, 1965; Birnbaum, 1972; Cusumano & Richey, 1970; Feldman, 1966; Hamilton & Huffman, 1971; Hamilton & Zanna, 1972; Hodges, 1974; Jordan, 1965; Levin & Schmidt, 1969; Miller & Rowe, 1967; Richey & Dwyer, 1970; Richey, 21 McClelland & Shumkunas, 1967; Warr & Jackson, 1970; Wyer, 1970). This research consistently demonstrates that when comparable positive and negative information is presented, the negative information has a disproportionate influence. One hypothesis which has been advanced to explain this negativity effect is the extremity hypothesis (Fiske, 1980). The extremity hypothesis states that attributes are considered to be informative insofar as their evaluations represent an extreme distance from a psychological neutral point. Evaluative valences (positive or negative) represent properties which define an attribute's informativeness for impression formation. This extremity hypothesis supports the position that extremely positive as well as extremely negative information should be weighed more heavily in evaluations. Fiske contends that the negativity effect is really more of an informativeness effect wherein extreme and negative events are the determinants of informativeness. Kellerman (1984) applied this extremity/informativeness hypothesis to initial interactions and found that both positivity and negativity effects were possible in social interactions due to skews in the underlying distributions of social actors' expectations about themselves and others. The main factor as to whether a negativity or positivity effect occurred appeared to be the degree to which information received was considered informative about the individual. 22 An underlying presumption for the existence of a negativity effect is related to bias. Research has demonstrated that even when the actual probabilities for pleasant and unpleasant events are identical, pleasant events have been judged to be more likely to occur than unpleasant events (Irwin, 1951; Marks, 1953). Likewise, positive interpersonal relationships are viewed to be more likely to occur than negative ones (DeSoto & Keuthe, 1959) and people tend to make positive evaluations of other people (Frauenfelder, 1974; Kleinke, Bustos, Meeker, & Staneski, 1973; Mettee, 1971a, 1971b; Sears & Whitney, 1972). This research would lend support to the idea that individuals often perceive of themselves as existing in a world of positive expectations (Katz, Gulek, Kahn, & Burton, 1975; Lau, 1980). This would seem particularly true in the area of initial interactions. In our culture, most individuals expect that others will seek to "put their best foot forward" in an initial interaction. It is usually expected that an individual will seek to make a positive first impression. In light of this, if someone were rude during an initial encounter, it would seem more surprising than if the individual were consistent with the cultural expectations of being courteous, friendly, and polite. This type of negative violation of a postive expectation would add to the informativeness and salience of the individual's negative behavior in forming an impression. 23 The Role of Expectations and Expectation Violation in Impression Formation In examining how individuals process information about others, it is significant to study the role of prior expectations and how they impact upon subsequent interactions. In an early study by Kelley (1950) subjects were given identical descriptions of a guest instructor except for the fact that one description mentioned that the instructor was "cold" while the other described him as "warm." It was found that this warm—cold variable played an important role in forming subjects' impressions. Subjects receiving the "warm" description rated the stimulus person as more considerate, informal, sociable, popular, humorous, and humane, as well as better natured. The findings also demonstrated that the frequency with which students took part in the discussion led by the guest instructor was also affected by the warm—cold manipulation. A larger prOportion of those given the "warm" preinformation participated in the discussion compared to those receiving the "cold" preinformation. Fifty-six percent of the "warm" students participated in the discussion, while only 32 percent of the “cold" students interacted with the instructor. Kelly concluded that the initial impression of favorability or unfavorability will serve to either facilitate or inhibit social interaction. In a similar study, Bond (1972) told students that they were going to have a conversation with a woman whom 24 they believed would be either cold and distant or warm and outgoing. The various women who protrayed the stimulus woman were given no particular instructions on how to behave and did not know whether the subjects thought she was cold or warm. Before interacting with the woman, the "warm" subjects had very different impressions of her than did the "cold" subjects. After the interactions, the ratings were much closer together, although the two groups were still significantly influenced by their initial impressions. That is, the "warm" subjects thought that the woman was a little less warm than they had previously believed, while the "cold" subjects thought that the woman was considerably less cold than they had previously believed. In this case, subjects were quite sensitive to the difference between the woman's behavior and their own expectations. In this study, it was also noted that the women behaved differently to the two subject groups. Since they didn't know which subjects were given "warm” preinformation and which subjects were given "cold" preinformation, their behavior must have been influenced by differences in the subjects' behavior. They behaved more warmly to the "cold" subjects than to the "warm" subjects as though they had guessed the subjects' initial impression and were making an extra effort to compensate. Several other studies supported the idea that perceivers do influence others to behave according to the perceiver's stereotype (Snyder, 25 Tanke, & Berscheid, 1977; Swann & Snyder, 1980; Ickes, Patterson, Rajecki & Tanford, 1982). Once an initial expectation is set, individuals form a set of normative expectations concerning behavior. Schneider, Hastorf and Ellsworth (1979) state: It is when the norm is broken that our attention is engaged. If we signal that we're ready for the other person to speak but the other person continues to gaze at us with the calm attention of a listener, or if we gesticulate to show that we're in the middle of a thought and want to continue but the other person begins on another topic, some of our attention may switch from the topic of conversation to the mechanics of even to the nature of our companion. If the conversation resumes on an orderly course with no further disturbances, we may think no more of it. But if these tiny offenses are repeated, we may end up with the impression that the listener, who fails to take an offered turn is a cold, dull, unresponsive person who is contemptuous of what we are saying while we may perceive the interrupter as being inconsiderate or overbearing. If the norms are broken consistently and regularly so that even the basic patterns of normal social behavior are violated, we may begin to suspect that our companiOn is not right in the head. (pg. 137) Schneider, Hastorf and Ellsworth go on to state that the clarity and awareness of the inferences which are made depend on the extremity and suddenness of the departure from baseline norms. If a person makes a series of minor violations from the norms which extend over the course of a conversation, the receiver may not consciously notice them or make an immediate attribution, but may feel a vague sense of annoyance. Schneider et a1. conclude by stating that we notice and interpret departures from norms of social desirability and from common statistical norms as meaningful. As long as our companion's behaviors run 26 "true to form," we probably don't think much about it. Whether we consider the behavior to be purposive or reactive, departures from our expectations cause us to stand up and take notice. The Impact of Inconsistent or Contradictogy Information on Impression Formation The role of prior expectanciescniour subsequent reactions to others makes the effects of an initial impression on perception and memory an important issue to address. It is interesting to examine how our initial impression of a person causes us to treat unexpected events, behaviors and information that are incongruent with our impression differently from impression-congruent events. It is significant to look at whether attention and encoding processes focus on incongruent information or avoid it and to examine what happens when individuals attempt to retrieve impression-congruent and impression-incongruent events from memory. In other words, which plays a bigger part in forming an impression: information which is consistent with the final impression or information which contradicts it? In organizing this type of information, problems generally occur when inconsistencies arise. That is, when an individual is described or perceived in both positive and negative terms, some resolution of the inconsistencies needs to occur. Concern with dealing with this type of inconsistent information led to the development of the 27 linear combination model (Anderson, 1965). In this model, the overall impression is seen as an additive combination of the properties of the stimuli. This linear combination hypothesis would suggest that information contained in the traits is combined by adding or averaging the value of each trait. Linear combination models concern themselves with responses to trait information that can be captured on a single dimension. The main assumption behind these models is that the response along the dimension is made by adding or averaging relevant information from the various traits. Evaluations of a person are often based upon the sum or average of evaluations made on individual traits. This model does not assume that relationships among traits impact on the final impression. As mentioned previously, the two simplest linear combination models are the summation and averaging models. While both models have an additive basis, they can imply different things. The summation model is based upon the premise that responses are dictated by the total favorability of the stimuli. For example, a person may like an individual who is considerate or a person who is generous. In this case, it would be logical to assume that such a person would really like an individual who would possess both qualities. In the same way, a person may dislike people who are stingy and rude and would doubly dislike people who are both stingy and rude. 28 The averaging model conceptualizes the impression formation process differently. As an example to illustrate this model, there may be times when a person would prefer to get one or two expensive gifts at Christmas (with a high average cost) over a greater number of less expensive gifts, even though the sum total of the less expensive gifts may exceed the total cost of the fewer more expensive gifts. Applying this to the judgment of traits would lead to the conclusion that if a person possesses a few strong traits, adding weakly favorable information would hurt the overall impression, rather than help it. Anderson (1965) did some research to test these various linear combination models. Anderson did a systematic study in which he varied the quality of different trait stimuli. He also varied the number and mix of the stimuli. A subject got either two or four traits as stimuli and there were four levels of favorability of stimuli. Negative stimuli were designated L for low, moderately negative stimuli were designated M-, moderately positive stimuli were designated M+, and positive stimuli were H for high. The two or four stimuli presented to subjects were either uniformly of one type or varied. According to the averaging model, adding moderately positive stimuli to highly positive stimuli will lower the final evaluation because the average of the set will decrease. The summation model, on the other hand, would predict a more positive evaluation for the larger set because of the 29 greater amount of information provided. The results tended to be mixed, as some data supported an averaging model, while other data did not. For example, evaluations of an HHM+M set are lower than those of an HH set. However, an HHHH stimulus person is evaluated more highly than an HH person, a result not always consistent with a straight averaging model. Anderson suggested a weighted averaging model to resolve inconsistencies. He added two additional factors: trait weights and an initial impression. The argument would be that weights for traits increase as trait information becomes more central or important, or as the person pays more attention to the trait. Since initial impressions are not always neutral, differences exist in terms of a "baseline" impression. One of the predictions of Anderson's model is that as more information is added, the "weight" of the initial impression shOuld lessen. Kaplan (1972) predicted that the effects of an initial impression should decrease as the number of stimulus traits increase and that differences between subjects with initially positive and negative dispositions would be lessened with larger sets of stimulus traits. Kaplan also predicted that if the weights of the new traits are somehow reduced, the weights of the initial impressions should somehow increase. Kaplan (1971) varied trait weights by telling subjects that traits were provided by either a highly valued or a low valued source. Results 30 supported his prediction that differences were greatest between positive and negative disposition subjects when the traits were given low weights due to assignment to a low valued source. These results are consistent with the idea that without any information, peOple have the predispositions to evaluate others positively or negatively and that these predispositions are given less virtue of credibility or amount. As new information is added, the "weight" given to the initial information became significantly lessened until it no longer carried any "weight." The implication is that given a large amount of information which contradicts an initial impression, this more recent information will supercede the impact of an initial impression. In further considering whether impression-congruent information or impression-incongruent information would be considered most relevant and best recalled regarding impression formation, it would seem that a great deal of the literature supports the notion that information which is consistent with a previous impression would best be recalled. In fact, simple versions of schema theories (Bartlett, 1932; Bransford & Franks, 1972; Mandler & Johnson, 1977), organization theories (Bower, 1970; Tulving, 1968) and prototype theories (Cantor & Mischel, 1977a; Rosch, 1973) would seem to predict that fitting, consistent information would be better recalled than incongruent information. 31 Studies performed by Hastie (Hastie & Kumar, 1979; Hastie, 1980) found, however, that information which was incongruent with an initial impression produced much better recall effects than information which was consistent with an initial impression. Hastie hypothesized that giving subjects an initial trait ensemble would produce a certain perceptual set of expectancy. Behavioral information which would appear to be incongruent with the initial ensemble would appear to be surprising, striking and especially informative. It was further hypothesized that incoming events would elicit attention and encoding elaboration to the extent that they were informative about the personality impression. That is, information which is incongruent With the initial impression is seen as particularly salient and informative. Because of the informative nature of the information, it receives deeper processing and, therefore, is more easily retained and retrieved. In looking at the persuasion literature, a study by Brooks (1970) demonstrated that receivers hold shared expectations about what kinds of behaviors a communicator should exhibit. When these expectations are violated, receivers overreact to the behaviors which are actually exhibited. If a communicator who is initially perceived negatively performs more positive behaviors than expected, receivers tend to overestimate the positiveness of the unanticipated behaviors. In the same way, if an individual is initially perceived positively and then performs a 32 series of negative behaviors, individuals will overreact to the negativeness of those behaviors. Further research by McPeek and Edwards (1975) demonstrated that positive violations of expectations increase persuasibility only for initially negatively perceived sources. In other research concerning the violation of expectations in communication situations, Burgoon and Chase (1973) found that when innoculation messages differed in linguistic structure from attack messages, the amount of persuasion resulting from those messages differed. This finding led Burgoon and Stewart (1975) to propose that when communicators positively violate linguistic expectations, attitude change toward the advocated position increases. At the same time, when communicators negatively violate linguistic expectations, receivers react strongly in the direction of advocating the position opposite to the one advocated by the communicator. This research led to the development (Burgoon & Miller, 1984) of an expectancy interpretation of language and persuasion in which it was prOposed that the use of language which negatively violates normative expectations about appropriate communication behavior inhibits persuasive effectiveness and that the use of language which positively violates normative expectations by conforming more closely than anticipated to normative expectations of appropriate communication behavior facilitates persuasive effectiveness. 33 The findings in the persuasion research are consistent with the notion espoused by Hastie (1980) that events which contradict a previous expectation are given more importance due to their surprising nature. The contradictory information becomes more salient, more easily recalled and is given more attention in forming a final impression of an individual. Rationale of the Current Study Hastie's (1980) idea that we attribute more importance to events which contradict a previous expectation due to their surprising nature seems intuitively sensible. For example, many individuals deliberately set their expectations low so that they are not surprised when things do not turn out positively, rather than setting expectancies high and then receiving severe disappointment. If a person hears that a movie or a book is supposed to be excellent and then it turns out to be poor, the judgments may be harsher than if he or she heard the movie or book was poor to begin with due to the contrast effect. In certain extreme cases dealing with person perception, if a person is said to be friendly and he or she turns out to be unfriendly, the surprising disappointment may cause a perceiver to judge them as being even more unfriendly than if the individual was unfriendly to begin with. A simple averaging model based upon Anderson's (1965) information integration theory would predict a series of 34 specific predictions concerning the combined effect of trait and behavioral information. Using Anderson's model of 82 = WOSO + W131 W i-VV O 1 we can predict the outcome of combining new information with initial expectations in the following way: Table 1 Predicted cell means using various weights for information integration Initial Info/ _ _ _ Cell New Information 1 - w1 w0 — 2w1 2w0 - W1 or wO 1/2lJ l +a/+a +a +a +a 2 +a/-a 0 +a/ 3 -a/ 3 3 -a/+a O —a/3 +a/3 4 -a/-a -a -a -a This leads to the following hypothesis: H1: Given a positive expectation followed by negative behavior, an individual will be perceived more positively than if he or she were given a negative expectation followed by negative behavior (see cells 2 and 4, Table 1). H2: Given a negative expectation followed by positive behavior, an individual will be perceived less 35 positively than if he or she were given a positive expectation followed by a positive behavior (see cells 3 and 1, Table 1). H3: Given a positive expectation followed by positive behavior, an individual will be perceived more positively than if he or she were given any other combination of expectation and behavior (see cell 1 compared to all other cells, Table 1). H4: Given a negative expectation followed by a negative behavior, an individual will be perceived more negatively than if he or she were given any other combination of expectation and behavior (see cell 4 compared to all other cells, Table 1). These hypotheses are based on some assumptions. One assumption is that the opposite evaluative poles would be perceived with equal extremity for both initial trait information and subsequent behavioral information. Another assumption is that ratings for trait information and behavioral information would be matched at each end of the dimension. This version of the information integration model predicts that although there will be a significant main effect for both the initial expectation and subsequent behavior, there will be no significant interaction effect. In this model, both sources of information maintain their original relative weights when combined. Graphically, this model could be represented in the following way: a Pos. r Neg. Neg. Pos. Initial Expectancy Figure 1. Graphic representation of information integration model. A second model, based upon the idea that the surprising nature of contradictory information makes it more important or salient, would lead to a different set of predictions. Although an amount of information integration would take place in this model, this model would predict a significant interaction effect between the initial expectation and subsequent behavior. Because of the surprising nature of the behavioral expectation violation, this second model would predict that individuals will pay greater attention to the violation and differentially weight its importance in forming their impressions of an individual. This model could be graphically represented the following way: 37 a POS’\ Neg. Pos. Initial Expectancy Figure 2. Graphic representation of interaction model. In order to test this model, a condition is included in which individuals are given no prior information about the individual. Presuming that an individual's own expectation about another person is either initially neutral or minimally biased, the mean differences in ratings between the consistent and inconsistent conditions should be significantly larger than the differences in ratings between opposite behaviors when given no prior expectations. In this second model, by violating expectations in an extreme way, the "surprising" nature of the violation causes an individual to overreact and not only discount the initial expectation, but also to heavily weigh the subsequent behavior. This would create a more extreme 38 evaluation in the direction of a violation. Most of the studies of impression formation models and processes have dealt amost exclusively with the evaluative dimension of person assessment. This particular piece of research seeks to determine the generality of the findings for evaluative judgments to the two other response dimensions of activity and potency. In testing these two other dimensions, the hypothesized models are tested in the same way, substituting active/passive and strong/weak for positive/negative (as stated in the initial descriptions of the models). The study by Hamilton and Huffman (1971) which examined the generality of impression formation judgments for these two additional dimensions found that the processes were similar, suggesting that similar processes may underlie these dimensions in cases of extreme expectancy violation. Chapter 2 Method Overview Three separate studies were performed, each examining a different dimension of person assessment. The three dimensions examined were the evaluative, activity and potency dimensions. Each study used a different confederate as a stimulus person. A female confederate was used for the evaluative dimension, while male confederates were used for the activity and potency dimensions. Each confederate was used in two conditins, playing the bi-polar opposite ends of the dimensions. Subjects were given eight questions to ask confederates in what they were told was an exercise in teleconferencing. In actuality, all of the confederates' responses to the questions were videotaped, so that when the subjects asked the questions, they received the videotaped responses through telemonitors. Prior to asking the questions, subjects received either no information about the confederate, or information which confirmed or contradicted the way that the confederate behaved. Subjects were then asked to rate the confederates on personality characteristics corresponding to the specific assessment dimension. 39 40 Subjects Subjects were undergraduate students recruited from various communication courses. Participation was voluntary but subjects earned extra credit for their participation. All subjects were told that they would be participating in an interviewing study. Ninety subjects were used in each of the three studies for a total of 270 subjects overall. Design Three separate studies were performed, each examining one of the three dimensions of person assessment (evaluative, potency, and activity). For each study, a 3 X 2 design was used to determine the impact of the consistency versus inconsistency of preinteraction expectancies and subsequent behavior on impression formation. One factor in the design was the initial information subjects received about a stimulus person. Subjects received information describing a stimulus person one of six ways: pleasant, unpleasant, strong, weak, active, or passive. The information was designed to create an initial expectation about the person which would either be confirmed or disconfirmed by a second factor, the confederate's behavior. Some subjects received no information before their "interaction" with the confederate. Fifteen subjects were used in each cell of the study. 41 Dependent Measures In each of the three studies, trait terms reflecting the various dimensions of person assessment (pleasant- unpleasant, strong-weak, active-passive) were used as central terms in creating a dependent measure. Because there are individual differences in perceived relationships among traits, related trait terms used on the dependent measure were generated by a group of individuals similar to the actual subject population. Groups of individuals were asked to generate lists of adjectives that they felt were closely related to pleasant—unpleasant, strong-weak, and active-passive. From these generated lists, the most frequently mentioned trait names were selected to appear on the dependent measures for each dimension. Trait names were not used, however, if they appeared frequently on lists for more than one dimension (for example, "quiet" frequently appeared on lists generated for both "passive" and "weak"). This was.to keep the dimensions as conceptually separate from one another as possible. Trait terms for each dimension were arranged in a 15 item, nine-point semantic differential format. Scales used for each dimension are presented in Appendix A. Internal reliability was measured for each scale, using pretest data. Reliability was calculated using Cronbach's alpha. Reliability for the evaluative dimension scale was .99, with inter-item correlations ranging from .84 to .98. Reliability for the potency dimension scale was .99, with 42 inter-item correlations ranging from .77 to .96, and reliability for the activity dimension was .99, with inter-item correlations ranging from .66 to .97. Independent Variable Manipulations Two independent variable manipulations were used in the study: preinteraction expectancies and confederate behavior. For purposes in this study, it was important that several criteria were met in creating the stimulus materials. The first criterion was that the bi-polar ratings on the dependent measures were equidistant for each dimension. That is, the absolute value for the subjects' ratings of the confederate's pleasant behavior could not be significantly different from their ratings of the person's unpleasant behavior. The same was true for the other two response dimensions. Furthermore, the confederate's behavior could not be seen as overly extreme for any of the six conditions because of a concern that a ceiling effect might be created. In addition, the same concerns were true for the initial expectancy information. Information was designed to be equidistant in strength for each of the three response dimensions. The information was also designed so as not to be perceived as too extreme and was intended to match ratings of the behavioral ratings in strength. The ratings had a maximum negative value of -60 and a maximum positive value of +60. In designing the stimulus materials for the behavioral 43 information, eight questions were created to be used in the study. These eight questions were representative of the types of questions an individual might ask another in an initial interaction setting. Scripts were generated to provide answers that reflected the six personality traits (pleasant, unpleasant, strong, weak, active, passive) used in the study. The eight questions used in the study, as well as the scripted responses used in each condition are provided in Appendix B. Four confederates (two male and two female) videotaped the scripted responses, each portraying the six personality roles. All tapes were pretested to insure that confederates were actually seen to represent the various personality roles. The tapes were also pilot tested to insure that the confederates' performances on one end of the dimension were equidistant from their performance on the other end of the dimension. In pretesting the tapes, six groups of 15 subjects were shown four tapes of confederates' responses to the eight questions. All tapes were randomly arranged so that no subject heard two confederates answer the same questions the same way. Subjects were told that these individuals had been asked these questions in an interview and that the answers given were their honest responses. Subjects were also asked to put themselves in the situation of having just asked the questions personally. This was done so that the information processing would be similar to an actual 44 initial interaction. Subjects were then asked to rate the individuals on the various dependent measures. As mentioned previously, the confederate's performance on one end of the dimension was designed to be equidistant from his or her performance on the other end of the dimension. The confederate's gender was treated as a random variable. Selection for use in the actual study was based upon subjects' ratings of the confederates' performances. One confederate was chosen to portray both ends of each dimension based upon equidistant ratings for the opposite ends of the dimension. It was also important that the pretest ratings were not too extreme due to concerns of ceiling effects. Confederates with more moderate ratings were selected in this case. Confederate ratings were approximately the same for each of the three dimensions. Based upon this pretesting procedure, one female confederate was selected to portray opposite ends of the evaluative dimension (pleasant-unpleasant), one male confederate was chosen to represent opposite ends of the activity dimension (active-passive) and one male confederate was chosen to portray the opposite ends of the potency dimension (strong—weak). The following table represents E-tests for equidistance of the confederate performances for those tapes actually used in the study. Although the means reported in the table are the actual rating means, the E-values represent tests for differences 45 in the absolute values of the means. Each test used 14 degrees of freedom. Table 2 The t-Tests for Equidistance of Absolute Values of Ratings for Response Dimensions. Standard Mean Score Deviation t-value p-value Pleasant 43.93 6.91 < 1 62 Unpleasant -45.47 9.67 ° Active 38.47 5.77 passive -40.67 6.78 < 1 '35 Strong 39.40 7.22 Weak —39.20 9.42 < 1 '95 In addition to the tapes representing the behavioral manipulation, information was also created to manipulate preinteraction expectancies. Information was created to produce an expectation that the confederate was pleasant, unpleasant, strong, weak, active or passive. This information was pretested by telling groups of subjects that they were about to meet an individual and that before meeting this person they should read some information about the person. All information was written so that the gender of the person was unspecified. Subjects were then asked to rate what they expected the person to be like, using the dependent measures. The information was modified several times before ratings between the bi—polar ends of the 46 dimensions were equidistant. The information was also designed to produce similar ratings as the behavioral manipulations. The information used for each condition of the study is presented in Appendix C. The E—tests performed to test equidistance among the expectancy manipulations are presented in the following table. Again the means presented in the table are the actual rating means, while the E-values represent the difference in absolute values of the means. Each test used 14 degrees of freedom. Table 3 The E-Tests for the Equidistance of Absolute Values of Expectancy Manipulations. Standard Mean Score Deviation t-value p-value Pleasant 41.87 9.78 < 1 44 Unpleasant -38.93 10.85 ° Active 42.73 9.49 passive -39.27 8.70 1'04 '31 Strong 42.00 8.05 Weak -41.33 9.08 < 1 '83 Finally, statistical tests were performed in order to test the equidistance between the initial expectancy manipulation and the behavioral manipulation for each condition. These results are presented in Table 4. The tests used 14 degrees of freedom. 47 Table 4 The T-tests for Equidistance Between Expectancy Manipulations and Behavioral Manipulations Standard Mean Scores Deviation t-value p-value Pleasant Expectation Manipulation Pleasant Behavior Manipulation Unpleasant Expectation Manipulation Unpleasant Behavior Manipulation Active Expectation Manipulation Active Behavior Manipulation Passive Expectation Manipulation Passive Behavior Manipulation Strong Expectation Manipulation Strong Behavior Manipulation 41.87 43.93 -38.93 -45.47 42.73 38.47 -39.27 ~40.67 42.00 39.40 10.85 1.74 -1.49 (table .51 .09 .15 .62 .36 continues) 48 Table 4 (cont'd.) Standard Mean Scores Deviation t-value p-value Weak Expectation -41.33 9.08 < 1 .53 Manipulation Weak Behavior -39.20 9.42 Manipulation In addition to seeking equidistance, it was also important to test for the fact that individuals paid attention to both the initial information and the subsequent behavioral presentation. Had this not been the case, the effect size would have been too small to make a strong case for any model. Therefore, additional pretesting was done in order to test for the fact that the attention paid to the given written and behavioral information was significantly greater than 0. A nine-point, five item scale was used (see Appendix D) as a means of measurement. The E-tests comparing the obtained mean to 0 are presented in Table 5. The probability values for the obtained p—ratios being equal to zero were all less than .001 (with 14 degrees of freedom). One final pilot test was performed to examine individuals' own initial expectations for each of the three dimensions without being given any initial information. 49 Table 5 The E-Tests Comparing Means Measuring Amount of Attention Paid to Manipulations Against a Value of 0. Standard Mean Score Deviation t-value Pleasant Expectancy Manipulation Unpleasant Expectancy Manipulation Active Expectancy Manipulation Passive Expectancy Manipulation Strong Expectancy Manipulation Weak Expectancy Manipulation Pleasant Behavior Manipulation Unpleasant Behavior Manipulation Active Behavior Manipulation Passive Behavior Manipulation 28.93 29.87 29.80 29.13 30.27 31.13 28.87 26.80 28.40 30.33 5.44 26.65 34.64 26.37 30.60 32.40 21.47 21.38 12.59 21.51 33.85 (table continues) 50 Table 5 (cont'd.) Standard Mean Score Deviation t—value Strong Behavior 27.60 3.92 26.31 Manipulation Weak Behavior 27.60 4.32 23.89 Manipulation Research has consistently demonstrated that individuals hold a positivity bias toward others, i.e., they naturally expect others to be friendly and tend to make positive evaluations (DeSoto & Keuthe, 1959; Frauenfelder, 1974; Irwin, 1951; Kleinke, Bustos, Meeker, & Staneski, 1973; Marks, 1953; Mettee, 1971a, 1971b; Sears & Whitney, 1972). If a bias on any of the three response dimensions was too strong, a "no information" condition could not be used as a basis for contrast. Subjects were asked to give their impressions of an average individual whom they might meet off the street by filling out the dependent measures for all three response dimensions. Results demonstrated that individuals held a "neutral" initial impression for the activity and potency dimensions. A mild positivity bias was confirmed for the evaluative dimension, but it was significantly different from the initial impression created by the expectancy manipulation. Therefore, one's own 51 initial impression was used as a third point of contrast. Results of E-tests comparing obtained means for one's own initial impression on each dimension to a neutral zero point are presented in Table 6. Table 6 The E-Tests Comparing Means for One's Own Initial Impression to a Value of 0. Standard Mean Score Deviation t-value Pleasant/ Unpleasant 13.53 8.36 6.07 Active/ Passive 2.20 8.57 < 1 Strong/ Weak 1.73 8.33 < 1 The obtained E-value for pleasant/unpleasant was significant at the .001 level, using 14 degrees of freedom. Procedure In this study, subjects were led to believe that they would actually be interacting with the confederates through television monitors. The room in which the research was conducted was located next to a technical room. The situation was set up such that a video and audio signal could be sent from the tech room to the monitors in the research room. Cameras were set up so that subjects could 52 be seen and heard in the tech room at all times. Upon entering the study, subjects were greeted by a researcher and given experimental consent forms. After the subject filled out the consent form, the researcher read the following instructions: As you may be aware, a great deal of interviewing is now being done through the use of long distance teleconferencing. Therefore, we are conducting a study which looks at how peOple form impressions through video monitors in a question and answer interview setting. What we would like you to do is ask another individual, who is sitting in the tech room, a series of questions. He/she is situated behind a similar camera setup. At this point, the subject was led over to the door of the tech room. Inside the tech room, the confederate was sitting behind a camera, dressed the same as he or she was in the prepared stimulus tape. The researcher opened the tech room door for a few seconds, just long enough for the subject to get a look at the confederate. Once the subject had an opportunity to View the confederate and the bogus camera setup, the researcher proceeded to read the following: The person in the other room is participating in the study as part of a class requirement. As a result, the person has agreed to be interviewed and to take a series of personality measures. Also, the person has provided us with the name of several family members and personal contacts who know the individual well. From this information, we've come up with a brief personality profile. The person has read this profile and has agreed that it is basically accurate. Since you will be interacting with this person and will be asked to make a series of judgments about the person, it is important that you have some background information about the person. At this point, the researcher handed the subject a 53 description (see Appendix C) of the confederate which portrayed the individual as pleasant, unpleasant, strong, weak, active or passive. The subject was allowed to thoroughly read the information before going on with the experiment. In the conditions of the study in which subjects received no information about the confederate, this section of the study was deleted entirely. The researcher simply went on to the next section in which the subject was handed questions to read to the confederate. Following the subjects completing reading the profile, the researcher stated: Here is a list of eight questions to ask the person in the other room. (At this point the researcher handed the subjects the questions.) Because we're concerned with consistency in the study and because we're pretty inexperienced with the camera equipment, we'd ask you to read the questions exactly as they are written. Do not ask any follow-up questions. When you're done with one question, go on to the next. The person in the other room has been asked to give a brief, honest answer to each question. Remember, when you ask the question, the video monitor will be on you and when the person in the other room answers, the monitor will be on him/her. That is why it is important to keep the interview very structured. After these instructions were given to the subject, the researcher went into the tech room. The subject was viewed on a monitor camera. As soon as he or she finished asking a question, the researcher played a prerecorded answer from the stimulus tapes in response to the subject's question. As soon as a subject finished asking a question, a pause button was pushed to stop the tape. This left a darkened screen for the subject. When the subject asked the next 54 question, the pause button was released and the next response was allowed to play. This procedure was repeated until all eight responses were played. The tape was shut off immediately after the segment was over. As mentioned previously, in the cases in which individuals were given initial information about the confederate, the responses to the questions, the confederate's appearance and behavior were intended to confirm or radically violate the subject's initial expectation. Following the subject's "interaction" with the confederate, he or she was given one of the dependent measures to fill out for the appropriate response dimension. Following completion of the forms, the subjects were debriefed and excused from the study. Chapter 3 Results Analyses were performed separately for each of the three studies. Therefore, the results will be examined in three separate sections. Results for the Evaluative Dimension Analysis of variance was performed in order to test for the effects of the initial information manipulation and the behavioral manipulation on impression formation for the evaluative dimension, as well as to test for significant interaction effects. Results for the behavioral manipulations on participants' ratings of pleasantness/unpleasantness indicated a strong main effect for the behavioral manipulation. Those being exposed to the "pleasant" confederate consistently rated the confedeate more positively than those being exposed to the "unpleasant"' confederate (F = 480.484, df = 1, 84, 5 = .91). While the analysis of variance revealed a significant main effect for the initial expectation manipulation (F = 5.091, df = l, 84, p_‘<.01), the effect size for the linear trend was not particularly large (3 = .13). The analysis of variance is presented in Table 7. 55 56 mm.vmvH mm mm.OOOMMH HouOB HN.HMN vm nv.HNme Hmsowmmm Hoo. mm.mm mm.mHhNN m Nv.mhmMHH Umnflmamxm men. mmm. om.mm N om.MMH .>msmm\0mcH mcoHuomnoucH Hog 039 em. Hoo. mv.oov ow.HmOHHH H om.HmoHHH H0H>mcom Hoo. mo. was. mm.HmH H mm.HmH Hoaumnomsov mHo. Hoo. mm.m mo.NhHm H mo.NhHN Hummcflqv moo. mo.m Ho.nhHH m No.vmmm .0mcH HmHUHcH Hoo. mm.mmH Hm.mHmhm m Nw.mvvaH muoommm ch2 one osHm>|m m mumsom cmmz mo moumswm mo Esm monsom soncoEHQ m>HumsHm>m osu How mommanm> mo mHmHHmsg a eases 57 As demonstrated by the analysis presented in Table 7, there were no significant two-way interaction effects between the initial information and the subsequent behavior, hence supporting the linear combination model and rejecting the interaction model. The obtained means were all consistent with an information integration model. The highest evaluations were given to the positive information/ positive behavior condition, while the lowest evaluations were given to the negative information/negative behavior condition. Obtained ratings are presented in Table 8. Table 8 Ratings for the Evaluative Dimension Unpleasant One's Own Pleasant Expectation Expectation Expectation Pleasant 32 = 33.40 SE = 44.73 x = 44.93 Behavior SD = 17.57 SD = 11.96 SD = 13.16 Unpleasant i = -35.67 SE = -28.93 x = -23.13 Behavior SD = 10.83 SD = 13.78 SD = 21.33 Note. Low Score = -60. High Score = +60. Least significant difference (LSD) contrasts were performed to identify differences among specific cells at the .05 level of significance. This analysis demonstrated significant differences between the information consistent and the information inconsistent cells at both ends of the 58 evaluative dimension. Scheffe tests and Newman-Kuels tests did not reveal these differences, supporting the idea that the effect of the initial information was weak. Results for LSD contrasts are presented in Table 9. Since results for the Scheffe tests and Newman-Kuels tests were identical, they are both reflected in Table 10. The trends did indicate that at least some integration of information took place in the conditions where information was inconsistent. There were no differences revealed by any of the tests for the information consistent cells and the cells in which individuals received no initial information. This would be consistent with an averaging model, which would state that individuals, when given two pieces of information about a person, will average the two pieces of information to come up with an evaluation. Assuming that the value of the initial information was consistent with the value of the behavior, we would expect a similar evaluation for the average of the two pieces of information and for the behavior itself. In addition, a priori contrasts were performed to test the initial hypothesis directly. Contrast 1 tests the hypothesis that those given a positive expectation followed by negative behavior will perceive the individual more positively than those individuals given a negative expectation followed by‘a negative behavior; contrast 2 tests the hypothesis that those given a negative expectation followed by positive behavior will perceive the Table 9 Results for Least Significant Difference (LSD) Contrasts for Evaluative Dimension Subset 1 Subset 2 Group 6 5 Group 5 4 Mean -35.67 -28.83 Mean -28.93 -23.13 Subset 3 Subset 4 Group 3 Group 2 1 Mean 33.40 Mean 44.73 44.93 Note. Group 1 represents the positive expectation/positive behavior condition; Group 2 represents the no given expectation/positive behavior condition; Group 3 represents the negative expectation/positive behavior condition; Group 4 represents the positive expectation/negative behavior condition; Group 5 represents the no given expectation/negative behavior condition; and Group 6 represents the negative expectation/negative behavior condition. The subsets represent homogeneous subsets (subsets of groups, whose highest and lowest means do not differ by more than the shortest significance range [.05] for a subset Of that size. 60 Table 10 Results for Scheffe and Newman-Kuels Contrasts for the Evaluative Dimension Subset 1 Group 6 5 4 Mean -35.67 -28.93 -23.13 Subset 2 Group 3 2 1 Mean 33.40 44.73 44.93 Note. Group 1 represents the positive expectation/positive behavior condition; Group 2 represents the no given expectation/positive behavior condition; Group 3 represents the negative expectation/positive behavior condition; Group 4 represents the positive expectation/negative behavior condition; Group 5 represents the no given expectation/negative behavior condition; and Group 6 represents the negative expectation/negative behavior condition. The subsets are homogeneous subsets (subsets of groups, whose highest and lowest means do not differ by more than the shortest significance range [.05] for a subset of that size). 61 individual less positively than those given a positive expectation followed by positive behavior; contrast 3 tests the hypothesis that those given a positive expectation followed by positive behavior will perceive the individual more positively than those given any other combination of expectation and behavior; contrast 4 tests the hypothesis that those given a negative expectation followed by negative behavior will perceive the individual more negatively than those given any other combination of expectation and behavior. In addition, two contrasts were performed to test for interaction effects between initial expectations and behavior. Contrast 5 compares the distance between evaluations for the negative expectancy/ positive behavior condition and the negative expectancy/ negative behavior condition with the distance between evaluations for the no given expectation/positive behavior and the no given expectation/negative behavior conditions (see Figure 2, Chapter 1). Likewise, contrast 6 compares the distance between evaluations for the positive expectancy/positive behavior condition and the positive expectancy/negative behavior condition with the distance between evaluations for the no given expectation/positive behavior and the no given expectation/negative behavior conditions (also see Figure 2, Chapter 1). The results of the a priori contrasts are presented in Table 11. The results of the a priori contrasts are similar to those of the overall ANOVA. These results support the 62 Table 11 A priori Contrasts for the Evaluative Dimension Pooled Variance Estimates Value S. Error t-value df t—prob Contrast 1 12.53 5.55 2.26 84 .03 Contrast 2 11.53 5.55 2.08 84 .04 Contrast 3 46.85 4.30 10.89 84 .00 Contrast 4 49.87 4.30 11.59 84 .00 Contrast 5 -4.60 7.85 -.58 84 .56 Contrast 6 -5.60 7.85 -.71 84 .48 linear combination model, finding significance for all four tested hypotheses related to this model. Once again, there were no significant interactions demonstrated between the initial expectancy manipulation and the behavioral manipulation. Results for the Activity Dimension Analysis of variance was performed in order to test for the effects of the initial information manipulation and the behavioral manipulation on impression formation for the activity dimension, as well as to test for significant interaction effects. Results for the behavioral manipulations indicated a 63 strong main effect on participants' ratings of activity/ passivity. Those who "interacted" with the "active" confederate saw him to be significantly more active than when he portrayed the "passive" confederate (F = 811.674, df = 1, 84, p = .117). The effect size for the linear trend was small (5 = .07). Once again, the results demonstrated no significant interaction between the initial expectation and subsequent behavior. The analysis of variance is presented in Table 12. Since there was no significant effect for initial expectancy on ratings of activity, it can be concluded that subjects basically disregarded the initial information that they were given. Activity judgments were based primarily on the confederate's behavior. Ratings for the activity dimension are presented in Table 13. A priori contrasts were performed for the activity dimension in order to test several hypotheses directly. Contrast 1 tests the hypothesis that those given an active expectation followed by passive behavior will perceive the individual as more active than if they were given a passive expectation followed by passive behavior; contrast 2 tests the hypothesis that those individuals given a passive expectation followed by active behavior will perceive the individual to be less active than if they were given an active expectation followed by active behavior; contrast 3 tests the hypothesis that those individuals given active expectations followed by active behavior will perceive the 64 mm.NHMH mm mo.memHH HMDOB hh.mNH om om.Hom0H Hmsowmwm Hoo. mm.mmH hm.mmHHN m mm.mHmmOH UOGHMHme mom. mmo. mm.v m mh.m HOH>mnmm\omcH mcoHuomnopsH Hex 039 om. Hoo. no.HHm Ho.mmmmoH H Ho.ommmoH H0H>mnom moo. on. H on. ADHumuomsov moo. mo. mm.v mm.ohm H mm.onm HumocHHo nHH. om.m wm.mmm m mo.Hhm .owcH HMHuHcH Hoo. mo.~>~ nm.mommm m oH.hommoH muoommo chz mum msHo>um m mumsom cums mo moumsvm mo 85m mousom cOHmcoEHo mufl>flwofl on“ MOM mUQMHHm> mo mHmmHmcfl NH eHsee 65 Table 13 Ratings for the Activity Dimension Passive One's Own Active Expectation Expectation Expectation Active X = 30.60 E = 33.67 '2 = 37.53 Behavior SD = 10.79 SD = 8.91 SD = 14.57 passive p = -32.20 x = —34.47 a = ~31.80 Behavior SD = 11.92 SD = 11.21 SD = 10.15 Note. Low = —60. High = +60. individual to be more active than if they were given any other combination of expectation and behavior; and contrast 4 tests the hypothesis that those individuals given a passive expectation followed by passive behavior will perceive the individual to be more passive than if given any other combination of expectation and behavior. Contrasts 5 and 6 test for interaction effects between expectations and behavior. Contrast 5 compares the distance between evaluations for the passive expectation/ active behavior condition and the passive expectation/ passive behavior condition with the distance between evaluations for the no given expectation/passive behavior condition and the no given expectation/active behavior condition. Contrast 6 compares the distance between evaluations for the active expectation/passive behavior condition and the active expectation/active behavior 66 condition with the distance in evaluations for the no given expectation/active behavior and the no given expectation/ passive behavior conditions. Results of the a priori contrasts are presented in Table 14. Table 14 A Priori Contrasts for the Activity Dimension Pooled Variance Estimates Value S. Error t-value df t-prob Contrast 1 -5.40 4.16 -1.30 84 .20 Contrast 2 6.93 4.16 1.67 84 .10 Contrast 45.37 3.22 14.08 84 .00 Contrast 44.31 3.22 13.75 84 .00 Contrast -.33 5.88 -.06 84 .96 Contrast 1.20 5.88 .20 84 .84 Results for the a priori contrasts are consistent with the results for the overall ANOVA. The conditions which presented behavior consistent with expectations produced the most extreme evaluations. However, information- consistent conditions did not produce evaluations which were significantly different from information inconsistent conditions. This also supports the idea that individuals basically disregarded initial information for the activity 67 dimension. A priori contrasts did not demonstrate any interaction effects between initial information and behavior. Results for the Potenpy Dimension As with the evaluative and activity dimensions, analysis of variance was performed in order to test for the effects of the initial information manipulation and the behavioral manipulation on impression formation for the potency dimension and to look for significant interaction effects between the two manipulations. The results for the potency dimension showed a significant main effect for both the behavioral manipulation and the initial expectancy manipulation. Subjects exposed to the "strong" confederate rated him significantly more potent than when he portrayed the "weak" confederate (F = 461.778, df = 1, 84, 5 = .89). While this held true for all conditions, there was a significant main effect for initial expectancy (F = 17.075, df = l, 84, g = .24). The analysis of variance is presented in Table 15. Newman-Kuels contrasts revealed differences significant at the .05 level between the information consistent and the information inconsistent cells. This suggests that a good deal of information integration took place. Results for the Newman—Kuels tests are presented in Table 16. 68 mo.mNNH mm oH.NHmmoH Hmuoe mm.mmH em o~.aoemH Hnseamem Hoo. ma.OOH mm.eeemH m om.memma emsHsstm He. Hme. ao.m sn.mem N as.esHH nOHansmm\oesH mCOHuomumch HMS O39 on. Hoo. mh.Hov wm.mommm H wm.oommm uoH>mcmm boo. mm.H H mm.H Hoflumuomooo mmo. Hoo. «H.wm Hn.vvmo H Hh.vomo Humocflqv Hoo. mo.>H mo.m~Hm N no.ovmo .owcH HMHuHcH Hoo. Hm.moH nw.mHhom m Hv.mHHmm muoommm :Hmz mum osHm>Im m mumsvm new: mo moumsom mo Sow oousom concmEHQ hosmuom map you OOCMHHM> mo mHmHHmsé mH OHQMB Table 16 Results for Newman-Kuels Contrasts for the Potency Dimension Subset 1 Subset 2 Subset 3 Group 6 Group 5 Group 4 Mean —4l.20 Mean -30.27 Mean -12.87 Subset 4 Subset 5 Group 3 Group 2 1 Mean 26.00 Mean 36.13 38.80 Note. Group 1 represents the strong expectation/strong behavior condition; Group 2 represents the no given expectation/strong behavior condition; Group 3 represents the weak expectation/strong behavior condition; Group 4 represents the strong expectation/weak behavior condition; Group 5 represents the no given expectation/weak behavior condition; and Group 6 represents the weak expectation/ weak behavior condition. The subsets are homogeneous subsets (subsets of groups, whose highest and lowest means do not differ by more than the shortest significant range [.05] for a subset of that size). 70 As indicated in Table 15, there was a nearly significant two-way interaction between the initial expectancy and the confederate's subsequent behavior (p = .051). However, the effect size of this interaction was quite small, as it accounted for only about one percent of the total variance. This near-interaction can be traced to the condition in which the confederate was described as strong, but behaved weak. In this cell, subjects tended to discount the confederate's behavior more than usual. It can be argued that since subjects were in close proximity to the male confederate, and since they were unsure as to whether or not the confederate would see the ratings, they may have been reticent to rate the male confederate as weak after hearing him described as strong. This is further supported by the fact that Newman—Kuels tests revealed significant differences between the cell in which subjects received no information before seeing the weak confederate and the cell in which subjects received information which was consistent with the confederate's weak behavior. The averaging model would state that if the initial information is weighted and valued equally with the behavior, there should be similar ratings between the information consistent and no information conditions. This was true in the cell in which subjects received information describing a "strong" confederate in manner consistent with his behavior and cell in which subjects received no prior 71 information about a "strong" confederate. However, the no information/weak subjects rated the confederate to be significantly less weak than when receiving information describing him as weak. When the confederate was described as weak and behaved in a weak manner, subjects had no problems rating him as weak. However, when the confederate was described as strong or not described at all, the subject tended to give him more benefit of the doubt. All ratings for the potency dimension are presented in Table 17. Table 17 Ratings for the Potency Dimension Weak One's Own Strong Expectation Expectation Expectation Strong “SE: 26.00 2': 36.13 3? = 38.80 Behavior SD = 12.82 SD = 8.73 SD = 8.94 Weak 3? = -41.20 34' = -30.27 p = —12.87 Behavior SD = 15.63 SD = 14.28 SD = 18.59 Once again, a priori contrasts were performed to test several hypotheses directly. Contrast 1 tests the hypothesis that those given a strong expectation followed by weak behavior will perceive an individual as being stronger than if given a weak expectation followed by weak behavior; contrast 2 tests the hypothesis that those given a weak expectation followed by strong behavior will 72 perceive an individual as being weaker than if given a strong expectation followed by strong behavior; contrast 3 tests the hypothesis that those given a strong expectation followed by strong behavior will perceive an individual as being stronger than if given any other combination of expectation and behavior; and contrast 4 tests the hypothesis that those given a weak expectation followed by weak behavior will perceive an individual as being weaker than if given any other combination of behavior. Contrasts 5 and 6 test for interaction effects between the initial information and behavior. Contrast 5 compares the distance between evaluations for the weak expectation/strong behavior condition and weak expectation/weak behavior condition with the distance between evaluations for the no given expectation/weak behavior condition and the no given expectation/strong behavior condition. Contrast 6 compares the distance between evaluations for the strong expectation/weak behavior condition and the strong expectation/strong behavior condition with the distance between evaluations for the no given expectation/strong behavior condition and the no given expectation/weak behavior condition. Results of the a priori contrasts are presented in Table 18. Again, the results strongly supported the linear combination model, as all four hypotheses tested in relation to this model demonstrated significant results. Contrast 6 demonstrated the previously described 73 Table 18 A Priori Contrasts for the Potency Dimension Pooled Variance Estimates Value S. Error t-value df t-prob Contrast 1 28.33 4.98 5.69 84 .00 Contrast 2 12.80 4.98 2.57 84 .01 Contrast 3 43.24 3.86 11.21 84 .00 Contrast 4 52.76 3.86 13.68 84 .00 Contrast 5 .80 7.04 .11 84 .91 Contrast 6 -14.73 7.04 -2.09 84 .04 interaction due to the strong expectation/weak behavior cell. Using this statistical method, this interaction was demonstrated to be significant. Summary Results for the evaluative and potency dimensions demonstrated significant main effects for behavior and initial expectancy, although the effect size for initial information was not particularly large for the evaluative dimension. These results suggest that a linear combination model best represents the data. Analyses for the activity dimension revealed a main effect for behavior but not for information. While subjects relied primarily on the 74 confederates' behavior in forming impressions for all three dimensions, the subjects appeared to integrate their initial expectations into their final impressions for the evaluative and potency dimensions, while largely ignoring initial expectations for the activity dimensions. There were also no indications of negativity or positivity effects for any of the three dimensions. Had either of these effects been present, the contrasts would have demonstrated this type of interaction effect. However, a valid test for these effects requires an underlying assumption of an initially skewed distribution. For the activity and potency dimensions, individuals' initial impressions were neutral. Subjects only held an initial positivity bias for the evaluative dimension. Therefore, the negativity/positivity effects were only validly tested for this dimension. Chapter 4 Discussion Much of the work in impression formation has used some kind of evaluative response as a dimension of judgment. Basically, most of the research has been concerned with the processes individuals use in judging whether or not a person is likeable. First impressions, however, also involve judgments such as how energetic or lazy an individual is, and whether or not an individual holds power over us. The current research examined whether or not we arrive at these different judgments by the same process, particularly in situations where subsequent behavioral information strongly contradicts previous expectation- setting information. Initially, two models were proposed to explain these judgment processes. One was an averaging type of information integration model, such as that suggested by Anderson (1968). A simple averaging model would consider each element (in this case, the initial information and subsequent behavior) to be of equal importance in forming an overall impression. This type of simple averaging model is somewhat untenable since previous research (Anderson, 1965; Anderson & Jackobson, 1965) has indicated that the 75 76 various elements of a stimulus set are differentially weighted in the integration process. Stimulus characteristics typically contain properties which result in their receiving disproportionately large or small weights in the judgment process. A second model predicted significant interaction effects between the initial information manipulation and the confederates' subsequent behavior. It was suggested that in situations where behavior radically contradicts initial expectations, individuals will find this inconsistent information to be especially informative and will weigh it to an especially heavy degree. In this sense, individuals react to the initial information and discount it to a large degree. The results did not support the second model in any of the three assessment dimensions. There were no significant interaction effects of the previously described nature. While the subjects did use the confederates' behavior as the main basis for their final judgments, there was no indication that they reacted strongly against initial inconsistent information. Rather, the initial information appeared to be integrated into the final impression in two of the three dimensions examined in the study. In examining evaluative judgments, a simple averaging model appeared to represent the data when information and behavior were consistent. That is, evaluative judgments for the combined information were similar to the average 77 pretest rating value of the expectancy information and the behavioral information when the two were added together and then divided in half. The evaluations for the consistent information conditions were also similar to the evaluative ratings when subjects were given no prior expectation. However, when information was inconsistent, subjects tended to discount the initial information to a greater degree. While subjects still integrated this initial information into their final impression, it was given less importance. This same type of discounting effect was demonstrated by Anderson and Jacobson (1965). In this study, Anderson and Jacobson simultaneously presented subjects with three traits. One was markedly more positive or negative than the other two. Anderson and Jacobson found that individuals assigned the discounted trait a lesser weight. In some conditions, subjects were told explicitly that one of the traits did not describe the person. In these situations, the final evaluation was closer to the evaluations of the other two traits. Evaluations of the potency dimension also produced data consistent with an information integration paradigm. As with the evaluative dimension, the consistent trait and behavioral information cells produced an evaluation consistent with a simple averaging model. For the potency dimension, there was an even greater amount of information integration which took place among the inconsistent cells, although the confederate's behavior still carried the most 78 amount of weight in forming the final impression. In one cell, in which the confederate was described as being strong but behaved weak, the subjects tended to discount the behavior to a greater extent than in any other cell of the three studies. In this condition, subjects' ratings were much closer to the neutral point than in the counterpart condition where the individual was described as weak but behaved strong. It is likely that individuals were hesitant to describe the individual as weak, due to their close proximity to the male confederate. They may have thought that the confederate would have access to the ratings and therefore, were concerned about rating the confederate in a manner he might have considered insulting. This is supported by the fact that evaluations of the "weak" confederate when given no prior expectation are significantly less extreme than when given a "weak" expectation followed by "weak" behavior. When the confederate was described as "weak" and reportedly agreed that this was an accurate description (see Appendix C), it perhaps gave the subject license to rate the confederate as he or she really saw him. A study by Higgins and Rholes (1978) presented subjects with written descriptions of a stimulus person. When subjects had to communicate their impression to another person who supposedly liked the stimulus person, the communicated impression was more positive than when they had to communicate the impression to someone who disliked 79 the stimulus person. Interestingly, when subjects were asked to reproduce their impressions at a later time, their reproductions were consistent with the impressions they had communicated. In the Higgins and Rholes study, the act of communicating a positive impression created a positive impression. In the current case, the fear of communicating to the confederate that he was perceived as "weak" may have caused subjects to actualLyseetfimlas less weak. ‘When subjects were told that others close to the confederate saw him as weak, they may have perceived him as being more weak also. This particular effect was not seen in the evaluative dimension study, perhaps because the subjects found the confederate to be so unpleasant that they did not care what they communicated to her. Finally, for the activity dimension, the initial information which subjects received appeared to have no bearing on their subsequent impressions. Subjects neither integrated this information nor reacted to it, basing judgments strictly on behavioral information. Initial information was basically ignored in the final impression. The current findings are similar to those of Hamilton and Huffman (1971). In their study, Hamilton and Huffman examined combinations of trait terms to test an averaging model against a summation model of impression formation. Hamilton and Huffman examined the evaluative, potency and activity dimensions and found that their results for the evaluative and potency dimensions were consistent with an 80 averaging model. For active-passive judgments, the results found that neither the averaging nor summation models adequately accounted for judgments. In this case, however, individuals still integrated information, although in inconsistent ways. One of the distinctions between the current research and previous studies is that previous studies of information integration have primarily examined how individuals integrate similar kinds of information. For example, most of the Anderson studies have examined how individuals integrate trait information. The Hamilton and Huffman (1971) study also looked at integration processes using trait terms. In the current research, written trait information is combined with behavioral information. Initially, when subjects were given written trait information and behavioral information separately, they indicated that they would weigh both pieces of information about the same. This appears to be true when the information is consistent with behavior. However, individuals tend to discount the trait information when the information is inconsistent with behavior. In the case of consistent information and behavior, the information has a confirming effect for perceptions of behavior. In situations of inconsistency, the information has a mildly disconfirming effect, although behavior still serves as the primary basis for subjects' judgments. One likely reason for this is that subjects are more inclined 81 to trust their eyes rather than their ears. In other words, even if the information appears to be credible, people will still depend more on their personal experience with an individual in forming judgments, rather than on what they hear about a person. In forming judgments of activity, subjects paid attention to behavioral information only. These types of judgments require little attitudinal inference when compared to evaluative and potency judgments. A person may need to only look at another's behavior to determine whether they are active or not, whereas an individual will typically look at both a person's behavior and his or her psychological state when making attributions about friendliness or dominance. Therefore, background information about an individual will be considered more important in forming evaluative and potency judgments and of lesser importance when forming activity judgments. An important factor which was not controlled in the study concerns the set size of incongruent behaviors. Hastie and Kumar (1979) argue that set size, the number of incongruent and congruent behaviors attributed to a person, will be a major determinant in how well incongruent behaviors are recalled. Hastie and Kumar found that the smaller the size of the incongruent set, the higher the probability of recalling an item from the set. In the current research, no attempt was made to limit the number of incongruent behaviors by the confederates. The lack of 82 an interactive effect in cases where expectancies were violated may have been due to the large number of incongruent behaviors present. While the uncontrolled set size of incongruent behaviors may have had a significant effect on the results of the current study, the methodology used may be argued to be a valid representation of the processes that actually occur since there is no limit to the number of incongruent behaviors that may be present in "real life" expectancy violations. Another potential factor that may have had an impact on the results is related to ordering effects. This may have contributed to the differential weighting of behavioral information. Considerable support exists to suggest that the recency of activation of a construct category influences the accessibility of the category (Forbach, Stanners, & Hochhaus, 1974; Higgins, Rholes, & Jones, 1977; Warren, 1972; Wyer & Srull, 1980). According to Wyer and Srull (1980), a key aspect in determining priming effects is the time interval involved between the occurrence of the prime (e.g., giving subjects trait information about a stimulus person) and exposure to the information (e.g., the confederate's behavior) that is to be interpreted. The storage bin model proposed by Wyer and Srull (1980) implies that when several alternative concepts and schema could be potentially used to interpret new information, the one nearest to the top of the bin from which the material is being drawn will be the one that is 83 used. This would indicate that the most recently activated or accessed schema would be the one used to interpret the information. Priming may lead a relevant trait schema to be used in interpreting or encoding the new information, although the schema at the top of the bin will be the most important. As a result, the accessibility and the use of a particular schema to encode stimulus material is likely to decrease over the time interval between its prior activation and presentation of new material. A recent study by Garlick (1987) demonstrated an initial impact for trait information when little additional information was available. As subjects received additional information about the stimulus person, the more recently received information became the most important in terms of influencing subjects' judgments. As the initial trait information was pushed further down "into the bin," its impact in influencing subjects' judgments became less relevant until it was no longer a factor. In the current research, written trait information always preceded behavioral information. A logical follow-up to the current research would be to expose subjects to the confederates' behavioral cues first, and then allow the subjects to read trait information about the confederate. In this case, the behavior would set the expectation and recency effects would pertain to the subsequent trait information. It would be interesting in this case to see if any differential re—weighting occurred 84 for either the trait information or the behavioral information. One other factor which may have been significant pertains to the notion of mutual influence in interactions. Previous research (e.g. Bond, 1972) demonstrates that when individuals are given a positive or negative expectation about another person, they behave in ways that reflect that expectation. This causes a behavioral response from the other person which may be, at times, compensatory in nature. In the current study, all behavioral responses by the confederate were preset, allowing for no modification in response to the interviewer's behavioral style. This may have some effect on the external validity of the study. This research did not find support for the idea that information which contradicted an expectation would be seen as being particularly informative in forming impressions of people. Perhaps this can be explained by the fact that the work by Hastie (1980) and Hastie and Kumar (1979) dealt primarily with the recall of information. Their findings stated that information which was incongruent with expectations would be seen as surprising, and therefore would be differentially recalled. In this sense, the incongruent information would be expected to be especially informative and play a particularly significant role in making judgments about people. The Hastie research is primarily cognitive in nature. 85 Issues dealing with memory and recall of person information can be conceptually separated from how someone affectively responds to an individual. For years, contemporary theorists who studied the relationship between cognition and affect believed that affect was a post-cognitive experience. That is, an individual will make up his or her mind about how he or she feels toward a certain person after some considerable cognitive processing has taken place. An affective reaction, such as liking or disliking, would be considered to be based upon a prior cognitive' process in which a variety of content discriminations are made and features (traits) are identified, as well as examined for their value and contribution. In other words, before we can like or dislike anyone, we must know something about them. Perhaps this view was best summarized by Asch (1946) when he stated, "there has been a tendency to neglect the fact that emotions too have a cognitive side, that something must be perceived and discriminated in order that it may be loved or hated." Applying this to impression formation, an individual must recognize traits or qualities about an individual before they can make an evaluative judgment. Zajonc (1980) wrote a controversial paper in which he attacked the aforementioned opinion and stated that an affective response could not only occur without extensive and perceptual and cognitive encoding, but it can often occur before such encoding takes place. He goes on to make 86 the claim that affect and cognition are controlled by separate and partially independent neural systems. As a result of this claim, Zajonc believes that cognition and affect should be regarded as relatively independent subsystems rather than as fused and highly interdependent. In arguing for a "dual" system of processing for cognitive and affective responses, Zajonc argues that the affective qualities in impression formation are processed differently and perhaps separately from the cognitive content which "carries" the impression. He cites works by Anderson and Hubert (1963) and by Posner and Snyder (1975) to support this idea. In Anderson and Hubert's study, an impression formation task demonstrated a strong primacy effect for trait adjectives given early in a list of terms in regard to their overall impact on impression formation. However, there was an equally strong recall effect for the later adjectives given in the list. Anderson and Hubert suggest that the "impression response is based on a different memory system than that which underlies verbal recall." Dreben, Fiske, and Hastie (1979) also demonstrate that the weights calculated for the adjectives did not predict their recall. That is, adjectives which are assumed to be the most important in impression formation are not necessarily the same ones which will be effective in helping to retrieve the evaluative aspects of that same content. Zanjonc argues that it is not unreasonable to presume that the major difference between these two types 87 of cues may be the difference between what he terms "preferenda" and "discriminanda." "Discriminanda" refers to that which we can discriminate cognitively, while "preferenda" refers to those things related to the affective component. The difficulty with this view pertaining to the current findings lies in the fact that Anderson's information integration models are cognitive models in nature. Perhaps an alternative explanation can be suggested by referring to the information processing models provided in the first chapter of this dissertation. If information is not processed sequentially, but globally, cognitive processing may take place at a rapid rate of speed. Perhaps there are dual cognitive systems which are operating. There may be rapid cognitive processing, closely linked with the activation of stored categories, which is used in making evaluative judgments about people. This may be a different type of cognitive processing which is involved in the memory of traits or characteristics about persons. One other observation about the findings was a lack of a negativity effect for any of the dimensions. Briscoe, Woodyard, and Shaw (1967), for example, found that an unfavorable first impression is more resistant to change than a favorable one. No evidence was found to support any differential effects of negative information for any of the three dimensions. The results demonstrated that subjects considered the 88 initial information about the evaluative and potency dimensions to be more important than the information about the activity dimension. Hamilton and Huffman (1971) suggest that people more typically make evaluative and potency judgments in their initial encounters with others, but that the activity dimension is only of secondary importance in forming impressions. They suggest that when we first meet a person, we quickly gain a sense of whether or not we like him or her. It is also important for us to determine whether the person holds some kind of social power over us or whether we can behave authoritatively toward him or her. Thus, initial judgments of evaluation and potency may be of considerable importance if we are to successfully interact with the person. On the other hand, estimations of the individual's activity level may not be so important. Therefore, subjects may have paid differential attention to the initial descriptions for those reasons. Also, being likeable and being strong (if a male) are culturally valued to a much greater extent than being active. Therefore, the affective component of person evaluation may have played a role in heightening the subjects' awareness to the initial information. Perhaps the most significant question raised by the current research is how the term "first impression" may be usefully conceptualized and Operationalized. Many times this term is used to refer to one's first actual behavioral encounter with another person. Yet, much of the previously 89 cited research demonstrates that the communication of descriptive trait information may cause an individual to form a "first impression" before even meeting the person. A question is raised as to whether manipulations based upon information communicated by others are robust. The findings of this study would demonstrate that at least for two dimensions of person assessment, trait descriptive information does have an influence on subsequent impression formation. A useful follow-up study might make the whole sequence of information behavioral to see if this changes the predictive efficacy of the two models presented in this study. While information communicated through others had some impact on impression formation, the findings demonstrated that behavioral information had the most influence on impressions for all three dimensions. This underscores the importance of actual communicative exchanges and gives cause to question typical social psychological manipulations relying on "feeding in" of verbal characteristics or traits. While trait descriptions were demonstrated to have some influence on impression formation on at least two dimensions, it is possible that manipulating first impressions behaviorally would have had an even greater influence. An issue to be raised is whether information communicated by others through social networks has as much influence in forming a first impression as does an individual's actual first communicative exchange with 90 another person. The impact of initial information on impression formation has been studied for several decades now and yet there is still room to learn much more. This is an important area for communication scholars since initial impressions set the groundwork for subsequent interactions. Approaching impression formation from the direction of studying social cognition and information processing seems like a reasonable direction to go. APPENDICES APPENDIX A Dependent Measure Scales Nasty Nice Friendly Unfriendly Mean Kind Unhappy Happy Warm Cold Gloomy Cheerful Cruel Kind Rude Courteous Positive Negative Obnoxious Polite Caring Uncaring Agreeable Disagreeable Unpleasant Pleasant Inconsiderate Considerate Likeable Unlikeable Scale 1. Evaluative Dimension Scale 91 Inactive Busy Lively Dull Calm Excited Listless Dynamic Energetic Slow Uninvolved Involved Active Passive Laid-back Hyper Unmotivated ____________ Motivated Relaxed Restless Indifferent Eager Extroverted Introverted Patient Impatient Uninhibited Reserved Unconcerned Enthusiastic Scale 2. Activity Dimension Scale 93 Weak Forceful Powerful Powerless Domineering Submissive Meek Bold Yielding Defiant Unafraid Frightened Potent Feeble Unconfident Confident Frail Sturdy Timid Defiant Secure Insecure Defenseless Tough Dependent Independent Delicate Hearty Helpless Resourceful Scale 3. Potency Dimension Scale APPENDIX B Questions and Scripts Interview Questions 1. Why did you come to Michigan State? 2. What is your major area of study? 3. What would you like to do when you graduate? 4. What kinds of things do you do with your Spare time? 5. Are you involved with any organizations? 6. What kinds of sports do you like? 7. What are your favorite movies? 8. What type of music do you like? 94 95 Script for "Passive" Why did you come to Michigan State University? Well, my parents really liked MSU, so I thought that it was a good idea. What is your major area of study? Accounting. I like to work with numbers. I'm really good with quantitative abilities. What would you like to do when you graduate? I'd like to raise a family more than anything else and have a secure future. What kinds of things do you like to do with your spare time? I like to be by myself. I like to think, read or sit by the river on a nice day. Are you involved with any organizations? I'm really more into staying at home. I have been interested in looking into the chess club, however. What kinds of Sports do you like? Like I said, I don't like to go out much. I'm really not into sports. I like to read more than I like to watch sports. What are your favorite movies? I don't go to many movies either. I did see "On Golden Pond" and "Brighton Beach Memoirs." I really liked those movies. One final question, what type of music do you like? I like classical or beautiful music. My favorite is Mantovani. 96 Script for "Weak" Why did you come to Michigan State University? My parents told me that I should come here, so I figured that it was a good idea. What is your major area of study? I'm thinking about chemistry, but I'm not sure that I can make the grades. I don't know if I'm smart enough. I hope things will work out O.K. What would you like to do after you graduate? HOpefully, I can find something to do. I'm just going to go through interviews and see who'll take me. What kinds of things do you like to do with your spare time? Reading, playing chess. I usually like to be around people because I don't like to be alone. Are you involved with any organizations? I'm not really involved with anything right now. I have thought about the computer science club, but I'm not sure that I'm good enough with computers even though I like them. What kinds of sports do you like? I don't like sports. I'm afraid I'll get hurt. What are your favorite movies? I don't go to too many movies. There is too muCh sex, violence and harsh language. When I do go, I usually like funny movies like the Muppett Movie. One final question, what type of music do you like? Well my favorite is classical music. You'll probably think this is funny, but I really like Liberace. 97 Script for "Pleasant" Why did you come to Michigan State University? I really liked the pleasant atmosphere here when I came to visit. With all of the people and activities on campus I thought that it was a great place for me to make all of the friends I'd like. What is your major area of study? Communication, I think that it's a really important area. If people knew how to communicate better, the world would be a lot better off. What would you like to do when you graduate? I'd like to do something that involves working with people. I really like people and would like to do something to have a positive impact on their lives. I'd also like to be actively involved with community service organizations like the United Way. What kinds of things do you do with your spare time? My favorite activity is spending time with friends. I‘m also Hall president for my dorm.and do a lot of things with my floor. I also volunteer on weekends to be a big brother/big sister to a boy/girl from a troubled home. Are you involved with any organizations? Like I said before, I'm active in the Big Brother/Big Sister program and residence hall council. I also used to be involved with Student Foundation. What kind of sports do you like? I like softball, tennis, swimming, sailing and windsurfing. What are your favorite movies? I mainly like comedies. My favorites are Ghostbusters, Caddyshack and Back to the Future. One final question, what type of music do you like? I generally like soft-rock Top 40 music. I guess my favorites are Lionel Richie, Kenny Rogers and Huey Lewis. 98 Script for "Unpleasant" Why did you come to Michigan State University? Not that it's any of your business, but I couldn't get into Michigan. So, I got stuck here. Actually, it's not so bad, it's so big that I don't have to worry about associating with anybody. What is your major area of study? Finance, all I want to do is get out of here as quick as possible and make money. What would you like to do when you graduate? Well, my motto is "survival of the fittest." I'm going to do everything I can to look out for number one. I'll make money and be secure. What kinds of things do you like to do with your spare time? I don't have much spare time around here. I usually go home a lot because I don't get along with my roommates. I try to spend as little time around here as possible. Are you involved with any organizations? I think that clubs and organizations are a waste of time. People who join those things don't have anything better to do with their time. They're really pretty superficial. What kind of sports do you like? Sports are a waste of time, too. If people don't have anything better to do than bat or kick a stupid ball, then they've got problems. I've got better things to do. What are your favorite movies? Scarface, Platoon, and Exorcist. Usually ones with a lot of blood and gore. One final question, what type of music do you like? My favorite bands are Black Sabbath, Motley Crue and AC-DC. I also like Sammy Hagar and Ozzy Osbourne. 99 Script for "Active" 3’0 3’0 Why did you come to Michigan State University? Well, there were sure a lot of parties on campus. There were a lot of activities I could get involved in to broaden my horizons. What is your major area of study? I have a dual major in communication and business, I'll have to work really hard to get out of here in four years. That's OK, I generally like to keep on the go. What would you like to do when you graduate? Well, actually there are a number of things I'd like to do. I'd like to start a couple of different businesses. I also like to travel a lot. One other possibility is politics, I'm thinking about running for office someday. What kinds of things do you do with your spare time? I do tons of things! I'm real active in a number of organizations, I play on some intramural sports teams. I also try to get out involved in the community. I don't like to relax, I feel that you should always be doing something constructive with your time. Are you involved with any organizations? Like I said, I'm involved in tons of organizations. I play intramural basketball and floor hockey. I'm also involved in Hall Council and Student Foundation. I also work 10 hours a week in the cafeteria. What kinds of sports do you like? I like to jog, I also like hockey and basketball. What are your favorite movies? I'm a real active movie goer. I try to get to the movies as much as I can. I just saw Top Gun and Aliens. I really liked them both. One final question, what type of music do you like? I really like dance music. K001 and the Gang are probably my favorite. 100 Script for "Strong" Why did you come to Michigan State University? I wanted to go to a Big 10 school. I'm very confident that it was the best decision for me regardless of what anyone else says. Being here will give me an opportunity to grow as a leader. What is your major area of study? At this time, I'm taking courses in both political science and business administration. I'd like to go into either corporate administration or politics. What would you like to do when you graduate? Like I said, I'd like to either be a U.S. senator or a corporate executive. First, however, I have to fulfill my military commitments required by my scholarship through R.O.T.C. What kinds of things do you like to do with your Spare time? I play on a co-ed football team and participate in sports. Most of my time, however, is spent in R.O.T.C. We go on a lot of survival weekends. Are you involved with any organizations? Well, R.O.T.C. takes up most of my time. I do play on those intramural teams like I mentioned before. What kinds of sports do you like? Boxing, wrestling. Mainly sports that are competitive and aggressive like hockey or football. What are your favorite movies? I like a lot of the slasher movies like Halloween. I also really get into Clint Eastwood and John Wayne films. One final question, what type of music do you like? Quiet Riot, Motley Crue, any of the heavy metal bands. I also like some country-western music like Waylon Jennings and Hank Williams Jr. APPENDIX C Expectancy Manipulations (Pleasant Manipulation) You are about to interact with an individual. Before meeting the person, it is important that you know a little about the individual you will be interacting with. This person has agreed to be interviewed and to take a series of personalityanuiattitude measurements. Also, person has provided us with the name of several family members and personal contacts who know the individual well. From this information, the following profile is provided: This person has a positive, upbeat attitude towards life and has many friends. He is well-liked by all of the people who were contacted for interviews. This individual has a history of being involved in service activities such as the United Way and the Big Brother/Big Sister program, as well as many other community organizations. This person says that spending time with friends is a favorite activity. Furthermore, in giving this person a series of personality and attitude measures, the results showed an extremely healthy personality profile and demonstrated particularly high ratings on the traits of sincerity, considerateness, and tolerance toward others. This person's friends and family members consistently report this individual to be a friendly, pleasant person. After reading this profile, this person agreed that this was a basically accurate profile. 101 102 (Unpleasant Expectancy Manipulation) You are about to interact with an individual. Before meeting the person, it is important that you know a little about the individual you will be interacting with. This person has agreed to be interviewed and to take a series of personality and attitude measurements. ,Also, the person has provided us with the name of several family members and personal contacts who know the individual well. From this information, the following profile is provided: This person appears to have few friends. This person has a downcast attitude toward life and appears not to be liked very much by the people we contacted, including the person's roommates. The person has had a history of getting into fights and other altercations with the law and other authority figures. The person states that being left alone is preferable to dealing with people. Furthermore, in giving this person a series of personality and attitude measures, the results showed a relatively unhealthy profile and showed particularly high ratings on traits of hostility, selfishness, and intolerance toward others. Family members we talked to have described this individual as basically unpleasant and negative. After reading this profile, the person agreed reluctantly that this was a basically accurate profile. 103 (Active Expectancy Manipulation) You are about to interact with an individual. Before meeting the person, it is important that you know a little about the individual you will be interacting with. This person has agreed to be interviewed and to take a series of personality and attitude measurements. Also, the person has provided us with the name of several family members and personal contacts who know the individual well. From this information, the following profile is provided: The person loves being on the go most of the time and says that time is something which shouldn't be wasted. The person has had a history of involvement in a wide variety of clubs and organizations. This person loves spending time out in the community when not working toward completion of a dual major at MSU. Furthermore, in giving this person a series of personality and attitude measures, the results showed an active profile and showed high ratings on the traits of extroversion, restlessness, and activity. The person's family members and friends describe this individual, for the most part, as being active and involved. After reading this profile, the person agreed that this was an accurate profile. 104 (Passive Expectancy Manipulation) You are about to interact with an individual. Before meeting the person, it is important that you know a little about the individual you will be interacting with. This person has agreed to be interviewed and to take a series of personality and attitude measurements. Also, the person has provided us with the name of several family members and personal contacts who know the individual well. From this information, the following profile is provided: The person reports that reading at home or spending time thinking are favorite activities. The individual says that going out is an infrequent activity and that spending a quiet evening at home is much preferable. This person likes listening to mellow, beautiful music and enjoys watching beautiful sunsets. Furthermore, in giving this person a series of personality and attitude measures, the results showed a very passive profile and showed particularly high ratings on the traits of calmness, introversion, and being subdued. The person's family members and friends consistently described the individual as being a mellow and laid-back person. After reading this profile, the person agreed that this was a basically accurate profile. 105 (Strong Expectancy Manipulation) You are about to interact with an individual. Before meeting the person, it is important that you know a little about the individual you will be interacting with. This person has agreed to be interviewed and to take a series of personality and attitude measurements. Also, the person has provided us with the name of several family members and personal contacts who know the individual well. From this information, the following profile is provided: This person is actively involved in the Army R.O.T.C. program and frequently participates in survival weekends which often involves strenuous physical and mental activity. This person has a strong belief that people are masters of their own destiny and control their own fates. This person is aggressively pursuing a degree which hopefully will lead either to a position as a corporate executive or in politics after the individual fulfills the military service required through R.O.T.C. Furthermore, in giving this person a series of personality measures, the results showed a particularly aggressive profile and showed high ratings on the personality traits of dominance, forcefulness, and intensity. The person's family members and friends consistently described the individual as being strong and forceful. After reading this profile, the person agreed that this was a basically accurate profile. 106 (Weak Expectancy Manipulation) You are about to interact with an individual. Before meeting the person, it is important that you know a little about the individual you will be interacting with. This person has agreed to be interviewed and to take a series of personality and attitude measurements. Also, the person hasprovided usvdth.the name of several family members and personal contacts who know the individual well. From this information, the following profile is provided: This person has had a long history of illness and physical problems. The individual reports being frequently afraid and insecure about the future. The person has very few goals and feels that the future is determined more by luck or fate than anything else. Furthermore, in giving this person a series of personality and attitude measures, the results demonstrated an extremely weak profile and showed particular trait characteristics of insecurity, fearfulness, and timidity. The person's friends and family members consistently described the individual as being weak and feeble. After reading this profile, the person agreed that this was a basically accurate profile. APPENDIX D Scale Measuring Amount of Attention Given to Information When given information about an individual such as you have been given, how much impact would it have in forming your impression of the person? No impact Total at all 1 2 3 4 5 6 7 8 9 Impact How much influence would information such as this have in forming your impression? No Total influence influence at all 1 2 3 4 5 6 7 8 9 How much would you rely on this type of information in forming an impression? Wouldn't Would rely rely on it on it at all 1 2 3 4 5 6 7 8 9 totally How important would you consider this type of information to forming an impression of the person? Not Totally important important at all 1 2 3 4 5 6 7 8 9 How much would you depend on this type of information in forming an impression of the person? Wouldn't Would depend Depend on on it totally it at all 1 2 3 4 5 6 7 8 9 Scale 4. Scale measuring amount of attention given to information. 107 LIST OF REFERENCES LIST OF REFERENCES Anderson, N. H. (1965). Adding versus averaging as a stimulus combination rule in impression formation. Journal of Experimental Psychology, 10, 394-400. Anderson, N. H. (1968).‘ A simple model for information integration. In R. P. Abelson, E. Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, and P. H. Tannenbaum (Eds.), Theories of cognitive consistency: A sourcebook. Chicago: Rand-McNally. Anderson, N. H. & Hubert, S. (1963). Effects of concomitant verbal recall on order effects in personality impression formation. Journal of Verbal Learning and Verbal Behavior, 2, 379-391. Anderson, N. H. & Jacobson, A. (1965). Effect of stimulus inconsistency and discounting instructions on personality impression formation. Journal of Personality and Social ngchology, 2, 531-539. Asch, B. E. (1946). Forming impressions of personality. Journal of Abnormal and Social Psychology, 41, 258- 290. Bartlett, F. C. (1932). Remembering. London: Cambridge University Press. Birnbaum, M. (1972). Morality judgments: Tests of an averaging model. Experimental ngchology, 93, 35—42. Birnbaum, M. H. (1974). The non-additivity of personality impressions. Journal of Experimental Psychology, 102, 543—561. Bond, M. H. (1972). Effects of impression set on subsequent behavior. Journal of Personality and Social Psychology, 24, 301-305. Bower, G. H. (1970). Organizational factors in memory. Cognitive Psychology, 1, 18-46. 108 109 Bransford, J.D. & Franks, J.J. (1972). The abstraction of linguistic ideas: A review. Cognition, 2, 211-240. Briscoe, M.E., Woodyard, H.D., & Shaw, M.E. (1967). Personality impression change as a function of the favorableness of first impressions. Journal of Personality, 22, 343-357. Brooks, R.D. (1970). The generalizability of early reversals of attitudes toward communication sources. Speech Monographs, 21, 152-155. Bruner, J.S. (1958). Social psychology and perception. In E. Maccoby, T. Newcomb, & E. Hartley (Eds.), Readings in social psychology. New York: Holt-Rhinehart and Winston. Burgoon, M. & Chase, L.J. (1973). The effects of differential linguistic patterns in messages in attempting to induce resistance to persuasion. Speech Monographs, 40, 1-7. Burgoon, M. & Miller, G.R. (1985). An expectancy interpretation of language and persuasion. In H. Giles & R.N. St. Clair (Eds.), Recent advances in language, communication, and socialgpsychology. London:‘ Lawrence Erlbaum Associates. Burgoon, M. & Stewart, D. (1975). Empirical investigations of language intensity: I. The effects of sex of source, receiver, and language intensity on attitude change. Human Communication Research, 2, 244-248. Cantor, N. & Mischel, W. (1977a). Traits as prototypes: Effects on recognition memory. Journal of Personality and Social Psychology, 25, 38-48. Cantor, N. & Mischel, W. (1977b). From personality impression to social inference: The role of pre-existing knowledge in structuring information processing. Unpublished manuscript, Stanford University. Cusumano, D.R. & Richey, M.H. (1970). Negative salience in impressions of character: Effects of extremeness of stimulus information. Psychonomic Science, 22(2), 81-83. D'Andrade, R.G. (1965). Trait psychology and componential analysis. American Anthropologist, 67, 215-228. D'Andrade, R.G. (1974). Memory and the assessment of behavior. In H. Blalock (Ed.), Measurement in the social sciences. Chicago: Aldine-Atherton. 110 DeSoto, C.B. & Kuethe, J.L. (1959). Subjective probabilities of interpersonal relationships. Journal of Abnormal and Social Psychology, 52, 290-294. Dreben, E.K., Fiske, S.T., & Hastie, R. (1979). The independence of evaluative and item information: impression and recall order effects in behavior based impression formation. Journal of Personality and Social ngchology, 21, 1758-1768. Feldman, S. (1966). Motivational aspects of attitudinal elements and their place in cognitive interaction. In S. Feldman (Ed.) Cognitive consistency: Motivational antecedents and behavioral consequents. New York: Academic Press. Fiske, S.T. (1980). Attention and weight in person perception: The impact of negative and extreme behavior. Journal of Personality and Social Psychology, 22(6), 889-906. \ Fiske, S.T. & Pavelchak, M.A. (in press). Category-based vs. piecemeal based affective responses: Development in schema triggered affect. In R.M. Serrentino & E.T. Higgins (Eds.), The handbook of motivation and cognition: Foundations of social behavior. N.Y.: Guilford Press. Forbach, G.B., Stanners, R.F., & Hochhaus, L. (1974). Repetition and practice effects in a lexical decision task. Memory and Cognition, 2, 337-339. Frauenfelder, K.J. (1974). A cognitive determinant of faborability of impression. Journal of Social Psychology, '24, 71-81. Frijda, N.H. (1969). Recognition of emotion. In L. Berkowitz (Ed.), Advances in experimental social psychology (vol. 4). New York: Academic Press. Garlick, R. (1987). Examining the lasting impact of trait descriptions on impression formation. Paper presented at the annual meeting of the International Communication Association, Montreal. Goodenough, F. (1930). Interrelationships in the behavior of young children. Child Development, 2, 29-47. Hamilton, D.L. & Huffman, L.J. (1971). Generality of impression formation processes for evaluative and nonevaluative judgments. Journal of Personality and Social Psychology, 22, 200-207. 111 Hamilton, D.L. & Zanna, M.P. (1972). Differential weighting of favorable and unfavorable attributes in impressions of personality. Journal of Experimental Research in Personality, 2, 204-212. Hastie, R. (1980). Memory for behavioral information that confirms or contradicts a personality impression. In R. Hastie, T.M. Ostrom, E.B. Ebbesen, R.S. Wyer, D.L. Hamilton, D.E. Carlston (Eds.), Person memory: The cognitive basis of social perceppion. Hillsdale, N.J.: Lawrence Erlbaum Associates. Hastie, R. & Kumar, P.A. (1979). Person memory: Personality traits as organizing principles in memory for behaviors. Journal of Personality and Social Psychology, 21, 25-38. Higgins, E.T. & Rholes, W.S. (1978). "Saying is believing": Effects of message modification on memory and liking for the person described. Journal of Experimental Social Psychology, 12, 363-378. Higgins, E.T., Rholes, W.S., & Jones, C.R. (1977). Category accessibility and impression formation. Journal of Experimental Social Psychology, 12, 141—154. Hodges, B.H. (1974). Effect of valence of relative weighting in impression formation. Journal of Personality and Social Psychology, 22(3), 378-381. Ickes, W., Patterson, M.L., Rajecki, D.W., & Tanford, S. (1982). Behavioral and cognitive consequences of reciprocal versus compensatory responses to preinteraction expectancies. Social Cognition, 1(2), 160-190. Irwin, F.W. (1953). Stated expectations as functions of probability and desirability of outcomes. Journal of Personality, 21, 329-335. Jordan, N. (1965). The "assymetry" of "liking" and "disliking": A phenomenon meriting further replication and research. Public Opinion Quarterly, 22, 315-322. Kaplan, M.F. (1971). Dispositional effects and weight of information in impression formation. Journal of Personalipy and Social Psychology, 12, 279-284. Kaplan, M.F. (1972). The modifying effect of stimulus information on the consistency of individual differences in impression formation. Journal of Experimental Research in Personality, 2, 213-219. 112 Katz, D., Gutek, B., Kahn, R.L., & Burton, E. (1975). Bureaucratic encounters. Ann Arbor, MI.: Institute for Social Research, University of Michigan. Kellerman, K. (1984). The negativity effect and its implications for initial interactions. Communication Monographs, 21, 37-55. Kelley, H.H. (1950). The warm-cold variable in first impressions of persons. Journal of Personality, 12, 431-439. Kelley, G.A. (1955). The psychology of personal constructs. New York: Norton. Kleinke, C., Bustos, A., Meeker, F., & Staneski, R. (1973). Effects of self-attributed and other attributed gaze on interpersonal evaluations between males and females. Journal of Experimental Social Psychology, 2, 154-163. Knutson, T. & Holdridge, W. (1975). Orientation behavior, leadership and consensus: A possible functional relationship. Speech Monographs, 22, 107-114. Lau, R.R. (1980). Negativity in political perceptions. Unpublished manuscript, Department of Psychology, University of California, Los Angeles. Levin, I.P. & Schmidt, C.F. (1969). Sequential effects in impression formation with binary intermittent responding. Journal of Experimental Psygholqu, 12(2), 283-287. Levy, L.H. & Dugan, R.D. (1960). A constant error approach to the study of dimensions of social perception. Journal of Abnormal and Social Psychology, 21, 21-25. Lucas, R. & Jaffe, C. (1969). Effects of high-rate talkers on group voting behavior in the leaderless-group problem-solving situation. Psycholggical Reports, 22, 471-477. Mandler, J.M. & Johnson, N.S. (1977). Rememberance of things parsed: Story structure and recall. Cognitive Psychology, 2, 111-151. Marks, R.W. (1951). The effect of probability, desirability, and privilege on the stated expectations of children. Journal of Personality, 12, 431-465. McPeek, R. W. & Edwards, J.D. (1975). Expectancy disconfirmation and attitude change. Journal of Social Psychology, 22, 193-208. 113 Metee, D.R. (1971a). Changes in liking as a function of the magnitude and effect of sequential evaluation. Journal of Experimental Social Psychology, 7, 157-172. Metee, D.R. (1971b). The true discerner as a potent source of positive affect. Journal of Experimental Social Psychology, 7, 292-303. Miller, J. & Rowe, P. (1967). Influence of favorable and unfavorable information upon assessment decisions. Journal of Applied Psychology, 51, 432-435. Morris, C. & Hackman, J. (1969). Behavioral correlates of perceived leadership. Journal of Personality and Social Psychology, 13, 350-361. Mulaik, S.A. (1964). Are personality factors raters' conceptual factors? Journal of Consulting Psychology, 22, 506-511. Nisbett, R.E. & Wilson, T.D. (1977). The halo effect: Evidence for unconscious alteration of judgments. Journal of Personality and Social Psychology, 22, 250- 256. Osgood, C., Suci, G.J., & Tannebaum, P.H. (1957). The measurement of meaning. Urbana, 111.: University of Illinois Press. Posner, M.I. & Snyder, C.R. (1975). Facilitation and inhibition in the processing of signals. In P.M. Rabbitt & S. Dornic (Eds.), Attention andgperformance. New York: Academic Press. Richey, M.H. & Dwyer, J.S. (1970). Negative salience in impressions of character: Sex differences. Psychonomic Science, 22(2), 77-79. Richey, M.H., McLelland, L., & Shimkunas, A.M. (1967). Relative influence of positive and negative information in impression formation and persistence. Journal of Personality and Social Psyghology, 6, 322-327. Riecken, H. (1958). The effect of talkativeness on ability to influence solutions to group problems. Sociometry, 21, 309-321. Rosch, E. (1973). On the internal structure of perceptual and semantic categories. In T.E. Moore (Ed.), Cogpitive development and the acguisition of language. New York: Academic Press, 1973. 114 Rosenberg, S. (1977). New approaches to the analysis of personal constructs in person perception. In A.W. Landfield (Ed.), Nebraska symposium on motivation, 1976. Lincoln: University of Nebraska Press. Rosenberg, 8., Nelson, C., & Vivekananthan, P.S. (1968). A multidimensional approach to the structure of personality impressions. Journal of Personality and Social Psychology, 2, 283-294. Rosenberg, S. & Olshan, K. (1970). Evaluative and descriptive aspects in personality perception. Journal of Personality and Social Psychology, 12, 619-626. Rosenberg, S. & Sedlak, A. (1972). Structural representations of implicit personality theory. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 6). New York: Academic Press. Rumelhart, D.E. & Ortony, A. (1977). The representation in memory. In R.C. Anderson, R.J. Spiro & W.E. Montague (Eds.), Schooling and the acquisition of knowledgp. Hillsdale, N.J.: Lawrence Erlbaum Associates. Schneider, D.J. & Blankmeyer, B. (1978). The effects of prototype salience on trait inferences. Unpublished manuscript, University of Texas at San Antonio. Schneider, D.J., Hastorf, A.H., & Ellsworth, P.C. (1979). Person perception (2nd edition). Reading, Mass: Addison: Wesley. Sears, D.O. & Whitney, R. (1973). Political persuasion. In 1. Pool, S.W. Frey, W. Schramm, N. Maccoby, & E.B. Parker (Eds.), Handbook of Communication. Chicago: Rand-McNally. Snyder, M., Tanke, E.D., & Berscheid, E. (1977). Social perception and interpersonal behavior: On the self- fulfilling nature of social stereotypes. Journal of Personality and Social Psychology, 35, 656-666. Swann, W. & Snyder, M. (1980). On translating beliefs into action: Theories of ability and their application in an instructional setting. Journal of Personality and Social Psychology, 38, 879-888. Thibaut, J.W. & Riecken, H.W. (1955). Some determinants and consequences of the perception of social causality. Journal of Personality, 24, 113-133. 115 Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.), Organization of memory. New York: Academic Press. Warr, P. & Jackson, P. (1976). Three weighting criteria in impression formation. European Journal of Social Psychology, 6, 41-49. -+ Warr, P.B. & Knapper, C. Thepperception ofgpeople and events. New York: Wiley. Warren, R.E. (1972). Stimulus encoding and memory. Journal of Experimental Psychology, 94, 90-100. Willard, D. & Strodtbeck, R. (1972). Latency of verbal response and participation in small groups. Sociometpy, .22, 161-175. Wishner, J. (1960). Reanalysis of "Impressions of personality." Psychologica1 Review, 21, 96-112. Woll, S. & Yopp, H. (1978). The role of context and inference in the comprehension of social actions. Journal of Experimental Social Psychology, 12, 351-362. Wyer, R.S. (1970). Information redundancy, inconsistency, ,and novelty and their role in impression formation. Journal of Experimental Social Psychology, 2, 111-127. Wyer, R.S. (1974). Cognitive organization and change: An information processing approach. Potomac, Md.: Erlbaum, 1974. Wyer, R.S. & Hinkle, R.L. (1976). Informational factors underlying inferences about hypothetical persons. Journal of Personality and Social Psychology, 22, 481-495. Wyer, R.S. & Srull, T.K. (1981). Category accessibility: Some theoretical and empirical issues concerning the processing of social stimulus information. In E.T. Higgins, C.P. Herman & M.P. Zanna (Eds.), Social cognition: The Ontario symposium on_personality and social psychology. Hillsdale, N.J.: Erlbaum, 1981. Zajonc, R.E. (1980). Feeling and thinking: Preferences need no inferences. American Psycholggist, 22, 151- 175.