in" 5 II :E D3 313933363 3.31.8333; 333 C33 33.335333 WRmP.33"€ 3? 3E3 E333 3033 SHARAC "35333 “‘AQF?‘ Q :PE- PG; tit-L1 fig :fil- 3‘ ll- 3335;333:335 .vi: 3 5‘33: \ :3'3E Jail; 'L. we {'33.- la "F“r’flfiifl Ampm.‘ .mm" 3.... "33*": A“?! H 3‘: ‘u 3:”! PI align”; '2 Rig £23. ¥':} 9 “'1‘ 3535 3333333333333”133333 3333 31293 01730 3417 r V“ n A a or hit-LA. ' 3 ' -: ‘_. University This is to certify that the thesis entitled The Dimensional Structure of Children's Perception of Television Characters presented by Byron B. Reeves has been accepted towards fulfillment of the requirements for Ph.D. Communication degree in ALMS 3 .33 r M4>profess<3rl February 20, 1976 Date 0-7639 Mi? 331332062002 321302 .ABSTRACT THE DIMENSIONAL STRUCTURE OF CHILDREN'S PERCEPTION OF TELEVISION CHARACTERS By Byron Bonn Reeves The purpose of this study was to examine the perceptual di- mensions used by Children to differentiate television Characters. The utility of these dimensions in predicting media effects and subgroup differences in dimension content were also examined. All past researCh in this area had used dimensions defined by the researCher as important in understanding television's imr pact on Children. Therefore, the methods used to define the dimen- sions without suggesting their content were very important in this study. MUltidimensional scaling of lh TV characters was accomplished by asking third, fifth, and seventh graders to judge whether all possible pairs of the sample of Characters were alike or different. This method.had the advantage of letting Children differentiate among the Characters by whatever attributes they chose to think about. These results were then compared with unidimensional evalua- tions of characters on attributes defined by pre-test and past re- searCh. Four interpretable dimensions were found.from.the multidimen- sional analysis. In decreasing order of variance explained in the final solution they were: unsupported.humor, masculine strength, feminine attractiveness, and activity. An hypothesis that older children would use more dimensions to differentiate TV Characters than younger children was not supported. One hypothesis concerning developmental differences in dimen- sion content was supported. Younger children were more reliant on dimensions descriptive of physical attributes (strength and at- tractiveness) and older Children depended more on the dimensions descriptive of TV character's behavior (unsupported humor and activity). Although no hypotheses about sex differences in dimension content were made, two sex differences were found. Males weighted support and strength more than females, and females weighted at- tractiveness more than males. Good was related to strength for males and to attractiveness for females. For the total sample of children, the four dimensions pre- dicted whether they wanted to be like or "do" like television characters. The strength and activity dimensions predicted these measures best for the males and attractiveness predicted best for the females. THE DIMENSIONAL STRUCTURE OF CHILDREN'S PERCEPTION OF TELEVISION CHARACTERS By Byron Bonn Reeves A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree 0f DOCTOR OF PHILOSOPHY Department of Communication 1975 ACKNOWLEDGMENTS Bradley Greenberg was my advisor and critic. Recognition goes first to him for his time and encouragement during the prep- aration of this dissertation and throughout my graduate education. Three other committee members, Charles Atkin, Edward.Fink, and.Ellen Strommen, are to be thanked for their assistance and review of the dissertation. Two school principals and numerous teaChers allowed us to in- terrupt classrooms and interview their students. For this assistance, thanks go to Mr. Doonan, principal of the Murphy Elementary SChool, Haslett, Michigan; Mr. Farr, principal of the Haslett Middle School, Haslett, Michigan; and teachers at both schools. I also appreciate the interviewing done by Roxanne Reeves and.Mark Milkovich, both graduate students at Michigan State University. Finally, I aCknowledge the help of’Mark Miller, a friend who shared the more infermal atmospheres where good ideas are often conceived. ii Chapter II III IV TABDE OF CONTENTS INTRODIICTION O O O O O O O I O O O O O O O O O 0 General Considerations in Person Perception. . . Dimensions of Perception . Sex. 0 O C O O O O O O O O O O 0 I O O O O Ageoooooooo 0.0.00.0... Physical Strength. . . . Physical Attractiveness. . Perceived Reality. . . . . Hmorooooooooooo GOOdneSSOOOOOOOOOOOOOI... Support From.0ther Characters. . . . . . . Developmental.Differences in Perceptual Dimensions . . . . . . . . . . . . . . . . Sex Differences in Perceptual Dimensions . . . . Perceptual Dimensions as Predictors of Media.Effects. . . . . . . . . . . . . . . Summary of Chapter I . . . . . . . . . . . . . . MODS O O O O O O O C O O O O O O O O O O O O 0 Individual Interviews. . . . . . . . . . . . . . Group Interviews . . . . . . . . . . Multidimensional Data and.Analysis . Sample of TV Characters. Similarity Judgments . . The INDSCAL Program. . . Unidimensional Measures. Media Effects Variables. Data Analysis. . . . . . RESUI'TS O O O O O C O O I C O O O O O O O O O O O Dimensionality of INDSCAL Solutions. . . . . . . Dimension Content. . . . . . . . . . . . . Subgroup Differences in Dimension Content. Sex Differences in Dimension Content . . . Perceptual Dimensions As Predictors of Media Effects. . . . . . . . . . . . . SUMMARY AND DISCUSSION . . . . . . . . . . . . . Dimensionality of Subgroup Spaces. . . . . . . . Dimension Content. . . . . . . . . . . . . . . . Differences in the Dependent Measures. . . . . . The INDSCAL.Method . . . . . . . . . . . . . . . FutureReseal‘ch................. Present Implications of Research Findings. . . . Relation of Results to Past ResearCh . . . . . . iii TABLE OF CONTENTS (cont'd) BIBLIOGRAPHY APPENDICES A Pre-test Questionnaire B Description of Television Characters Used for Similarity Judgments C Group Questionnaire for Similarity Judgments D Subgroup Distance Matrices for Similarity Ratings E Group Questionnaire for Unidimensional Measures F Mean Scores for Fourteen Television Characters on Eight Unidimensional Attributes G Plots of Two-Dimensional Comparisons of Fourteen Television Characters (Total Sample) H INDSCAL Coordinates fOr Fourteen Television Characters on Four Dimensions (Total Sample) I Zero Order Correlations Between Nine Uni- dimensional Ratings for Each Grade and Sex iv LIST OF TABLES Table Page 1 Percentage of Variance Accounted for by INDSCAL Analysis with Different Dimensional Solutions (Total Group and Third, Fifth, and Seventh Grades). . AB 2 Zero Order Correlations Among Four INDSCAL DimenSionS . O O O O O O I O O O O O O I O O O O O O O 53 3 Zero Order Correlations for Entire Sample Among Nine Unidimensional Ratings of Fourteen Television Characters. . . . . . . . . . . . . . . . . . . . . . 55 h Zero Order Correlations and Standardized Regression Weights for Predicting Mean Ratings on Unidimensional Scales from Four INDSCAL Dimensions (Total Group) . . 56 5 Zero Order Correlations and Standardized Regression Weights for Predicting Mean Ratings on Unidimensional Attributes from Four INDSCAL Dimensions (Third Grade, Fifth Grade, and Seventh Grade) . . . . . . . . . . . 63 6 Salience Scores for Three Age Groups from an INDSCAL Analysis of Data on Similarities Among Fourteen Television Characters . . . . . . . . . . . . . . . . 65 7 Canonical Correlations of’Four INDSCAL Dimensions for Three Separate Age Levels (Third Grade, Fifth Grade, and Seventh Grade). . . . . . . . . . . . . . . . . . 7O 8 Zero Order Correlations and Standardized Regression Weights for Predicting Mean Ratings on Unidimensional Attributes from Four INDSCAL Dimensions (Males and Females). . . . . . . . . . . . . . . . . . . . . . . 71 9 Zero Order Correlations and Standardized Regression Weights for Predicting Mean Ratings on Two Media Effects Variables from Four INDSCAL Dimensions (TotalGroup)....................75 10 Zero Order Correlations and Standardized Regression weights for Predicting Mean Ratings on Two Media Effects Variables from.Four INDSCAL Dimensions (Males and Females). . . . . . . . . . . . . . . . . . . . . 77 Table 12 LIST OF TABLES (cont'd) Zero Order Correlations and Standardized Regres- sion Weights for Predicting Mean Ratings on Two Media Effects Variables from Four INDSCAL Dimen- sions (Third Grade, Fifth Grade, and Seventh Grade)......... Zero Order Correlations and Standardized Regression Weights for Predicting Mean Ratings on Two Media Effects Variables from Four INDSCAL Dimensions (Grade by Sex Breakdown) . . . . . . . . 78 92 LIST OF FIGURES Total Group Concept Space (Hypothetical) . . . . . Concept Space for Subgroup 2 (Hypothetical). . . . Concept Space for Subgroup 3 (Hypothetical). . . . Space of Salience Scores for Six Subgroups (Hypothetical).................. Percentage of Variance Accounted for by an INDSCAL Analysis With Different Dimensional Solutions. . . Dimensions 1 and 2 of the Four Dimensional Group Concept Space From an INDSCAL.AnalySis of Data on Perceived Similarities Among Fourteen Television Characters . . . . . . . . . . . . . . . . . . . . Dimensions 3 and h of the Four Dimensional Group Concept Space From an INDSCAL Analysis on Data on Perceived Similarities Among Fourteen Television Characters . . . . . . . . . . . . . . . . . . . . Subgroup Spaces for Three Grades on Four INDSCAL DimenSionS O O O O O O O O O O O O O O O O O O O 0 vii A9 51 52 67 Chapter I INTRODUCTION In studying cognitive processes that intercede between children's exposure to television and their use of TV information, researchers have assumed that several perceptual dimensions are operative. Studies have examined the effects of intervening variables such as the perceived reality of TV portrayals (cf. Greenberg and Reeves, 197A; Feshbach, 1971); differential impact of male and female Characters (cf..Atkin and.Miller, 197M; Miller and Reeves, 197M); amount of violence attributed to characters (of. McLeod, Atkin, and Chaffee, 1971); the amount of support received from other characters (cf. Walters and Parke, 196A; Bandura, Grusec, and Menlove, 1967); and the character's race (cf. Greenberg, 1971; Greenberg and Hannenman, 1969). Children recognize and sort different portrayals using per- ceptual dimensions that may shift with age, sex, or any of several environmental influences. To young children, a TV character may be primarily funny; to an older child the same character may be active; to a male, strong; or to a female, physically attractive. The assumption has been that an understanding of these dimensions is an important prerequisite to accurately describing media effects. A criticism of this assumption is that perceptual dimensions defined a priori by the researCher may be inaccurate. Dimensions such as reality, sex, violence, race, etc., are forced upon children regardless of Whether they represent the primary evaluations that children make of television characters. Even if they do enhance prediction of media effects, there is still no way to insure the ac- curacy of their use in relation to other dimensions Which cannot be anticipated. Mass communication literature is void of studies Which attempt to empirically derive the dimensions children use to differentiate TV characters. Psychological research, however, does deal with the more general issue of person perception, including developmental differ- ences in the quantity, type and salience of attributes used to de— scribe other people. The research reported here will attempt to integrate the find- ings from research on person perception with children's perception of television characters. Although there may be perceptual dif- ferences between TV characters and people in real life, these studies deal with cognitive processes most likely generalizable to a variety of phenomenon. Specifically, the study will deal with three issues: 1) identi— fication of the underlying perceptual dimensions children use to differentiate TV characters; 2) a description of developmental and sex differences in the use of perceptual dimensions; and 3) the relation of these dimensions to Children's acceptance and application of TV dharacters' behaviors. General Considerations in Person Perception Person perception research recognizes that attributes of an in- dividual are more a function of the perceiver than of the person being perceived (Dornbusch, ct al., 1965). There are few inherent qual- ities of a person or television character that will be recognized in the same way across observing individuals. If this were not true, content analysis of character traits would adequately describe the prominent qualities that distinguish TV characters, regardless of whether the perceiver was male, female, child or adolescent. Furthermore, it is likely that one person's description of sev- eral people will be more consistent than several people's description of one person. This individual consistency implies that people use a uniform category scheme to describe others (HaStorf, Richardson, and Dornbusch, 1958). There appears to be no reason not to assume that children use consistent attributes in categorizing and describ- ing TV characters. It is the pattern of perception across individuals, however, that allows the researcher to make the most general use of results. Al- though individuals may be more consistent than aggregates of people, findings that apply to groups will have the most implications for understanding television's impact on children. Consistencies among aggregates are more descriptive of cultural processes and, therefore, are more useful in understanding the societal impact of a mass medium. To the extent that individuals conform to group perceptual processes, this understanding will be more accurate. One of the most important issues in person perception concerns the nature of an attribute or descriptive category. Some cognitive psychologists (e.g. Bruner, 1957; Gibson, 1963) suggest that cate- gories of description are nominal and qualitative.1 This simply 1These studies are reviewed by Wegner in a dissertation entitled "The Development and.Articulation of.Attributes in Person Perception," Department of Psychology, Michigan State University, l97h. means, for example, that if a Child thinks a TV character is funny, that character is categorized with other funny TV characters. Math— ematically, nominal categories of perception are best represented using set theory (Bieri, et al., 1966, cited in Wegner, 197M). A nominal view of perception does not require that categories or attributes be relevant to every person. Some attributes could be unique to subsets of people. For example, the category "serious" may not apply to any of a young child's favorite TV characters. A perceiver's response to a new person is defined by the other people in the category or "set." Past responses associated with other members of a category are attached to new people and old re- sponses are generalized to new stimuli. If a child perceives a TV Character to be funny, for example, responses to past characters per- ceived as funny will be applicable to the new portrayal. The relative "funniness" of the new character compared to others already in the category Should not alter the child's responses. It is not difficult, however, to imagine situations Where an attribute is applicable to different people as a continuous evalua- tion. This would be especially true for TV characters. The majority of portrayals represent a rather homogeneous group of people. All dramatic heroes, for example, may be good looking, but most children could probably further differentiate the characters in terms of their relative attractiveness. In this sense, people are assigned a value on some attribute dimension rather than merely being grouped with others haying similar attributes. It is this attribute value that correspondingly defines one person's response to another. This dimensional model of perception holds that people are as- signed a value along some attribute dimension and responses to people are based on that value. In contrast to a set theory representation of nominal attributes, the dimensional model is often represented by a geometric space (cf. Carrol and Wish, 197A; Woelfel and Burnett, 197A; Sherman, 1972). .Attributes are dimensions in a space, and objects, or in this case people, are points in the space. Psycho- logically, the dimensions are assumed to correspond to fundamental perceptual processes. The projection of a person on a dimension in- dicates the perceived value for that person on the attribute repre- sented by the dimension. The dimensional model of perception will be used in this study. This model is more consistent with available literature on children's perception of TV characters and it represents a more descriptive way of defining multidimensional perceptual processes. To accept a nominal classification of attributes would mean that no additional information is obtained by differentiating people within a given category. The relative "fUnniness" of two characters, for example, should be no more predictive of the Characters impact than the knowledge that they are both characterized as funny. There are, however, media studies which attribute differential effects of TV characters to different amounts of a perceived quality. For example, while several TV characters are perceived by children as real, dif- ferent amounts of perceived reality will better predict media effects than a dichotomous analysis of whether the quality "real" is present or absent (Greenberg and Reeves, 197A). By arraying TV characters in a multidimensional space, the characters' values on several of these attributes can be Simultaneously examined. Description of the continuous dimensions used by children to evaluate TV Characters is the primary emphasis of this research. Television characters may vary with respect to one or several under- lying perceptual dimensions. These dimensions may be restructured during development. They may individually or collectively predict which TV characters will have the greatest impact on Children's at- titudes and behavior. The following sections will discuss dimension content, developmental changes in the number and content of dimen- sions, and the relation of perceptual dimensions to television effects. Dimensions of Perception In this section, hypotheses will be formulated that project which dimensions children use to differentiate TV characters. While psy- chological research on person perception offers guidelines relevant to the process of dimensional development and Change, no literature has examined the dimensional content regarding perception of TV Characters. The only studies that deal with perception of TV char~ acters begin with one or occasionally several dimensions defined by the researCher as important. No attempt has yet been made to empir- ically derive a set of the dimensions from data on children's evalua- tions of TV characters. Two different types of research will be used to suggest percep- tual dimensions. Some studies explicitly ask children to volunteer attributes that describe TV characters. These studies are obviously the most helpful because they do not assume a perceptual process but rather attempt to define the process. Other researCh has tried to determine whether hypothesized per- ceptual dimensions will result in differences on some dependent measure. In these studies the research emphasis is usually on dif- ferences in behavior caused by a perceptual dimension rather than whether the dimension is actually used. If, by manipulating some attribute of a TV model, differences in attitude or behayior occur, then the dimension is assumed to exist. These assumptions have validity to the extent the dimensions indeed produce differences in behavior. However, the addition of other confounding influences in these studies makes their application to this research somewhat less direct. The dimensions hypothesized in this section are founded on both types of research. Even these studies, however, are few. Other di- mensions and rationales for their inclusion will be based on personal speculation and interviews with elementary school children. Two major types of dimensions will be hypothesized to emerge: 1) dimensions that describe the physical aspects of TV characters, and 2) dimensions that describe TV characters' behaviors. It would be unreasonable to expect that all of the hypothesized dimensions will be operative for large groups of children. Given the exploratory nature of this analysis, however, the concern is to identify a set of attributes whiCh could reasonably occur rather than predict the subset that will occur. This method provides 1) an oversampling of dimensions to maximize the probability that dimensions actually used by children are included in the research and 2) a rationale for why a dimension might have emerged if it is later found in the data. Four dimensions of physical description will be hypothesized. They are sex, age, physical strength and physical attractiveness. S25, Several studies, especially recent efforts, Show that children do differentiate TV characters on the basis of sex and that this distinction is related to the impact TV characters have on children's own perceptions. Miller and Reeves (1975) found that third through sixth grade children used sex to distinguish occupa- tional portrayals on TV. The distinction between females and.males in these roles pre- dicted Children's perception of how appropriate it was for a female to work in a traditionally male position and how many females the Child thought were actually working in the area. For both sexes of children the perception of appropriateness and the number of females in the occupation increased as exposure to the characters increased. Atkin and Miller (1975) found in an experimental setting that children who saw women portrayed in counterstereotypical occupations were more likely to endorse those occupations as appropriate for women than were Children Who did not see the portrayals. Beuf (197M) found that among 3-to—6-year-olds heavy television viewers hold more stereotypical perceptions of sex roles than do light viewers. Chil- dren are also more likely to imitate same sex models (Bandura, 1969), and to choose same sex TV characters as people they would most want to be like (Miller and Reeves, 1975). It is therefore hypothesized that: H1: Children will differentiate television char- acters on the basis of the Character's sex. Age. No studies have explicitly examined Whether children dis- tinguish old from.young TV characters or whether Children are more likely to be affected by characters from certain age groups. Psy- chological research on imitation, however, does Show that children will imitate real life models they perceive to be like themselves more than models who they think are different (Stotland, 1961; Bandura, 1969). Age should be an obvious candidate to contribute to observer-model similarity. ResearCh on observer-model similarity says that children either told.they have qualities in common with a.mode1 or who perceive they are similar to a model are more likely to imitate the model's be- havior than those Children who perceive no common characteristics. Stotland (1961) explains the increased imitation in terms of needs for cognitive consistency. If Children perceive a model as Similar to themselves, they are more likely to adopt other characteristics of that model because a model like themselves would only perform ac- ceptable behavior. It is hypothesized that: H2: Children will differentiate television char- acters on the basis of the character's per- ceived age. Physical strength. The inclusion of strength as a perceptual dimension is based on personal interviews with third through sixth graders about which TV character they most want to be like and why (Miller and Reeves, 1975). Physical strength was cited several times by males as a rationale for modeling, indicating at least a recogni- tion of variation in TV character's strength. 10 Considering the characters most frequently viewed by elementary school children, strength is a dimension which also has a large variance among portrayals. Differences in cartoon characters, sit- uation comedy roles, male vs. female characters, and dramatic heroes and heroines should increase the likelihood of a physical strength dimension. Thus: H3: Children will differentiate television char- acters on the basis of the character's per- ceived physical strength. It is less clear whether strength is more a physical than be- havioral attribute of television characters. For some portrayals strength could only be inferred from characters' behaviors. A character may have to perfbrm an act of strength before the attribute is applicable. As used here, strength is meant to be indicative of physical qualities such as height, weight, or other nonverbal cues which suggest superior physical abilities. Physical attractiveness. Just as males cited physical strength as a stereotypical evaluation of TV characters, females were likely to justify wanting to be like a favorite character because the person (usually a female) was attractive (Miller and Reeves, 1975). Similar to the strength dimension, it seems reasonable that children will dif- ferentiate characters on the basis of attributes that are highly re- inforced in their lives. Strength and attractiveness are attributes which early in life have positive referents and are qualities which children should notice in TV presentations that often maximize their importance : 11 HA: Children will differentiate television char- acters on the basis of the perceived physical attractiveness of the Character. Dimensions hypothesized to describe TV characters' behavior include the perceived reality of the behavior, hOW'funny and good the character is and.how muCh support the Character receives from other people on the same program, Perceived reality. Compared to other perceptual dimensions, the perceived reality of TV portrayals has been frequently researChed. Several studies show that TV characters are differentially perceived as being like people in real life (cf. Greenberg and Reeves, 197M; Lyle and Hoffman, 1971). It has also been shown that reality perceptions vary across dif- ferent subgroups. Children from.disadvantaged families believe TV is more true-to-life than children from middle class homes (Reeves, 197M; Greenberg and Gordon, 1971); blaCk children believe TV is more real than white Children (Greenberg, 1971); perceived reality of TV decreases for older Children (Lyle and.Hoffman, 1971; Greenberg and Reeves, 197k); and girls tend to perceive TV as more realistic than boys (Greenberg and.Dervin, 197M). Other studies have shown that perceived reality will predict sub- sequent behavior. Greenberg (197A) found.more aggressive behavioral intentions for children who thought TV was real. Feshbach (1971) ex- perimentally found a similar relationShip with children exposed to a "real" news story of the Viet Nam.war becoming more aggressive than I children exposed to a "fantasy' war movie. Hypothesis five is: 12 H5: Children will differentiate television char- acters on the basis of the perceived reality of the character's behavior. .Although this hypothesis is included with others that deal with a Character's behavior, it has still not been empirically determined whether perceptions of reality are based on character's actions, on other personality and physical attributes, or both. It is assumed here that Whatever qualities do influence perceived reality, they are consistent for any single Character. Hgmgg. There is little empirical evidence that a funny dimension exists. In the study asking Children to explain their choice of a favorite TV model (Miller and Reeves, 1975), some children did say they wanted to be funny. Funny was primarily a quality used by males. Despite the lack of researCh on TV characters and.humor, this dimension seems to be one of the more obvious attributes to include. Making children laugh is probably one of the surest ways of attract- ing their attention. One only need examine the TV shows containing the most advertising directed exclusively at children. .Audience rat- ings confirm the most popular shows for Children (i.e. Saturday morn- ing cartoons, afternoon and early evening situation comedies) are based on humor. Hypothesis six is: H6: Children will differentiate television char- acters on the basis of perceived humor. Goodness. The quality "good" in TV Characters has never been researChed nor is there any evidence that this attribute is given as a reason to model TV Characters. It will be included.here as a more abstract Characteristic that should correlate with other qual- ities generally understood by most children to have positive 13 connotations. For'example, of the attributes already discussed, good should be related to strength and physical attractiveness. The final categories that do correlate with this evaluation can be more ac- curately combined into a composite dimension if this abstract quality is included. Hypothesis seven is: H7: Children will differentiate television char- acters on the basis of‘how good the Character is perceived to be. Support from other characters. Three studies on the effects of televised violence indicate that characters supported fer their ac- tions are different from and imitated more than characters who are not. Bandura, Ross and Ross (1963) exposed four groups of nursery sChool children to filmed aggressive models. Children Who observed the ag- gressive model rewarded displayed more aggression in a play situa- tion than did children Who saw an aggressive model punished, an active but non-aggressive model with no reaction, or no exposure to a model. Those Who saw the aggressive model punished did not differ from.the remaining two groups. These findings substantiate that vicariously experiencing a reward will influence a model's impact on children. Two other studies (Walters and Parke, 196M; Walters, Leat, and Mezei, 1963) examined the effects of a filmed model on pre-school and kindergarten males' ability to resist temptation. In both ex- periments, Children Who witnessed disobedient models being rewarded aggressed.more than those Who saw the model being puniShed. Bandura, Grusec and Menlove (1976) applied similar experimental manipulations to the imitation of a model's prosocial behavior. 1h Social rewards dispensed in response to usually positively sanctioned behavior increased the likelihood of Children's imitation in mag- nitude similar to the aggression studies. These findings suggest that Children will discriminate between TV characters on the basis of other Character's reactions to their behavior. Hypothesis eight is: .H8: Children will differentiate television char- acters on the basis of the perceived support the character receives from other characters. In summary, eight perceptual dimensions were hypothesized. Four ‘were related to physical qualities (sex, age, physical strength, and physical attractiveness) and four related to TV character's behavior (perceived reality, humor, goodness, and support of the character's behavior by other Characters on the program). Certainly some dimensions will be more important than others. At this time, however, there is no basis for predicting the relative contribution of eaCh attribute to an overall dimensional structure. The research cited on eaCh attribute merely suggests that such a dimension might exist. It also must be remembered that most of the studies forced chil- dren to make judgments about characters along an attribute continuum, No attempt was made to empirically derive the dimensions from data whiCh did not name specific attributes. It could very well be that characters do differ on some of these qualities and that children recognize these differences, but that the differences are not opera- tive fer children in any systematic way. 15 Developmental Differences in Perceptual Dimensions Considerably more data is available on the process of dimen- sional change than on the content of perceptual dimensions. .Al- though none of these data are specifically related to perceptions of media content, changes in dimension structure should be applicable across different content areas. This section will deal with developmental changes in children's perception of television characters. Two hypotheses will be made about age-related changes in 1) the number of perceptual dimensions used to distinguish TV characters and 2) qualitative differences in the dimensions. One aspect of development which most psychologists agree on is that children generally progress from simple to complex modes of understanding. For learning theorists, this progression is dependent on experience with the environment and different patterns of rein- forcement (cf. Skinner, 1953; Bandura, 1969). For cognitive develop- mental theorists, changes in complexity of perception are linked to developmental stages whiCh define the upper bounds of understanding (Piaget, 1953; Bruner, 196M). Despite these fUndamental differences, however, both kinds of theories predict that children become more complex as they age. With this theoretical support, a prediction that the number of perceptual dimensions used by children will increase as age increases seems well founded. Studies examining this question, however, dis- agree depending on the methods used to test the hypothesis. The use of naturalistic descriptions of other people (Peevers and Secord, 16 1973) or checklists of descriptive traits (Yarrow and Campbell, 1963; Livesley and Bromley, 1973) show that the number of attributes used to describe people increase as children become older. These findings agree with the establiShed idea that complexity of person perception is associated with the use of a larger number of descriptive qualities (Peevers and Secord, 1973). A multidimensional study of person perception in Children, how- ever, did not find an increase in dimensions. The dimensional struc— ture of third, sixth and ninth graders was similar with respect to the number and nature of perceptual dimensions (Olshan, 1971). By using Kruskal's non-metric multidimensional scaling program.(Kruskal, 1968), the author showed that for all three age levels a two or three dimensional structure adequately represented a sample of 30 traits. Similarity between the traits was determined by having children sort them.into ten piles. A measure of trait co-occurence was then ob- tained for each possible pair of traits and these measures were input into the multidimensional scaling program, This contrary finding has been criticized because the method used to determine the number of dimensions (multidimensional scaling) is a teChnique designed to achieve maximum parsimony from a set of concepts (Peevers and Secord, 1973). The two methods for analyzing the number of operative perceptual dimensions, however, are possibly even more different than they first appear. Analyzing the mean num- ber of adjectives used by individual children to describe others is very different from scaling concepts in an n-dimensional geometric space. The first measure would be quite sensitive to shifts in the 17 use of descriptive words and it is possible that differences across age could be explained in terms of language development alone. Multidimensional scaling, however, would be less sensitive to age- related changes in vocabulary because the dimensions are determined by scaling similarity ratings of concepts. For structural changes to occur, children would have to differently judge two concepts in terms of their proximity to eaCh other. Although the multidimensional technique used by Olshan seems to be most representative of true developmental Shifts in the number of perceptual dimensions, this one study cannot negate the notion that dimensional complexity increases with age. This study does show, however, that for this research question, different methods have yielded opposing results. It will be hypothesized here that additional dimensions will emerge for older children. The available evidence on developmental Changes in cognitive complexity is too overWhelming at this time to support a null version of this expectation. Hypothesis nine is: H9: Older children will use more dimensions to differentiate television characters than young children. Another result suggested by increasing cognitive complexity is a Shift in the content of perceptual dimensions used by children. The pattern of these Changes generally has been described in terms of children's decreasing dependence on concrete physical cues (Wohlwill, 1962). As children develop they are more able to make inferences about the qualities of people and will rely less on phys- ical properties to define others. 18 A major theme of cognitive developmental theorists suggests that early cognition is tied directly to concrete events and later cogni- tion is more abstract, inferential, and generalized (Wegner, 197M). The stages of Piaget's theory of intellectual development (Piaget, 1953) follow this pattern. During the first stage, the sensorimotor period, the dominant activities are physical and motor probes into the child's immediate environment. During the second stage, children learn to symbolically represent concrete objects and to logically manipulate these representations. The final stage is devoted to propositional thinking Where children learn to make inferences about objects and people based on their experiences with them. Applying this developmental change to the perception of tele- vision Characters, older Children Should be more capable of making inferences about character's behavior than younger children. Further- more, younger children should be more dependent on attributes which are directly perceivable from the Character's portrayal. Differential reliance on the two groups of dimensions hypoth- esized in this study should test the validity of this developmental change. The dimensions which represent physical properties of char- acters (sex, age, physical strength, and physical attractiveness) should be used more by young children who rely on concrete physical objects to represent and react to reality. Perceptual dimensions which require an evaluation of a.TV char- acters' behavior and a subsequent inferential judgment about the character should be used.more by older children. These character attributes (perceived reality, humor, goodness, and support received 19 from other Characters) require children to observe a behavior and then, on the basis of the observation, evaluate the character on a relevant attribute dimension. The specific hypothesis is: H10: Younger children will be more dependent on perceptual dimensions describing physical properties of TV Characters (i.e. sex, age physical strength, physical attractiveness) than older children Who will depend more on dimensions describing TV characters' be- havior (i.e. perceived reality, humor, good- ness, and support received from other Characters). These hypotheses do not identify specific ages. The expecta- tion is that these Changes will be continuous developments from the time Children begin watChing television consistently (about age 2-3) to early adolescence (approximately age 13). Tests of these hypoth- eses would ideally sample children in this entire age range. How- ever, the most fundamental information processing changes should occur for school age Children (ages 5 to 12). These ages roughly correspond to the second and third stages in Piaget's conception of intellectual development. _It should also be noted that age is used only as an approxima- tion of development. It is entirely possible that same-age children would interpret TV information differently due to different develop- mental experiences. Cognitive stages are not universally applicable to certain age groups and Should not be confused with biological maturation. Connections between age and stage of cognitive develop- nmmt reported by various researChers are only guidelines reflecting the development of a particular sample under study. 20 Using age rather than level of cognitive development has, how- ever, two advantages. First, a reliable measure of cognitive de- velopment has not yet been developed. Very elaborate measurement techniques have yielded inconsistent results. Also, different techniques hate disagreed depending on whether self-report or be- ‘havioral measures were obtained. Second, age differences are more directly applicable to policy decisions. It would be much easier, for example, to implement an in-school media program for third graders than for children at a certain stage of development. Sex Differences in Perceptual Dimensions As mentioned earlier, there is evidence that males and females are different in terms of the saliency of some perceptual dimensions (e.g. physical strength and attractiveness). There is, however, very little literature on person perception and on sex differences in gen- eral that would suggest profound sex differences in the perception of TV characters. One of the few sex differences that is agreed upon by most psyChologists and.has been documented in several studies is that males are more aggressive than females.1 In media research, this finding is common in the studies on television and aggression (cf. McLeod,.Atkin and Chaffee, 1971). This Would lead to the prediction that perceptual dimensions related to aggression would be more im- portant for males than females. From among those dimensions For an extensive review of this literature, see Maccoby and Jacklin, The Psychology of Sex Differences, 1975. 21 hypothesized in this study, physical strength should be more ime portant for males than females. Sex differences in the other hypothesized dimensions do not have empirical support from past studies. Even the obvious di- mensions such as physical attractiveness (when operationalized as "good looking") have not consistently shown sex differences (Maccoby and Jacklin, 197k) . No formal hypotheses will be made concerning sex differences in the perception of TV characters. There is too little theoretical or empirical evidence to predict sex differences in the process of di- mension changes at different ages. While there is some basis for predicting differences in the con- tent of dimensions, these hypotheses would have to assume that the particular dimensions will emerge. As has already been mentioned, it is not known what dimensions will be used by children to differen- tiate TV characters. The dimensions suggested represent reasonable expectations, not empirically confirmed results. Differences in the dimensional structure of males and.females will be analyzed, however, because perceptions of TV characters have never been examined in the present manner. Some aspects of TV pro- gramming make sex differences a likely result. For example, the ratio of male to female characters on prime time TV is 3 to 1 (Miller and Reeves, 1975). The fact that females' perceptions are mostly of males, and.males' perceptions are mostly of same sex Characters could account for different dimension structures. The anaLysis, therefore, will test for these possible differences. 22 Perceptual Dimensions as Predictors of Media Effects What do perceptual dimensions tell us about the impact TV characters have on children's attitudes and behavior? Interest in the description of these dimensions is, of course, based on an ex- pectation that they will be useful in understanding which TV char- acters will have the greatest affect for which children. The general expectation is that the method used to categorize a TV character will determine a child‘s reaction to the character and will be influential in the decision to apply the character's behavior to real life situa- tions. A central part of all cognitive theories deals with the effect of mental representation of stimuli on behavioral response (Bruner, 1969). Representation is not merely a memory process responsible for the re- trieval of usable experience. The information processing formats by which experience is categorized for later use are just as important in determining behavior as associating past experiences with the present. If television characters are differentiated on the basis of funniness, for example, then humor should be an important referent for applying the Characters' behavior to real life. The information processing methods used to catalog character's behavior define the most salient aspects of the behavior when it is later retrieved. If this reasoning applies to perceptions of television portrayals, TV characters would be just as funny or not funny when they are first seen as when they are later remembered. The types of situations that characters' behaviors are relevant to, therefore, are determined by 23 the dimensions originally used to evaluate their behavior on the screen. The hypothesis is: H12: The dimensions children use to distinguish television characters will be related to Children's imitation of TV character's be- havior in real life situations. Hypothesis 12 says that perceptual dimensions will be related to children's imitation of TV characters' behavior. Prior research on children and.imitation offers several conceptual and operational def- initions of imitation and numerous other concepts that subsume es- sentially the same phenomena. Learning by vicarious experience has been labeled imitation, observational learning, copying, social facilitation, vicarious learning, contagion, identification and role- playing. Distinctions are made between learning on the basis of types of responses, antecedent variables controlling matching responses, fidelity of imitation, generality of learning, and whether matching responses occur in the presence of a model (Bandura, 1965). Several researchers (e.g. Gerwirtz and Stingle, 1968; Flanders, 1968; Bandura, 1965) question the validity of these arbitrary con- ceptual differences because essentially the same learning process is operative regardless of content, generality of what is learned, or the models responsible for the original behavior. These authors argue that suCh distinctions are counter-productive and should be incorporated into a single, more simple conception of behavioral alteration. In considering the mediating effects of perceptual di- mensions on Children's behavior, imitation will refer to behavior modifications resulting from exposure to modeling stimuli (Bandura, 1965). 2h One other note should be made about this hypothesis. The word- ing implies that ell perceptual dimensions will predict children's imitation or positive inclination to imitate TV characters' behavior. There is actually no basis for expecting that all of the dimensions ‘will fulfill this expectation, nor is there evidence to predict the relative association of each dimension with the utilization of TV characters' behavior. The hypothesis is only meant to suggest that among the dimensions used to distinguish TV characters, one or some will predict the impact of TV portrayals on real life situations. Summagy Three sets of hypotheses were made about the perceptual dimen- sions children use to differentiate television characters. The pri- mary purpose of the research is to describe the continuous dimensions that are used. Hypotheses were made for four dimensions of physical attributes (sex, age, physical strength, and physical attractiveness) and four dimensions of TV characters' behavior (perceived reality, humor, goodness, and support received from other characters). Two predictions were made about the differences in dimensional structure for children at different ages. It was hypothesized that 1) older children would use more dimensions to distinguish TV char- acters than younger children and 2) younger children would rely on dimensions describing physical attributes more than older children Who would rely more on dimensions describing TV behavior. Finally, it was hypothesized that the perceptual dimensions used to differentiate TV characters would be used by children to determine which Characters to imitate in real life situations. CHAPTER II METHODS Data for this researCh were collected in several different for- mats. The methods used to collect and relate these data will be pre- sented in this chapter. Before the specific methods are discussed, however, an attempt will be made to first illustrate how the dif- ferent results will be related to eaCh other. The description and validation of perceptual dimensions used by children to distinguish TV characters will involve relationships among three independently collected sets of data. One set of data will be children's similarity judgments between all possible pairs of a sample of TV characters. The pairs of characters will be rated by third, fifth and seventh graders on a five point scale ranging from very similar to very different. This type of measure has the advantage of not suggesting which perceptual dimensions to use. These "proximity" measures will then be used to locate the TV Characters in an n-dimensional geometric space which represents a conceptual map of the perceived psyChological distances between char- acters. Each concept or character will have a coordinate value for eaCh dimension and eaCh dimension will have some relation to how the group differentiates TV characters. This analysis will be accomp- liShed using multidimensional scaling. Another set of data consisted of nine univariate measures of dimensions hypothesized to emerge from.the multidimensional space. 25 26 As these dimensions will be conceptually identifiable,1 their cor- relation with distances in the multidimensional space will assist in the interpretation of the dimensions in that space. Although this analysis will be described later in more detail, it is useful now to understand exactly What is being correlated. For example, if a sample of 15 TV characters were rated by children on a five point scale measuring how funny characters were, these ratings could be averaged across children to obtain a mean "funny" score for each of the TV characters. If, for the same characters, children were asked to judge how similar each possible pair of characters was, the characters could also be arrayed in an n-dimensional space based on how far apart the children perceived the characters to be. For each dimension in the space, all of the Characters would have a coordinate score. If the mean scores for each TV character on the unidimensional scale funny correlated highly with the coordinate scores for dimension 1, then it could be argued that the Spatial dimension represented the attri- bute humor. An analysis similar to this example will be performed for each dimension hypothesized to emerge in the multidimensional space. The relative correlation of each dimension with each unidimensional at- tribute will suggest the conceptual make-up of the dimension. The Lauestions will ask, for example, How funny do you think Fred Flintstone is? Very Funny Not very Not funny funny funny at all 27 multiple correlation of all of the dimensions with each unidimensional attribute will indicate the extent to which the attributes can be pre- dicted from the dimensions that emerge from the space. This type of analysis does, however, pose at least two problems. First, the data relevant to perceptions of TV characters is collapsed across several individuals. It is possible that the aggregation of several Children's perceptions will result in values that are unrep- resentative of individual children within the group. This would be especially true if subgroups of children exist that differ a great deal in their perceptions. If, for example, males use only the di- H mension "strengt to differentiate TV characters and females use only "attractiveness," an aggregation of the two sexes would yield two dimensions - strength and attractiveness. The two-dimensional solution, however, would not be representative of either sex. This problem is solved to the extent that subgroups of children with unique perceptions can be anticipated. Grade and sex analyses of TV character perceptions are an attempt to account for subgroup differences. It must be remembered, however, that other subgroups may exist which were not included. .A second problem involves the relationship of the unidimensional measures with the spatial dimensions. To the degree that the at- tributes used as unidimensional measures are pep representative of the dimensions that emerge in the Space, they will not be useful in the interpretation of the space. The tentative basis for the hy- pothesized perceptual dimensions makes this outcome even more prob- able. 28 Given this possibility, a third set of data was collected. As mentioned earlier, in hypothesizing which dimensions will emerge, an attempt was made to include all the dimensions Which ggpld reason- ably be used by children to distinguish TV characters rather than predict exactly which ones gill occur. Interviews were conducted with children to determine if there are dimensions of character dif- ferences which were not included. These interviews also helped confirm the validity of the dimensions already hypothesized. A discussion of the methods used to collect and analyze these three sets of data follows. The discussion will progress in the same sequence that the data were collected. The first section reports on the pre-test interviews conducted to probe for additional unidimen- sional attributes. Second, the similarity judgments and multidimen- sional scaling will be presented followed by a discussion of the unidimensional measures. .A final section will deal with the measures of children's use of TV Characters' behavior. Individual Interviews Individual interviews were conducted with elementary school chil- dren to determine if other dimensions than those postulated in Chapter I should be included as unidimensional measures. Third, fourth, and fifth grade students (N=98) from an elementary school in Mason, Michigan were interviewed during regular classes on two consecutive days in May, 1975. The interviews were conducted by fifteen under- graduate students at Michigan State university. Interviewers first asked each child to name two favorite TV char- acters and two characters they did not like. The six questions that 29 followed asked them to describe the differences between each possible combination of characters that were named. The specific question was, "How is character A different from.character B?" This question forced comparisons between liked and disliked Characters, maximizing the po— tential for different dimensions of comparison to occur. Every difference description given by the children was written down by the interviewers. Each comparison also included a probe that asked "Is there anything else?" in reference to additional differ— ences between the characters (see Appendix.A for a complete copy of the interview schedule used). In analyzing children's responses, an attempt was made to de- fine perceptual dimensions that were not hypothesized in Chapter I. Each response was examined to determine if it was similar to any of the previously hypothesized dimensions. .A separate count was made of those attributes not included.among the previous set. One coder analyzed all of the responses. Several of the attributes named were not in the set of original- ly hypothesized dimensions. Most of these, however, were named by less than three children. These qualities were usually unique to a particular character in the comparisons. Attributes named by more than three children that were not covered by the original dimensions are listed below in order of frequency of occurence (numbers refer to frequency of mention): Does a lot of different things (it) Moves fast (9) Ownership of property (6) Different methods of killing or using violence (5) Leadership (5) Differences in occupation (h) Differences in ability to perform a special behavior (h) 3O Including all of these additional attributes would have been difficult. Assuming a sample of 1h TV Characters, to measure each character on the eight attributes already hypothesized would have required asking each child 112 questions. Additional attributes in- crease the number of questions by 1h. Given the large number of responses about Characters "doing a lot of different things" and "moving fast," it was decided to add a dimension called activity. None of the other responses from the open-ended questions were included due to the number of additional questions that would have been necessary. Detailed age and sex breakdowns of these responses were not made. The purpose of this pre-test was only to see if additional dimensions should be added to the study. The final group of attributes to be used now includes four dimensions of physical description (sex, age, physical strength and physical attractiveness) and five dimensions that describe TV characters' behavior (perceived reality, humor, goodness, support received from other characters and activity). GroupVInterviews At this point the final two sets of data -- similarity judg- ments of TV characters for the multidimensional scaling and ratings of the Characters on the unidimensional.measures -- were collected. The sample consisted of two third, fifth, and seventh grade classes from an elementary and middle school in Haslett, Michigan (total N=2lO). There was an approximately equal number in each grade and sex (exact breakdowns will be given separately for each set of data). 31 Separate questionnaires were used for the multidimensional and unidimensional data. For the third and fifth graders, questionnaires were group administered in two‘Sessions during regular class hours. Each administration lasted about 30 minutes. For the seventh graders, both questionnaires were given at the same time. The one administration lasted MO minutes. All data col- lection was monitored by graduate students with considerable exper- ienCe in collecting data from children. Multidimensional Data and.Analysis The primary methodological concern in this study was not to sug- gest dimensions with the questions used to obtain the data. If you ask a child if character A is funny, you force an evaluation of the character on the attribute funny even if that quality is not a pri- mary dimension used to differentiate TV Characters. Multidimensional scaling is in most cases based only on the per- ceived similarity between concepts. According to Helm, Messick and Tucker (1959): The fundamental concept in multidimensional scaling is psyChological distance, Which is usually estimated in terms of judgments of similarity among stimuli; i.e., two stimuli judged to be very similar are considered to be psychological- ly closer than two stimuli judged to be very different. Given judgments of similarity among all stimuli in a set, nethemetical models exist Which provide an interpretation of these psychological distances in terms of Euclidean geometry. The stimuli are treated as points in a Euclidean space, and analytical techniques are available to obtain the dimension- ality of the space as well as stimulus scale values deter— mined within a rotation and translation. The procedure is analogous to physical distances(as, for ex- ample, between cities). If it were known how far apart in miles 32 each city in a state were from each other city, there would be only one way to geographically display the distances. The space would correspond perfectly to a geographical map of the area. This is exactly What was done with a sample of television char- acters. Each child made proximity or similarity judgments between all possible pairs of TV characters and the resulting similarity matrix was converted into a multidimensional space. This was ac- complished using INDSCAL, a multidimensional scaling program. The following sections will describe: 1) the sample of TV characters used; 2) how the similarity judgments were obtained; and 3) the specifics of the INDSCAL program. Sample gilTV_characters. Fourteen TV Characters were chosen for the analysis. They were selected on three criteria: 1) maximization of children's awareness of the Characters; 2) representativeness of the sample in relation to other TV characters seen by children; and 3) maximization of variance on the hypothesized dimensions. Since the children used in the survey spanned a considerable age range, only Characters from programs beginning before 9:00 p.mt were con- sidered to further increase familiarity with the entire sample. The characters Chosen were: Laura on "Little House on the Prairie" Mary Tyler Moore on "The Mary Tyler Moore Show" Reed on "Adam-l2" Fred sanford on "Sanford and Son" Fred Flintstone on "The Flintstones" Gilligan on "Gilligan's Island" Samantha.on "Bewitched" Hawkeye on "M*A*S*H" ArChic Bunker on "All in the Family" Chico on "Chico and the Man" Steve.Austin on "The Six Million Dollar Man" Fat Albert on "Fat Albert and the Cosby Kids" John-Boy Walton on "The Waltons" Marshal Dillon on "Gunsmoke" I 33 A brief description of the characters is in Appendix B. The maximum.number of characters chosen was mostly governed by the number of paired similarity comparisons that are required for a given number of concepts (defined by n(n-l)/2). With four- teen characters the number of judgments is 91. It was believed that the addition of any more characters would have made the questionnaire burdensome, especially for third graders. Also, this sample seems to adequately represent the available pool of recognizable char- acters. Neilsen audience ratings indicated the programs these char- acters were taken from were among the most popular children's shows. Only a few additional shows received the same audience percentage and these were mostly other Saturday morning programs. It is important to use popular Characters to minimize the amount of missing data in the similarity ratings. For all grades, only three characters were not known to all of the children; Laura, Chico and Reed. For these characters, there was less than 2% miss- ing data. Choosing Characters that maximized variance on the hypothesized perceptual dimensions was more difficult. It is quite possible that in the multidimensional analysis different samples of TV characters could result in different dimensional structures. If the sample was all male characters, for example, a sex dimension would not emerge; if all were policemen, a funny dimension would probably not be ap- parent, etc. 3h The best argument for the representativeness of the sample is the fact that all rejected candidates are similar to characters al- ready included. For example, Steve Austin, Reed and Marshall Dillon represent most all of the police/detective Characters; Chico, Archie Bunker, Mary Tyler Mbore and Fred Sanford represent the situation comedy characters, etc. Nevertheless, it should be remembered, especially for the multidimensional data, that the dimensions that do emerge are definitely a function of the sample of characters used. Similarity judgments. Judgments were obtained for the 91 pos- sible pairs of TV characters to determine how "far apart" characters were perceived to be. The specific question used to Obtain the judgment was: "What do you think about Fred.Flintstone and Gilligan?" Are they: very alike I'm.not different very muCh sure different alike This format was repeated for all of the character pairs. The judg- ments were coded from 1 to 5, with the higher number representing a greater perceived difference between the characters (see Appendix C for a complete copy of the questionnaire). With the completion of this questionnaire, a.matrix of simil- arity scores was computed for eaCh child. Individual subject's similarity matrices averaged across subjects in a particular sub- group (e.g. each of the three grade levels) served as input for the multidimensional scaling program, 35 In the administration of this questionnaire, each item was read aloud for the third graders. The fifth and seventh graders were allowed to proceed at their own pace after the instructions were read as a group. Completed questionnaires were obtained for 202 children. The grade and sex breakdown is as follows: Third Fifth Seventh grade grade grade Total Males 36 33 26 95 Females 31 33 1+3 107 Total 67 66 69 Overall, children judged TV characters to be more different than they were similar. The mean similarity ratings for all subgroups were well above the midpoint of the scale (scale ranged from 1 to 5, with 5 representing the greatest perceived difference). Characters were most different for younger children and females. Averaging across Children in eaCh age group and sex gave the following mean similarity scores and average standard deviations for the 91 paired comparisons: Third Fifth S eventh Row grade grade grade means Males i=3.68 £3.58 £3.50 $3.59 ’ s.d.=l.l6 s.d.=l.ll s.d.=l.03 s.d.=1.10 Females $3.99 £3.86 £3.79 £3.86 s.d.=l.05 s.d.=1.07 s.d.=0.97 s.d.=l.02 Column $3.82 $3.72 36:3 .68 Means s.d.=1.lO s.d.=l.09 s.d.=0.99 The complete distance matrix for eaCh grade by sex subgroup is in Appendix D. 36 A reliability estimate for the similarity ratings was obtained by repeating three paired comparisons throughout the questionnaire. The two responses to eaCh question were correlated across subjects. The average zero-order correlations for these three measures are shown below: Male/ Third Fifth Seventh female grade grade grade average Males .50 .71 .57 .61 Females .37 .73 .65 .63 Grade .h8 .73 .63 average Overall .62 average These reliability estimates indicate a rather unusual pattern. The similarity comparisons were most consistent for the fifth graders followed by the seventh and third graders. This is counter to the general notion that reliability estimates for data obtained from children will increase with the age of the children. The values for the third graders in particular are low enough to cause some concern. These estimates, however, are based on the consistency of individuals and not groups. All of the analyses in this study used values aggregated across several children. Although estimates of the group reliabilities are not possible because the children were not interviewed at two different times, the group values would likely be more encouraging. The INDSCAL program. Multidimensional scaling in general is a technique for arraying concepts in an n-dimensional space. The 37 output is typically a single configuration of points, one point for each stimulus, in one or more dimensions. The typical input to multidimensional programs is an n x n matrix whose cells are sim- ilarity measures aggregated across subjects (of. Shepard, 1962; Kruskal, 196M; Woelfel and Barnett, 197M). Distance between con- cepts in these models is computed using the standard formula for ordinary Euclidean distances (Torgeson, 1958): k=£(yjt- ykt)2 where djk is the distance between stimulus j and k, and yjt and ykt are the coordinates of stimulus j and k on dimension t. The INDSCAL model (for INdividual Differences SCALing) uses as input an m.x n x n (subjects x concepts x concepts) matrix of sim- ilarity scores (Carroll and Wish, 1971+). This means that for each subject (or in this study, subgroup) a separate concept by concept similarity matrix is input. The output for INDSCAL consists of two matrices as opposed to one for the n x n matrix input. All of the subgroup similarity matrices are first combined into a group space and a coordinate value is computed for all concepts on each dimension. tA second matrix defines subgroup weights for eaCh dimension based on the similarity ratings for each subgroup. An example of the two types of output are shown in Figures 1 and 2. Figure la shows nine concepts (e.g. TV characters) arrayed in a two dimensional space. This space was determined by combining similarity matrices from six different subgroups. 38 DIM 2 DIM 2 1 A B C DIM 1 __.D E: F 3 6 G H I h DIM 1 Figure la Figure 2 Total Group Concept Space Space of Salience Scores (Hypothetical) for Six Subgroups (Hypothetical) DIM 2 DIM 2 A B C I DIM 1 D: E 4F A B C I I DIMl D l G W I G T I Figure lb Concept Space for Subgroup 2 Figure 1c (Hypothetical) Concept Space for Subgroup 3 (Hypothetical) Figures taken from J. Douglas Carroll and.Myron Wish, "Models and Methods for Three4Way'Multidimensional Scaling," in Contemporany Developments in.Mathematical PsyChology, 197M. 39 Figure 2 Shows the second form of output, subgroup salience scores. The points in this space represent six subgroups (e.g. three grades by two sexes). The weights for each dimension indicate the extent to which each subgroup uses the dimension in distinguish- ing the concepts in the sample. These weights, called salience scores, are approximately equal to the proportion of variance accounted for in the similarity score matrix for eaCh particular subgroup. Examples of the spaces for subgroups 2 and 3 are shown in Figures 1b and lo. They show how the group concept space is re- structured to maximally conform to the subgroup space.l Subgroup 2 has a high value for dimension 2 and a low score for dimension 1. Consequently, the space for that subgroup is stretched on dimension 2 to indicate the higher salience of that dimension (Figure lb). The reverse is true for subgroup 3 (Figure 1c). The INDSCAL model assumes then that all subjects use the same dimensions but that there are subgroup or individual differences in the extent to which each dimension is used. It is theoretically pos- sible, however, for a subgroup to use only one dimension from a multidimensional solution. Subgroup 1 in Figure 2, for example, has a high salience score for dimension 2 and a score of zero for dimension 1. Mathematically, the use of salience scores adds a weighting factor to the computation of ordinary Euclidean distances. The 1Dimensions in the group concept space are actually stretched and shrunk until distances in the space correlate maximally with the similarity ratings for each subgroup in the analysis. ho formula for distance in INDSCAL is: f' 2 da'k = r21 wit (Vjt " 3%) Where wit is the weighted.metric fer subgroup i. Using INDSCAL to analyze the similarity ratings for 1h TV characters will yield two spaces. First, each TV Character will be arrayed in an n-dimensional space and coordinate values for the character will be computed for each dimension. Second, the six subgroups (three grades by two sexes) will be arrayed in an n- dimensional space where the coordinate values represent the extent to which the subgroup uses each particular dimension to distinguiSh TV characters. Uhidimensional measures. To validate and describe the dimen- sions that emerged in the multidimensional space, unidimensional rat- ings were obtained for eaCh character on the nine hypothesized perceptual dimensions. These questions were on a separate question- naire that was administered the day after the similarity ratings for the third and fifth graders, and.immediately after the completion of the first questionnaire for the seventh graders. This questionnaire was administered second so the attributes would not suggest dimensions that should be used for the similarity judgments. Seven third and fifth grade students were not present for the second day of the study which lowered the total sample filling out both questionnaires to 195. There was one question for each of the eight hypothesized per- ceptual dimensions. Each child rated all 1h Characters on all of hl l the attributes. The specific questions which appeared in a constant order were: How How How How How How How How funny do you think character A is? active do you think character A is? good looking do you think character A is? strong do you think character A is? much like a real person is character A? good do you.think character A is? old do you think character A is? much do the other people on (name of show) like character A? There were four possible responses for each question ranging from.very much of the attribute to none of the attribute (e.g. very funny/funny/not very funny/not funny at all; very active/active/not very active/not active at all). Responses were coded from h to l, with the higher number meaning the attribute was maximally applicable. A complete copy of the quesionnaire is in Appendix E. For the six subgroups to be analyzed, the means and standard deviations on these unidimensional measures were as follows: 1Children were not asked to identify the character's sex, which was the ninth hypothesized attribute. A2 Third Fifth Seventh Males Females grade grade pgrade Age £2.38 £2.32 i=2.3u i=2.uo i=2. 31 s.d.=.55 s.d.=055 s.d.=.5h s.d.=.56 Sod-=055 Good looking ié2.7h i;2.63 ie2.75 i;2.78 ié2.53 s.d.=.hh s.d.-.66 s.d.=.h5 s.d.=.56 s.d.=.63 + Strength X=2.75 is2.7l ie2.76 ié2.79 ie2.ou s.d.=.56 s.d.—.5O s.d.=.56 s.d.=.50 s.d.=.52 Funny ie2.80 ié2.80 ie2.89 ie2.86 i%2.65 s.d.=.63 s.d.=.55 s.d.=.65 s.d.=.59 s.d.=.57 Active ié3.l7 ie3.07 ié3.15 ie3.2o £53.00 s.d.=.29 s.d.=.32 s.d.=.22 s.d.=.29 s.d.=.38 Good ie3.23 ié3.16 ie3.25 ié3.28 ié3.07 s.d.=.2O s.d.=.2O s.d.=.2l s.d.=.17 s.d.=.23 Real ié2.9l ie2.87 ie2.98 §;2,92 ie2.77 s.d.=.h6 s.d.=.h5 s.d.=.h8 s.d.=.h3 s.d.=.h6 Support ié3.53 ié3.58 ie3.56 ié3.57 ie3.5h s.d.=.16 s.d.=.l6 s.d.=.l7 s.d.=.l6 s.d.=.l7 All of the mean ratings exceeded the midpoint of the scale ex- cept those for age. The ratings for good looking, strength, funny, and reality generally fell between 2.50 and 3.00. Those attributes perceived to exist most in the 1h TV characters (mean ratings above 3.00) were active, good, and support received from other characters. The mean scores for each character on the eight attributes are in Appendix.F. Media effects variables. Two questions were used to measure the extent to which Children would imitate TV characters' behavior. The questions were: M3 How much do you want to be like character A? a lot a little not very not at much all Are there things that character A does that you would like to do? a lot of some almost nothing things things nothing at all These questions are self-report measures of anticipated be- havior. Responses to these items are valid to the extent that children are able to indicate their desire to be like and "do" like TV Characters and to the extent that these reports are actually translated into behaviors. Both questions are considered operationalizations of the same dependent variable, imitation. Although there are subtle differ- ences in the two items they both deal with reports of behavior mod- ifications associated with exposure to television characters. These questions were included in the second questionnaire and followed those for the nine attributes. The means and standard de- viations for these measures across the 1M TV characters were as follows: Third Fifth Seventh Males Females , grade.» grade grade . Want to i=2.l+l £2.12 £2.32 $2.111 112.06 be like s.d.=.53 s.d.=.59 s.d.=.h8 s.d.=.h0 s.d.=.h0 Want to Sir—2.70 x=2.26 i=2.l+9 112.72 i=2.22 do like Sodo=oh3 Sodo=osl+ Sod-=0LI'LI' Sodo=ohl Soda-oh]- 141+ Data.Analy§is The similarity judgments and unidimensional ratings were col- lapsed across subjects to give each TV character a value or values for each particular set of data. First, the similarity judgments were input into INDSCAL which resulted in a dimension coordinate for eaCh TV character on n dimensions. Second, the unidimensional rat- ings were averaged across subjects so that eaCh character was as- signed a mean score for a particular attribute. It is the comparison of these values for each TV character that comprised most of the data analysis. Most of the hypotheses were tested using correlational tech- niques. To test the extent to which the spatial dimensions can be defined as hypothesized, the dimension coordinates for each char- acter were correlated with the unidimensional means for each char- acter. The multiple correlation of the dimension coordinates with each unidimensional attribute indicated the extent to which each at- tribute was represented in the multidimensional space. The extent to which the dimensions predicted the media use variables was also tested by correlating dimension coordinates with the mean ratings on the two media effects variables. Hypotheses concerning subgroup differences in dimension content and structure were tested using the salience scores from.the INDSCAL output. These scores were also obtained by collapsing data acr as subjects; however, the salience scores represent a characteristic of the subgroup and not an attribute of any specific TV character. AS It should be noted that for all correlations the sample size for evaluating the significance of the coefficient is 1h (equal to the number of TV Characters). By most survey researCh standards, this sample size would be inadequate. It should be remembered, however, that the values for eaCh character are not based on a single observa- tion. Since all of the values for TV Characters were collapsed across subjects, the minimum number of ratings used to comprise the value is never less than the size of the smallest subgroup (n=26 fer seventh grade males) and is usually considerably larger. Reported significance levels should be regarded as conservative estimates. CHAPTER III RESULTS The results will be presented in four sections that generally conform to the format in which the hypotheses were originally dis- cussed. The sections are on 1) the dimensionality of the multi- dimensional space; 2) the content of the dimensions; 3) subgroup differences in dimensions; and 1+) the relationship between the di- mensions and children's use of TV character's behavior. Dimensionality of INDSCAL Solutions Although nine separate attributes were hypothesized as potential dimensions in the multidimensional solution, the exact number of di- mensions that would emerge was not specified. There was no theory or data which suggested how many dimensions children would use to dif— ferentiate TV characters. Age differences in the relative number of dimensions that would emerge were hypothesized. The specific hypothesis was: H9: Older children will use more dimensions to differentiate television characters than younger children. The number of dimensions that should be retained for further analysis was determined by comparing the proportion of variance ac- counted for in the original similarity data by adding more dimensions in the INDSCAL solution. While the overall proportion of variance will always increase with the addition of dimensions, higher dimensions may provide only a negligible improvement in the goodness of fit measure. 1+6 M7 The decision to stop adding dimensions is probably just as muCh a subjective judgment as a statistical one. .A three dimensional solution, for example, may add only 3% additional explained variance, but the dimension may be readily interpretable or it may be ime portant in explaining media effects. Table 1 shows the amount of variance accounted for by one, two, three, four and five dimension solutions. Figure 3 represents a graph of the change in explained variance for the different solutions. For the total group and all three age groups, the addition of the fifth dimension adds a negligible amount of variance. More than two digits would be needed to show the increase in the overall correla- tion. This eliminates from consideration anything greater than a four dimensional structure. A two dimensional solution for the entire group accounts for 7M% of the variance in the original data. .A third dimension increases the percentage to 82% and a fourth dimension adds another 5% result- ing in a total of 87%. ‘While the third and fourth dimension do not add a substantial amount of explained variance, they will be re- tained at this point in the analysis to determine if they are either identifiable or predictive of the media effects variables. The table also shows that no substantial differences occur among the dimensionality of the three age group spaces. While the amount of variance explained by the same dimensional solution in- creases slightly with age, (probably because the more reliable var- iance is in the older groups), the percentage change across solutions is almost the same. Addition of dimension 3 and h for the third #8 .ooogm opm>fiaa m.@sonw o£p_aoam oo>fiaoo mpodpoam Madmen to XHApda_m use xaapoanhpfinsaflawm m.g§omwnom onp_aoam eo>waop mpoopoam meadow mo Rename a somzpon ma aboawosm a mom oon> Hoeowpwaoahoo oxen am. mm. am. mm. mm. om. as. mm. mm. as. moose dance om. mm. om.. mm. mm. mm. ow. mm. on. :m. compo spoo>om mm. mm. mm. so. mm. Hm. me. mm. mm. as. moose cream mm. Hm. mm. Hm. ,ma. am. mm. mm. msw‘ we. momma cease mm m m mm a m mm m m mm m m mm a m nowesaom Ga moowmomEHm go amoeba mAmoomaw sumo>om pom .QPMfim eohflne one mSoaw Hmpoev msoaesaom assoamcosam psoaoeean new: mamaass¢.q.uo ommpsooaom H manna Percenta- of Varies; Accounte for 100 90 80 Percentage 70 of Variance l L 1+9 .’”o—o~oq o-—' O .00000" ..0000 '“OOOOOOCCO J Total group Third grade Fifth grade Seventh grade Accounted for 60 50 to Figure 3 l 2 3 h 5 Number of Dimensions in Solution Percentage of Variance.Accounted for by INDSCAL Analysis With Different Dimensional Solutionsa a‘The correlational value for a subgroup is between a matrix of scalar products derived from.the subgroup's similarity matrix and a matrix of scalar products derived from the group's pri— vate space. 50 grade increased the overall correlation by .Oh for each addition. For the fifth grade the increases were .05 and .03; and for the seventh grade .03 for each addition. Hypothesis nine is, therefore, not supported. Dimension Content This section will present analyses that attempted to describe the four dimensions that were retained. Essentially, these analyses relate children's perception of TV characters' qualities along uni- dimensional attributes to spatial configurations of the same group of characters. Nine attributes were hypothesized as reasonable ex- pectations of the content of these dimensions (sex, age, attractive- ness, strength, reality, goodness, support from other Characters, and activity). Figures h and 5 Show the fourteen TV characters plotted in four dimensions. Figure h represents dimensions one and.two and Figure 5 represents dimensions three and four. .All possible two dimensional comparisons in the four dimensional solution are in Appendix G. The actual coordinate values for eaCh character on eaCh of the four dimensions are tabled in.Appendix H. The INDSCAL program does not compute totally orthogonal di- mensions. The dimensions in Figures h and 5 are drawn as orth- ogonal for clarity of presentation. Correlations among the four INDSCAL dimensions are Shown in Table 2. In many cases, the interpretation of a single dimension will involve more than one unidimensional attribute. If more than one attribute has a high relationship with a single INDSCAL dimension, 51 . Steve Austin Reed. 0 Marshal Dillon DIM 2 +,50 ‘ +.25 * O Hawkeye O Chico . Fred Sanford I O Archie Bunker I "050 -025 O . Fat Gilligan Albert Fred Flint stone -.25 " -050 - Figure 1!. John-Bo . Y 0 Samantha Mary Tyler Moore 0 . Laura +.50 DIMl Dimensions 1 and 2 of the Four Dimensional Group Concept Space From an INDSCAL Analysis of Data on Perceived Similarities Among Fourteen Television Charactersa a The INDSCAL program does not compute totally orthogonal dimensions. The dimensions in Figure 2 are drawn as orthogonal for clarity of presentation. The correlation between dimension 1 and dimension 2 is .02. 52 DIM 1+ +.50 " Gilligan ' John-Boy 0 +25 ” .Laura Chico . .Hawkeye Samantha 0 Reed 0 Steve Austin DIM 3 I | 1 l -050 -025 +025 +.5O Marshal O Dillon . Mary Tyler Moore 0 Fred Sanford Fat Fred -.25 " Albert" Flintstone . Archie Bunker -50 '- Figure 5 Dimension 3 and ’4 of the Four Dimensional Group Concept Space From an INDSCAL Analysis of Data on Perceived Similarities Among Fourteen Television Characters a The INDSCAL program does not compute totally orthogonal dimensions. The dimensions in Figure 3 are drawn as orthogonal for clarity of presentation. The correlation between dimension 3 and dimension 1+ is .214. 53 Table 2 Zero Order Correlations Among Four INDSCAL Dimensionsa DIM l DIM 2 DIM 3 DIM 1+- D114 1 - DIM 2 .02 - DIM 3 —.3o .09 - DIM A -.hl -.26 .2A - a‘I‘hese correlations are based on the coordinate values for fourteen TV characters on each of four dimensions. The n for the correlations is 11+, the number of TV characters. 5h the attribute should be correlated in a magnitude and direction that is consistent with the regression analysis. Table 3 shows the zero order correlations among the nine unidimensional attributes for the entire sample. Similar matrices for each grade and sex are in .Appendix I. Description of the four dimensions was guided by results from multiple regression analysis. These results are presented in Table h. The columns in the table represent zero order correlations and standardized regression weights for predicting mean ratings of fourteen TV characters on the nine hypothesized unidimensional scales. The multiple correlations in the righthand column shOW'how well the mean ratings can be predicted using all four of the INDSCAL dimen- sions as independent variables. The correlations and regression analysis are based on an N of 1h, the number of TV Characters. Each character has a coordinate value (based on the INDSCAL analysis) and a value on the unidi- mensional attributes (based on the mean rating given the character by subjects). These two numbers are the X and Y values in the correlations. Geometrically, the regression coefficients represent direction cosines between a vector (corresponding to the unidimensional measure) in the four dimensional space. Projections of TV characters on the vector correlate maximally with mean ratings of the Char- acters on the unidimensional scales. Because the number of variables in the multiple regression equa- tions (5) approaehcs the sample size used to compute the correlations 55 Table 3 Zero Order Correlations For Entire Sample Among Nine Unidimensional Ratings of Fourteen Television Charactersa‘ 1 Funny - 2 Active -.u2 - 3 Good looking -.35 .8h**- h Strength —.50 .56* .h5 - 5 Reality -.19 .21 .26 -.oo - 6 Good -.u1 .89**.92**.h6 .h6 - 7 Age .15 -.3o -.25 .12 -.oh -.3h - 8 Support -.7h* .36 .hh .32 .29 .56*-.6o* - 9 Sex (of character) -.18 -.ou -.35 -.39 .oo .19 -.38 .AS - *p<.05 **p<.OOl aThese correlations are based on the mean rating of each of fourteen TV characters on nine attributes. The n for the correlations is therefore fourteen, the number of TV characters. The mean value of each attribute for each character is, however, collapsed across sub- jects. The attribute value for each character is then based on an n of 202. ' 56 .momumom .mg «mood a.onH .hoaflz «weapmfiQSpm Hoowmoaonohmm “mofimzoz coded comp .msowpmamaaoo map aw modam> w was x map one mucosa: 039 omega .Asoae uawao mo mamadm Deanne esp hp nopomaseo one so>ww weapon some can so domsnv mopdnfiapps HsSOfimooaHofles one so oddw> m use Amfimhamss HmomQZH one so oommpv QOfimsoaHe zoom pom moHs> opmqaoaooo s we: hopomamno zoom .mnopossmao >9 mo hogan: one .za mo o as no women was mwmhassm soammoamoa use mQOfipsHoSAOC ones Hoo.va** mo.vm* om. mm. **wm. *wm.u ma.+ **mm.+ *bm.+ **>>.n *mm.n *mm.+ o:.+ onom mm. an. *Jw. Ho.+ hm.+ Ho.l :m.+ 00.: 00.: *fiw.+ **:®.+ pHOQQSm om. as. me. am.- mm.- ma.+ mo.+ mm.+ os.+ sm.- mm.- omm om. mm. **mm. *om.+ *me.+ amm.+ *se.+ *mm.+ om.+ mm.+ *oo.+ eooo ma.- om. ms. om.+ mm.+ mo.+ mm.+ mm.+ sH.+ mo.+ mm.+ seafloor mu. :m. **mm. mo.+ mo.u Ho.u mH.+ **mw.+ **mm.+ *o:.+ mm.+ newsmhpm be. mm. **Hm. sm.+ *sm.+ **mo.+ **ma.+ Hm.+ ea.+ mH.+ *mm.+ monsooa ooou so. as. *sw. *mm.+ *ob.+ ma.+ m:.+ *b:.+ mm.+ mm.+ *sm.+ o>fipo< mm. mm. **mm. HH.+ Hm.- aa.+ sH.- mH.- ma.- *amo.a- **mm.- asssm pmm mm m spam a seem a seem a seem a Ammo : sz m 2Hm m ZHQ H EHQ -acaacm sou ecpooanUV sagsoaw Hopoev moowmsoaan Q.+ **mm.+ 3pm ma. ms. 00. 00.- mH.+ om.- 00.. mo.- 40.- H>.+ mo.+ cam psoaasm mo. mm. 00. mm.n m:.u ma.+ m0.+ mm.+ mm.+ :m.| mm.u nub 0m. 3:. >0. Nm.n. 0m.u HH.+ 00.: mm.+ m:.+ :N.n mm.n 2pm mm. mm. mu. m:.- *sm.- ma.+ :o.+ sm.+ Hm.+ :m.- am.- cam mm< on. :m. **mm. mo.u *H>.+ *mm.+ *mw.+ *om.+ 0H.+ ma.+ *wm.+ spa m2. ow. *wa. *Hm.u *:0.+ :m.+ >:.+ ma.+ mo.+ 0m.+ *mm.+ 3pm m0. :5. *0w. o:.+ m:.+ *0:.+ *m0.+ *::.+ mm.+ :m.+ mm.+ cam doom 0H.- mm. m:. mm.+ o:.+ ma.+ mm.+ aa.+ mo.+ oo.+ am.+ no» oa.- mm. m:. o:.+ mm.+ mo.+ :m.+ mm.+ mH.+ oo.+ mm.+ rpm om.- ma. 0:. om.+ mm.+ mo.+ ma.+ mm.+ mH.+ mo.+ mm.+ cam assasmm we. mm. **Hm. mo.+ mo.+ mo.+ om.+ **ow.+ **mu.+ *H:.+ m:.+ mom as. am. *om. mo.- ma.- 0H.- :o.+ **Hm.+ **ma.+ m:.+ mm.+ how we. mm. **Hm. uo.+ mo.u Ho.+ Hm.+ **>m.+ **:m.+ om.+ Hm.+ cam npmsmspm 0w. mm. **mm. *mm.+ *mm.+ **mw.+ **0m.+ mm.+ ba.+ 0m.+ *mm.+ 59% as. ow. *mm. *m:.+ *om.+ *oo.+ *w>.+ HH.+ mo.+ mH.+ mm.+ mom mqflxooa m0. :5. *mw. 0m.+ o:.+ *om.+ *m>.+ mm.+ om.+ ma.+ >:.+ cam @000 on. ma. *mw. **Ha.+ *wa.+ ma.+ m:.+ *mm.+ :H.+ :m.+ mm.+ now we. mm. **Hm. **Hs.+ *Hs.+ ma.+ m:.+ *s:.+ mm.+ ma.+ *mm.+ mom mw. 0:. mo. Hm.+ :m.+ m0.+ mm.+ *00.+ mm.+ mH.+ Hm.+ cam m>flpo< 0m. mm. **mm. Hm.+ ma.n mm.+ 00.: >0.a 50.: **mo.Hn**mw.n Spa mm. 0m. *imm. m0.+ 0m.n mH.+ ma.n *mm.n mm.u **mm.u **mw.u 3pm mm. om. **mm. mo.+ Hm.- \uo.+ sm.- mH.- ‘ImH.- **mm.- **mm.- cam amend pmm mm m mpmm a spam a mpmm a dpmm a Amws : SHQ m sz m qu H 2H9 -xsssnm sad mpfl>spo< mmmsm>sposspo< npmqmspm nossm vmpomaaoov omflofifimm mafiadomsz cophommdmpb sAmesso assm>mm .mssso_esflna .mesno spasmvmqosmqmssm geomQZH ssom.soso mmssnssps4.asaonmsmssessp co waprm one: wqapofldmam pom mvgwflmz.GOfimmmawmm_Umedadvndpm can mQOfipmeaaoo hmoao oamN m 36nt 63a domfiom .aa .89 32m .333 .moflmflspm floswoaofiama Jessi/62 saga mmmp .mQOHpmHmasoo map afl medas> N can x amp was mamassa 039 omega .Amowaw zone an ampomamzo one am>flw wqflpna Gems esp no dmmnnv mopsnflpppn assowmnmaficflqs mnp no msam> a dam AmfimanQd q<0mQZH may no Ummmnv seamsmaflw nose mom msas> oprHUaOOU a men ampowamno Seem .msmpowamno >5 mo Hogans map .2H 00 a as no woman was mamaanos scammmawoa was mnoapsaoaaoo mafia 6h attractiveness and the weights for dimension 3 regressed on the at- tribute good looking are .60 for third and fifth graders and .62 for the seventh grade. The lack of grade differences on dimension 3 is not supportive of hypothesis 10 Which predicted physical di- ‘mensions to be more operative for younger children. Dimension h, activity, was the second dimension descriptive of character's behavior. 'While the regression weights for dimension h were identical for fifth and seventh graders (.71), the weight for the third grade was smaller and insignificant (.31). In fact, the attribute "activity" is better predicted by dimension 2 for the youngest children. Dimension h is, therefore, also supportive of hypothesis 10. .Activity is a behavioral dimension and is more ob- viously present in the spatial configuration of the older children. The results for at least three of the dimensions (1, 2, and 1+) support the hypothesis that younger children will use physical di- mensions and older children will rely more on behavioral dimensions. The salience scores for each subgroup provided by the INDSCAL analysis offer another method for testing this hypothesis. The sali- ence scores represent the use or dominance of each dimension for all subgroups.l If the dimensions are used differently by each grade level, this Should be reflected in the salience scores. Table 6 shows the salience scores fbr three grades on the four INDSCAL dimensions. The changes in the scores generally conform to the results of the regression analysis. For the first behavioral IlThe salience scores are approximately equal to the proportion of variance accounted for in the similarity matrix for each subgroup. Table 6 65 Salience Scores for Three.Age Groups From an INDSCAL.Analysis of Data on Similarities Among Fourteen Television Characters Unsupported Masculine Feminine Activity Humor Strength Attractiveness DIM 1 DIM 2 DIM 3 DIM h Total Group .596 .372 .3h0 .326 (n=202) Third Grade .h36 .h56 .MOO .387 (n=67) Fifth Grade .692 .372 .325 .320 (n=66) Seventh Grade .72h .289 .295 .272 (n=69) 66 dimension (1), the salience scores increase with age (from..h36 to .72h) demonstrating dimensions 1 is more important in the space for older children. The reverse pattern is present for dimension 2. Masculine strength is more important for younger children (.h56) than older children (.289). Differences in salience scores for dimension 3, feminine at— tractiveness, show grade differences in use of the dimension that were not indicated by the regression analysis. The salience scores, unlike the regression weights, decrease with age (from .hOO to .295) which is supportive of hypothesis 10. The scores for dimension h, activity, change about the same amount as those for dimension 3 (.387 to .272). These changes, however, are not predicted by hypothesis 10 and run counter to the regression results which indicated this dimension predicted activity better for older children. The salience scores for each grade are plotted in Figure h. The coordinates in these plots represent the salience values. These plots graphically show that as grade increases, dimension 1 becomes more im- portant and dimensions 2, 3, and h are less important. Despite some inconsistent comparisons between the two methods, there is good support for accepting hypothesis 10. For the two di- mensions which account for the most variance in the subgroup spaces (unsupported humor and.masculine strength), grade differences most ob- viously support differential emphasis on behavioral vs. physical attributes. 67 DIM 2 (Strength) THIRD GRADE " o FIFTH GRADE . SEVENTH GRADE 0 l i L , ,4 DIM l (Unsupported Humor) DIM u (Activity) ' THIRD GRADE 0 O FIFTH GRADE . SEVENTH GRADE 1 1 1 DIM 3 (Attractiveness) Figure 6 a Subgroup Spaces for Three Grades on Four INDSCAL Dimensions aThe coordinate values for the three grades on each dimension are equal to the salience score on that dimension. 68 Furthermore, the two methods of analysis support the hypothesis in different ways. The regression analysis examined grade differ- ences in values associated with only the individual attributes that were best predicted by each dimension. The analysis of salience scores, however, looked at entire dimensions irrespective of the unidimensional attributes which loaded on the dimension. Actually, large grade differences in the regression analysis would make the salience scores uninterpretable because the latter are computed by relating each grade to the overall group. If the dimensions were conceptualized differently for each grade, the sal- ience scores would be meaningless because they refer to the same dimension across grades, regardless of differences in content. While there were grade differences in the prediction of attributes from the dimensions, the separate regression analyses did not suggest that dimension content changed with age. Two other results from Table 5 are important. While age and perceived reality were not significantly predicted for any age group, the multiple correlation for age almost reached significance for the third graders (.73, p<.10). Secondly, while age differences in the description and impor- tance of some dimensions did occur, the spatial configurations for the three grades were generally very similar. Table 7 shows the results of spatial comparisons using canonical correlation 69 . l . . analySIS. For all poss1ble comparisons of the separate four dimen- sional spaces, four significant canonical variates were obtained. Sex Differences in Dimension Content Although no specific hypotheses were made about sex differences in dimension content, reasons were given why some male/female dis- crepancies could be expected. The most general comment on the spaces for the two sexes, however, is that despite subtle differ- ences, the dimensional structures are very similar. .A canonical I correlation analysis similar to the one between grade levels resulted in four significant canonical correlations (.99, .98, .9h, .87; p<.OOl for all four correlations) between the four-dimensional spaces for males and females. Sex differences in dimension content were found for: l) predic- tion of the attributes active and good, and 2) the magnitude of re- gression weights for masculine strength and feminine attractiveness. Results for the regression analysis by sex are in Table 8. While the attribute "active" was predicted by dimension h for both males and females, the regression weight for dimension 2 was even greater (.71, p<.OOl), but only for males. The regression weight for females was only .20 (n.s.). lCanonical correlation analysis (Cooley and Lohnes, 1962) is a tech- nique used to test the relationship between two sets of variables (in this case two sets of coordinate values derived from two dif- ferent grades). A linear combination of variables is found in eaCh set of variables that maximizes the correlation between the created factors or canonical variates. .A new linear combination is then found which best accounts for the residual relationship between the two sets. The analysis finds as many canonical vari- ates as there are variables in the smaller set of the two (Fink and Walker, 1975) . Table 7 70 Canonical Correlations of’Four INDSCAL Dimensions For Three Separate Age Levels (Third Grade, Fifth Grade, and Seventh Grade)a Canonical Degrees Variate Canonical of Number Correlation 3L2 Freedom pjs Third grade 1 .985 88.03 16’ .001 with 2 .972 5h .28 9 . 001 Fifth grade 3 .916 26.60 h .001 h .787 9.18 l .002 Third grade 1 .991 92.1h 16 .001 with 2 .963 53 .011 9 . 001 Seventh grade 3 .907 27.9h h .001 h .837 11.05 1 .001 Fifth grade 1 .991 108.10 16 .001 with 2 .985 69.65 9 .001 Seventh grade 3 .957 35.8h h .001 h .851 12.28, 1 .001 a Separate INDSCAL solutions were obtained for each subgroup. The values in the canonical analysis are the coordinate values for each of 1h TV characters calculated separately for eaCh grade level. 71 H00.v9** mo.va* o e o 0' e o e 0' 0' O o m m mm. mm. ”mm. mm... mm.“ mm.” all...“ m... m... new... nan sea phommom om. 13.8. m. - mm: ma... mo.+ mm.+ 91+ mm: am: massed 0m. :z. 00. m. - mm.- H.+ 00.+ :m.+ 0:. + mm.- :m.- canWw¢ om. $.18. Imm.+ *3... $00... imp; 00.+ 8i 5... * m.+ madam.“ mm. 13.100. K .+ *mm.+ so... mm; *wm.+ .E.+ amt. * m.+ mawwoo ma; mm. 3. mm... o .+ om.+ mm.+ d...+ mo.+ mo.+ 0m.+ mamas-H ma: mm. 3. am... a .+ mo.- ma.+ mmr. .3... 91+ em... mass hpaammm me. 8.10m. 00.+ mo.+ 0H. + emf. ii.+ *i..+ *mm.+ Sf. magma Hm. mm. **mm. m0.+ mo.- ma. - H0.+ **Hm.+ **0®.+ * .+ pm.+ mama newcoapm ow. $.10? *mm.+ *mm.+ immJ tan-m... SJ :o.+ ma.+ 139+ madam-H so. . *5. *m:.+ m:.+ 1.0:; it... 113% mm... S.+ we; mass woaxooa @000 2.. * *m .+ *m .+ mm.+ *.m + o .+ mm.+ ma! mm.+ A madam mu. mm. *6 . *Hw.+ a .+ 12.- H. + *3 .+ * .+ om.+ me... mass 333. mm. mm. **0m. 00.+ H.- *mm.+ m0.- *mm.- 0m.- **m0.H-**mm.- marshy mm. mm. i8. ma... w;- 8... mm: ma: 3.- i8. H- - mass mason 9mm mm m team a mpom a seem a seem a flees 11 En m 23 m Em H Ea nxowanm ~00 hpfi>wpo< mmooobwpooapp< npwqoapm hoabm .gSmoD ompooaaoov sAmoaoawm one moadzv moOHmooaflm A<0mnzH Adom seam wmnspfiupp¢ HsoOflmomaflofioD no mwoapsm c602 onpOHomam Mom mpaww03.oowmm0aw0m ooNfloadoodpm one anewesamnaoo hooao chem m manna 71a sow-mom .dd .89“ 28H .333 .83meer Heeamoaoneamd Jess/El stage. meme .aowpmaoasoo one QH meoan> M one x one can wannabe 030 whose .Axom 3060 an sweetness map ombflm wsfipea mama can no oomnnv woponfiappo Hecoflmooaflowqd map so osHs> a can Amfimhanos q<0mQZH one no ommspv mommomaflo £960 mom 05Hs> opoqadaooo n was severance zoom .maoposasSO >9 mo 90985: on» .:H mo a on no woman one mfimszsd QOflmmmawmn one moowpnaoaaoo ogBd 72 This difference shows that for boys activity was part of the "masculine strength" dimension. The correlations between the uni- dimensional attributes calculated separately for each sex also con- firm.this finding. For males, r=.73 (P<.05), and for females, r=.h3 (n.s.). In the total group analysis, the attribute good was related to three of the four dimensions. The male/female breakdown, however, shows that good is predicted best by dimensions 2 and h for the males and dimension 3 and h for females (Table 8). Furthermore, for the two dimensions which weight differently on good, one sex weights heavily on the dimension and the other not at all. For dimension 2 the regression coefficient for good is .56 (p<.05) for males and .06 (n.s.) for females. The pattern for dimension 3 is reversed. The weight for females is .60 (p<.001) and for males .07 (n.s.). This indicates that good is a part of masculine strength for males and a part of feminine attractiveness for females. For the remaining attributes, significant weights were found simultaneously for males and females, although the magnitude of the weights was occasionally different. The weight for predicting strength from dimension 2 was greater for males (.91, p<.001) than for females (.71, p<.001). The weight for dimension 3 on good look- ing was greater for females (.68, p<.001) than for males (.h6, p<.05). Finally, dimension I predicted support better for males (.9h, p<.001) than for females (.68, p<.05). 73 In summary, the following results were found in the subgroup analyses: 1) 2) 3) 1+) 5) Older children tended to use dimensions which described TV character's behavior more than young children who used dimensions descriptive of character's physical attributes. Activity was part of dimension 2 for males but not for females. Males weighted support and strength more than females, and females weighted attractiveness more than males. These differences, however, were for weights that were significant for both sexes. Good was related to masculine strength for males and to attractiveness for females. Despite the differences cited above, the spatial configurations for the subgroups were generally very similar as determined by canonical correla- tion analysis. Perceptual Dimensions as Predictors of Media Effects This final section will show hOW well the four INDSCAL dimen- sions were able to predict whether children wanted to be like or do like characters on television. The analyses are similar to those previously discussed. Each of the two dependent measures was sep- arately predicted using multiple regression. There are two specific versions of hypothesis 11 being tested: Hlla: The dimensions children use to distinguish tele- vision characters will be related to how much children want to be like TV characters. Hllb: The dimensions children use to distinguish tele- vision characters will be related to how much children want to do things that TV characters do. No specific hypotheses were made about which dimensions would be the best predictors or about subgroup differences in the predictions. 75 .mom-mom .da .momH ..eeH aean .eeHemnpepm HsenmoHoeeawa .nesmzez sense meme .mnOHunaonnOo map nw monas> w_onn N 039 one mnonann 039 omone .Aoannam annoy can an nonconono one no>flm mnflpnn name 039 no oomnpv moabmwnn> mpoomao aflooa map mo some now onae> a one Amfimmaonn gnomQZH one no oomopv noemnmaoo some now onan> onwnflonooo a men nopowanno gnaw .mnoponnono >9 mo nonann 039 «:H mo n no no oomsp one mflmsznw neammoawon one mnoapoHoHAOO ones Hoo.va** mo.vn* me. om. *ms. m:.+ *Ho.+ as.+ *Ho.+ mo.+ mo.- mo.- mm.- eaaH oe op ens: me. am. **Hm. *Hs.+ Hm.+ **om.+ *ms.+ *mm.+ mm.+ 6H.- om.- ennH we won one; 9mm mm m seem a seem a seem a deem H Aewe s znn m an w znn H 2H9 nxnflnnm now hpfi>flpo< mmono>fiponapp¢ newnonpm noabm oopoonnooV onwnonom onHHSOmnE oophomgnmnb 039 no mwnfipnm snnnono anoav wnOHmnoaHQ anomQZH Adom.aonm moapmfinm> mpoommm nfiooz new: mnfipoflooam pom mpnmfioz,nowmmoamom_ooNHonsonnpm one mnowpoaohaoo noono onoN m manna... 76 The most profound differences were between the sexes (Table 10). While the multiple correlations for predicting "want to be like" for males and females were almost the same (.90 for males, p<.OOl; .91 for females, p<.001), they are dependent on entirely different di- mensions. For the males, "want to be like" is predicted by dimen- sion 2, masculine strength (Beta=.89, p<.001), and dimension ’4, activity (Beta=.56, p<.05). Dimension 3, attractiveness, was the only significant predictor for the females (Beta=.7’4, p<.001). The multiple correlations for the variable "want to do like" were not equally as high for males and females (.7’4 for males, p<.09; .89 for females, p<.05); however, the same dimensions were operative by sex as for the first dependent measure. Strength (Beta=.59, p<.05) and activity (Beta=.65, p<.05) predicted "want to be like" for males, and attractiveness (Beta=.62, p<.05) pre- dicted for the females. Grade differences in predicting "want to be like" were less striking (Table 11). Three changes occurred from third to fifth grades. First, the overall multiple correlation increased from .86 to .88 to .92. Second, dimension h, activity, did not predict whether third graders want to be like TV characters (Beta=.28, n.s.). Dimension 1+ was, however, a significant predictor for both fifth and seventh grades. Finally, the magnitude of the regression weights for dimension 3, attractiveness, varied greatly with age. Attractiveness was most predictive for seventh graders (Beta=.63, p<.001) and least predictive for fifth graders (Beta=.39, p<.05). Third graders 77 .eom-mom .on .memH ..oon .aean..aeneenneem HeoneoHonoaae .nesezez geese memo .enoweeaennoo enp nw menae> w.one x esp ene mnenann 03p whens .Amodenwm one meaen_hp nepoeneno on» ne>ww wnfipen news can no oemenv meanewne> own ewomn en» mo noes new enHe> e one Amfimhaene q<0mQZH esp no omeepv nOHmnenHo noes now enHe> onenflonooo e men nepoeneno Anew .mnepoeneno >9 mo neonnn map and mo n ne no oomep one mamhaene nowmmenwen one encapeaennoo enae Hoo.va** mo.vaa on. me. *me. me.+ *en.+ *me.+ *ma.+ ne.- sm.- mH.+ Hm.+ eeHesme mm. mm. a». *m0.+ 0:.+ m0.+ mw.+ *mm.+ m:.+ HH.- mH.+ meaea mafia oo ou.pne3 0p. :m. **Hm. 00.+ ::.+ **:>.+ **mw.+ HN.- NH.- pm.+ *mm.+ meaeaem me. He 100. 109+ mm.+ oo.- mH.+ $.me 135+ mo: mH.+ amass weed mp op pnez pmm mm m epom n epem n epem n epem n Aces : znn m znn e znn H znn nxnfinnm now zpe>wpo< mmono>apoenpp< newnmnpm noanm oopoennoov mnwnaaem wnHHnomez _oepnomnnmnb one no emnfipem enmeaenem one meaezv mnOHmnenfin geomQZH Anch.aonm moanefine> m poo new: mnfipofloenm now mpnwfle3.n0Hmmenwom ooufioneonepm one encapeaenno new eeooz 0 noono oan OH mHneB 78 now-mom do .memH teen .0an .moflafieom HecneoHonoamm Jeanna: none memo .mnoapeamnnoo enp na menHe> M.one x can ene mnonnnn 03p emena .Aooenw seem an nepoenene enp nw>Hw wnaeen neon enp no oomepv moanefine> own_edoma any we noee mom onae> e one AmfimaHene AdUmQZH may no ooeepv noamnonflo noes pom onHe> onenwonooo e we: nepqeneno noem .mnepoeneno >9 mo nonann on» .:H mo n ne no oeeep one mwmhaene nonmeenwen one mnoaeeaennOo enBe Hoover; moves S. tr. *5. *mn.+ $3.1. in; *ms... mo... oo.+ :o.+ ma... negate 0:. m0. *mw. *H0.+ *wm.+. 0m.+ *zm.+ NH.+ 00.- :0.+ o:.+ apnea mm. we. om. ma; mm.+ wm.+ mm.+ 01+ 00.+ HA.+ 9.1+ onus“ mead oo on pnez om. 0m. imm. *oa.+ mm.+ creme... tow; *Hm.+ we! mH.+ ma.+ 5.55m we. we. *3. *mm.+ *8; *mm.+ *m0.+ *mm.+ we}. om.+ mm.+ nfiam H0. me. *8. mm? mm.+ *mm.+ *E..+ *m:.+ mm! mH.+ m:.+ oann. efiH so 8. one: pmm Nm m epom n epem n epom n epom n Gee :an m as man Henn unnannm now hpo>flpo< mmene>flpoenpp¢ npmnmnpm nonnm ompoonHOOV mnflnonmm mnfldnomez oopnonnnmnb eAeoenw negate Jesse ans-HE 635 2125 83895.9 iomnzn noon seen negated, 383m eses: can. no mwnfleem_neoz wanOHoonm non mpnwfio3.n0Hmmenwom oeufioneonepm one encepeaonnoo neono onoN HH magma 79 ‘were between the two, although closer to the seventh grade (Beta=.55, P<-05)- Grade differences in predicting "want to do like" were more ob- vious. None of the dimensions predicted this variable fer the third graders (R=.69, n.s.). The multiple correlation increased to .79 (p<.05) for the fifth graders and to .87 (p<.05) for seventh grade. Only one dimension, activity, predicted."want to do like" for the fifth grade (Beta=.6l, p<.05) while two dimensions, attractiveness and activity, were significant predictors for the seventh grade (Beta=.57 for attractiveness, p<.05; Beta=.52 for activity, p<.05). To summarize, these are the results from using four dimensions to predict hOW'much children want to be like and.do like a sample of f0urteen TV characters: 1) For the total sample of children, the four dimen- sions yielded significant multiple correlations for predicting both dependent variables. 2) The strength and activity dimensions predicted the dependent measures best for males and attractive- ness predicted best for females. 3) For third graders only, the activity dimension did not predict "want to be like," and none of the dimensions predicted "want to do like." h) For all.males, the multiple correlations for "want to be like" were insignificant, while the same cor- relations for all females were significant. Dimen- sion 3, attractiveness, was the best predictor of this variable for the females. CHAPTER IV SUMMARY AND DISCUSSION The purpose of this study was to examine the perceptual dimen- sions used by children to differentiate television characters. The utility of these dimensions in predicting media effects and sub- group differences in dimension content were also examined. All past research in this area.had used dimensions defined by the researcher as important in understanding television's impact on children. The methods used to define the dimensions without suggest- ing their content were, therefore, very important in this study. Multidimensional scaling of IA TV characters was accomplished by asking third, fifth, and seventh graders to judge whether all pose sible pairs of the sample of characters were alike or different. This method.had the advantage of letting children differentiate among the characters by whatever attributes they chose to think about. These results were then compared with unidimensional evaluations of characters on attributes defined by pre-test and past research. Four interpretable dimensions were found from.the multidimen- sional analysis. In decreasing order of variance explained in the final solution they were: unsupported.humor, masculine strength, feminine attractiveness, and activity. An hypothesis that older children would use more dimensions to differentiate TV characters than younger children was not supported. One hypothesis concerning developmental differences in dimen- sion content was supported. Younger children were more reliant on dimensions descriptive of physical attributes (strength and 80 81 attractiveness) and older children depended.more on the dimensions descriptive of TV character's behavior (unsupported.humor and activity). Although no hypotheses about sex differences in dimension con- tent were made, two sex differences were found. Males weighted support and strength more than females, and females weighted at- tractiveness more than males. "Good" was related to strength for males and to attractiveness for females. For the total sample of children, the four dimensions pre- dicted whether they wanted to be like or "do" like television Characters. The strength and activity dimensions predicted these measures best fer the males and attractiveness predicted best for the females. Discussion The results of the study will be discussed generally in the same order as the presentation of results. Comments about the methods used in the study will be discussed first, followed by dimension content, subgroup spatial differences, and prediction of media effects variables with the perceptual dimensions. The relationship of two independently collected sets of data ‘was the basis for most of the results. Children were asked to rate the similarity of all possible pairs of 1% TV characters and to judge the same characters on eight specific attributes.1 The 1Sex of characters was not a judgment made by subjects. It was the ninth attribute. This attribute was entered by the researCher with the assumption that Children could tell the difference between males and females on TV. 82 coordinate values for the TV characters, as determined by multi- dimensional scaling, were then correlated with the mean ratings for each character on the unidimensional attributes. Generally, these correlations were very high. Of the nine unidimensional attributes, the multiple correlation of four dimen- sions on each attribute was greater than .90 in five cases. Two other multiple correlations were greater than .80 and two were in— significant (age and perceived reality). Given the frequent concern about survey measurement with chil- dren, these results are very encouraging. The reliability of the similarity judgments was not much greater than .50, yet the dimen- sions from the INDSCAL analysis explained over 80% of the variance in the unidimensional measures in the majority of cases. Ideally, the two methods should have produced isomorphic re- sults. Each operation was attempting to measure which dimensions children use to differentiate TV characters. It is, therefore, im- portant to note that the high multiple correlations do not rep- resent a theoretically independent set of variables predicting a dependent variable. The correlations indicate the relationship in a multi-method approach to measurement. These results suggest added confidence in simple, Likert type measurement with children. There is an important instance, however, when this confidence can be misleading. If it would have been pos- sible to know-a priori which dimensions would emerge in the multi- dimensional analysis, the entire study could have been based on univariate measurement. This, however, was not the case. At least 83 two of the hypothesized unidimensional attributes (perceived reality and age) were not highly predictable fran the dimensions in the space. To the extent that dimensions are known, unidimensional measure- ment may be within a range of precision that would make more compli- cated multidimensional analysis unnecessary. Even with a sample of only 1’4 TV characters, 91 similarity comparisons were required of each child. If the dimensions are not known in advance, methods such as the multidimensional technique used in this research seem imperative. Media researchers may have already allocated considerable effort to exploring perceptual dimensions which possibly do not exist or are at best only minimally operative. Dimensionality of Subgroup Spaces Before dimension content was examined, the three grade levels were analyzed to determine if the same number of dimensions existed for each age group. Based on a considerable past literature on de- velOpmental differences in the dimensionality of person perception, it was hypothesized that older children would use more dimensions to distinguish TV characters than young children. There was no evidence , however, to smart this expectation. This is only the second known study to report no age differences in the dimensions children use to evaluate other people. It is also only the second study to employ multidimensional scaling. Olshan (1971) reported no dimensionality differences between children in 8h a similar age range. Although her research dealt with person percep- tion in general and not with TV characters, the same number of di- mensions adequately represented similarity data for third, sixth, and ninth graders. This multidimensional analysis and the one by Olshan at least suggest that changes in dimensionality need to be verified with more rigorous methods than adjective check lists. Clearly, similar re- search must be done with yomger children, however, methods that rely on judgments dependent as much on vocabulary skills as true cognitive processing changes may be misleading. Similarity in the concept spaces for children at three different age groups is not very supportive of extensive developmental shifts in the dimensionality of cognitive processes. There was, however, a limitation of this study that may have minimized the possibility of finding developmental differences. The range of children's ages in the sample did not represent all of the critical developmental periods. In fact, it could be argued that by third grade, socializa- tion to television may be complete to the extent that third and seventh graders §1_19_1_Jl__d_. perceive TV characters similarly. There are at least two other plausible explanations for the similarity in the number of perceptual dimensions. First, a max- imum number of usable dimensions may have been reached by third grade. It is possible that three or four dimensions represent the most com- plex system for categorizing TV characters available to children or even to adults. The addition of new dimensions may progress only 85 through early development and achieve a ceiling level after con- siderable experience with the medium. Second, the similarity between age groups may be due to other effects of socialization that are countering the development of more complex cognitive processes. At the same time that complexity of evaluation is increasing for children, they may also be learning the relevant dimensions which society expects us to use in evaluat- ing others. The ability to use more dimensions then, may be offset by the discovery that only certain ones should be applied. Dimension Content The most obvious dimension which emerged in the multidimensional space was humor. It accounted for the majority of variance in the four dimensional solution. Due to the high negative association with support, humor here was labeled somewhat differently than might be expected. Although no previous research has looked at the role of humor in imitation, it seemed reasonable to assume that children would positively evaluate humorous behavior. Funny people are generally reinforced for their talent, making them desirable models. The humor which emerged in this analysis was, however, unsup- ported.humor. The characters most representative of this attribute were those laughed at rather than laughed.with -- the classic "boobs." Characters at the other end of the dimension were not funny, but very much supported by television peers. These results suggest that the humor dimension which differen- tiates TV characters is not a positive, sought after attribute. The 86 total inability of the unsupported.humor dimension to predict which characters children wanted to do like or be like supports this find- ing. While this dimension represents the primary characteristic that differentiates TV characters, it is not related to children's application of television to real life. While these results are not contradictory to previous research on humor, several studies would predict that the characters liked most by their TV peers would be most eligible for modeling. Bandura's notion that vicarious reinforcement is as much a determinant of be- hayior as direct reinforcement is not supported in this study; Highly supported characters are no more desirable than the comical boobs. It is possible, however, that for Children there is a difference between liking someone (the wording used in the questionnaire) and being supportive. Since it was assumed that "support" would not have a Common referent for all age groups (and maybe no referent for the third graders) the word "like" was substituted as a reasonable synonyms Since this was the only dimension which did not to some extent predict either of the two dependent measures, it can also not be assumed that all dimensions used in information processing are re- lated to subsequent behavior. Testing of the dimension's relation- Ship to attitudes and'behavior was in this case, however, only a beginning. Other researCh employing perceptual dimensions as pre- dictors of media effects should further explore the extent to which humor mediates Children's modeling of TV characters. 87 The humor dimension was most supportive of the hypothesized de- velopmental shift in the use of dimensions descriptive of TV char— acter's behavior. Humor is used.muoh more by older children to differentiate TV characters. This particular shift could be related to the "laugh at" con- text of the humor dimension. Younger children may not be as able to recognize which characters are comically portrayed. The distinction 'between the talent or ability to make people laugh and.humor associ- ated with mistakes and.misfbrtune may require considerable experience with television entertainment. .Another explanation for the increased use of humor by older children may be dependent on differences between person perception in general and perception of people on TV. Younger children may be more inclined to evaluate TV characters in the same way they do people in real life. Older children, however, may have been social- ized that television is entertainment. Their perception of people on TV may be influenced by their recognition of a primary function of the medium. Dimensions 2 and 3 are probably best discussed together because they both represent stereotypic evaluations of other peeple. While these dimensions do not account for as much variance in the multi- dimensional space as humor, they are most predictive of children's desires to be like and do like TV characters. The content of the two dimensions, masculine strength and fem, inine attractiveness, are both descriptive of physical attributes of TV characters. They are also more present in the spaces of younger 88 children. These attributes are easily recognizable by young chil- dren and their increased use is generally supportive of research on developmental changes in information processing abilities. The stereotypic use of these dimensions is probably most associ- ated with sex differences. Males should use the strength dimension most and females the attractiveness dimension. This was the find- ing, although both dimensions were significant predictors of the two attributes for both sexes. Furthermore, the spatial configurations for males and females were generally very similar as determined by canonical correlation analysis. This finding is most interesting when the predictive power of the two dimensions is considered. For males, strength was the pri- mary predictor of wanting to do like and be like TV characters while attractiveness was totally unrelated to the two dependent measures. For females, the opposite was true. Attractiveness was highly pre- dictive of the two measures and strength was unrelated. The magni- tude of the multiple correlations, however, was equal. The conclusion is that cognitive structures for males and females are the same while the use of the structures is completely different. This is different from saying that only males use the dimension strength and, therefore, strength is only predictive of media effects for males. Males use both strength ar_1_d_ attractiveness, but only one dimension is applicable to their modeling decisions. The same is true for females' use of attractiveness. The relation between the attribute good and the two dimensions is also consistent with sex differences in the prediction of the 89 dependent measures. Good was most related to strength for the males and to attractiveness for the females. This indicates, rather ob- viously, that children are most interested in imitating good people. These results may be more expected by parents than psychologists studying sex differences. In an extensive current review of psy- chological research on sex differences, Maccoby and J acklin (1975) find no support for assuming that male and female children differ- entially use dimensions traditionally associated with their sex. Increased aggression among males is a well documented finding and this could possibly explain their dependence on the strength dimen- sion to evaluate TV characters as suitable for imitation. Regarding the stereotype that females are dependent on attributes related to social interaction (i.e. physical attractiveness), Maccoby and Jacklin conclude there is no empirical support. This study would seem to be a glaring exception. The regression weight for dimension 3, attractiveness, as a predictor of wanting to be like TV characters was .7h (p<.OOl) for females and .00 for males. A note should be made about attaching sexual descriptions to strength and attractiveness. Masculine and feminine were included in the dimensions because they were also highly predictive of the sex of TV characters. The lack of female TV characters in the sample (and also on TV in general) may make these labels misleading. For masculine strength, the females were the three characters most unrepresentative of the dimension. If different characters were included (for example, Christie Love -- female police officer and judo expert) this may have altered the dimension's correlation with sex. 90 For feminine attractiveness, only the two older female char- acters (Mary Tyler Moore and Samantha) ranked high on the dimension. .Attractive males included in the sample (e.g. Steve Austin and Hawkeye) were rated.almost as high. Again, with a different sample of Characters, the tendency for females to dominate the high end of dimension may be less obvious. These speculations are further cau- tion that the dimensions uncovered in this study are totally related to the television characters which were used. The fourth dimension was activity. Although this dimension ac- counted for the least amount of variance in the final solution, it did predict which characters certain subgroups most wanted to be like and do like. Activity, a behavioral attribute, was most present for older children. This result was supportive of the hypothesis that older children would rely more on behavioral descriptors. This result could have been more a function of age differences in the meaning of activity. The label "activity" was an attempt to operationalize comments from children in the pre-test interviews that certain characters "moved around a lot" or "did a lot of different things." It is possible that for fifth and seventh graders, the word "activity" adequately represented these characteristics, but for the the third graders the word did not have as concrete a meaning. Furthermore, the attribute "active" gguld be predicted for the third graders. Dimension 2 (strength) predicted activity as well for the third graders as dimension h did for fifth and.seventh 91 graders. Perhaps active is more of a physical attribute for the younger children. Although "activity" does not account for a substantial propor- tion of variance in the multidimensional solution ( 5%), it is highly predictive of the dependent measures. It provides a good example that the criteria for eliminating dimensions, factors or clusters from a solution should not depend only on the percentage of variance accounted for in the final solution. Activity was most predictive of wanting to be like and "do" like TV characters for the males. The only group of females for which activity was a significant predictor was the seventh graders. Activity predicted both measures for all groups of males except third graders. The above conclusions were taken from a grade by sex breakdown in the regression analysis. Both sex and grade analyses suggested that a grade by sex breakdown would further illustrate subgroup differences in the predictiveness of the four dimensions. Table 12 shows the results of this analysis. For the variable "want to be like," male/female differences are similar to previous analyses, except for dimension h, activity, as a predictor for the males. Activity is only a significant pre- dictor for fifth and seventh grade males. Forfemales at all ages, attractiveness is equally predictive of which TV Characters they want to be like. A breakdown for the variable "want to do like" revealed greater sex by grade differences. For males at all ages, the four dimensions 92 .oomumom in .83 205 £32, .meflmfleem Heeamoaoeeawm .eeseznez gee 9% . n .mQOfiomHoaaoo opp CH modam> N one x onp on muonasa 03p omega .ngsoumnSm xom an modem SQmo h nopodnmso mew sobww weapon some one so oommpv moanoflam> ems mwooa one we some pom odao> m . one Amamhamem A oewofioaooo m was uopooaoSo seam manpomhono.>a mo genes: may eza we : am so comma one wflmhadqm scammonwon one msoapwaoaaoo mafia Sofia; mo.vm* Wk... mm. :80 *wm.+ N on? o 0 0| 0' o o e. e. a. a; HOW; ammaemw.“ mm- m: m. m. “Mammalian we :o *8 om... em... Ems. to; 5.- mm... mm. me. $38M Em a:. on. as. *mm.+ mm.+ mm.+ o .+ s . . .u . Wm” mm” mm“ *st+ m:.+ 00.+ mfl.+ mu.“ mm.H HH.: mm. “WNW“ “mm m Hm + em + 20.- Ha.+ *mw.+ *mm.+ oa.u so. modes_eem mxHH oo op eses mm. mm. *lx-Jmo HN0+ Hmo+ FF. 0 0| ol 0 o a.” e: wane e; a; e. ”Maia w o *>® mo I Hm.+ **m>.+ **m>.+ :m.u wH.n Hm. *mm. moamEmm ohm 0N... Jmo immo *Nmo+ Omo+ NH. o o o o 0 me“ e. *5. a; a; a.” we. Mme m? N? m. mime o E *5 was. was. 8.- 3; $83. *3... ma... 6. amines gm 33 en 0 Cd 9mm we mm m mpom a mpom a poem a mpom A p p 3 Am :23 memo memo HES uxcflhnm Mom >pw>flpo< wmoao>fipowspo¢ npmsoapm nossm empeefioev manage 83832 eefioeebmes 8 803559 .2083 Bee scam meansaaa, 382m Sees wAszodxwonm xom hm moan Hmz.noammohwom ooufideUdem one mnoapdaohhoo hooao OHoN ado mg 0. ohm Mom mfiflm 039.. GO mmfiwfiwm 2 .3. 8 NH MHQQB 93 yielded insignificant multiple correlations. These correlations for females, however, were significant at all three age levels. Di- mension 3, attractiveness, increased in magnitude as a predictor of "want to do like" as age increased. For the seventh grade females, dimension h, activity, was also a significant predictor. A general comment should be made about the conceptualization of the four dimensions. The labels given each dimension came from the unidimensional attributes which were hypothesized to emerge in the space and which were measured separately in the study. If a uni- dimensional attribute could be predicted by one of the dimensions from the multidimensional space, then that attribute was included in the conceptualization of the dimension in the same form as it ap- peared in the unidimensional question. For example, strength was predicted by dimension 2, therefore, dimension 2 was called "strength." It is possible that the inclusion of other attributes would have changed the labels attached to each dimension. There could be attributes which are highly correlated with those used that would add to or more clearly define the dimensions. Dimension 1 is a good example. If support were not included in the study, dimension 1 would have been labeled just "humor." The high negative association between support and dimension 1, however, suggested the dimension be interpreted as unsupported humor. There could be other attributes not included that might change the labels placed on any dimension. Is violence related to mas- culine strength? Are attractive females also nurturant? 91L These dimensions should also not be interpreted as having polar opposites for anchors. For example, it should not be assumed that the opposite of highly funrw characters are serious characters un- less the attribute serious is later found to negatively correlate with funny. In fact, by examining where various characters rank on the unidimensional ratings, it may not even be safe to say that a low rank means the character does BEE possess the attribute. The only conclusion that can be safely drawn is the relative position of each character in relation to other characters in the sample. All characters, for example, were perceived as being highly supported by their peers, only some were more supported than others. Differences in the Dependent Measures Among subgroups of children, the extent to which children wanted to be like and "do" like TV characters was definitely related. There were differences in these two measures, however, that are worth noting. Generally, the four dimensions accounted for more variance in children's desire to be like than to do like TV characters. For every sex by grade subgroup (Table 12), the multiple correlation for be like was larger than for do like. This is perhaps a little surprising since to be like a character seemingly requires a greater commitment. There could be numerous characters who may not be ac- ceptable in general, but who possess certain attributes or do dif- ferent things that would be suitable for imitation. In fact, the mean ratings for each character on "do" like were in all cases higher than the means for be like. The lower multiple 95 correlations for do like only signify that children's desires to do like characters were more difficult to predict from the four dimensions . The grade by sex regression analysis showed the greatest dif- ferences in the two dependent measures (Table 12). For males in all three age groups the multiple correlation for the four dimensions as predictors of "want to be like" were insignificant. All of the other subgroups had significant values. It is difficult to say why the extent males want to do like char- acters is unpredictable while it is for females. Perhaps there is no systematic means by which males select portions of TV character's be- havior to model. For males it seems to be all or nothing. The INDSCAL Method Many different metric and non-metric multidimensional scaling procedures and programs are becoming available. Each program incor- porates new ideas about multidimensional scaling and because of the relative newness of this methodology some comments about INDSCAL seem appropriate . Evaluating the face validity of INDSCAL is probably most im- portant. If the program is not able to create from the similarity data a configuration of concepts that makes sense, the method is of little use. The first three INDSCAL dimension were readily interpretable even before the regression analysis. Just by examining the plots of TV characters, it was apparent that humor, strength, and attract,- iveness, or very similar concepts, would describe the dimensions. 96 This is not to say that the interpretation of the spaces can be totally credited to INDSCAL; however, the program provided no obstacles in the interpretation. In literature discussions of the multidimensional programs, serious criticisms are leveled against various procedures (i.e. use of non-metric programs, ordinal vs. interval/ratio similarity judgments). The INDSCAL program has been technically criticized in regard to use of specific algorithms used to derive the concept spaces. Some methodologists maintain that subtle differences in the calculations have the potential of greatly changing results. To check the external validity of the INDSCAL program, the similarity judgments used in this study were input into a different multi- dimensional program. The other program used was Galileo (Serota and Woelfel, 19714). It is a computer package for metric multidimensional scaling. Canonical correlation analysis was used to test the similarity of spaces generated by the two programs. The first four canonical cor- relations were .99, .98, .9u, .81, all significant beyond .001. This analysis indicated a veryhigh degree of isomorphism between the two programs and increased the confidence which can be placed in the INDSCAL solutions. Similar comparisons done between INDSCAL and other non-metric programs (e.g. the Kruskal program) have shown similar results (Sherman, 1971). One unique aspect of the INDSCAL procedure which was not util- ized in this research may be especially appropriate for future media studies. The INDSCAL program not only provides for the input of 97 subgroup similarity matrices (in this study, three grades and two sexes) , but also for each individual subject's similarity matrix. The program will calculate comparable individual spaces for each sUbject. The analysis of individual spaces may answer one question not dealt with in this study. Although the subgroup spaces were gen- erally very similar, how representative is the subgroup space of the individuals within that group? For example, it could reasonably be hypothesized that individual variance in the space for seventh graders would be less than for third graders. Greater chance for socializa- tion may decrease individual differences in how children perceive television. This hypothesis can be tested only by analyzing in- dividual spaces. Future Research It seems most apprOpriate to discuss future research related to this study before discussing the implications of the present find- ings. Because this study was a first attempt to describe and apply the dimensions children use to differentiate TV characters the im- plications of the findings depend a great deal on further empirical confirmation. Probably most important is the validation of these dimensions using different samples of television characters. Because dimension content in multidimensional scaling is a function of the concepts used, these dimensions may be related only to this particular sample of characters. 98 One of the primary considerations in selecting characters for this sample was familiarity across a large age range. This method of selection, however, left out characters which are very popular with only younger or older children. Future samples should include dramatic characters from late evening television (e.g. Christie Love, Mannix) and other cartoon and late afternoon characters. It is even possible that the similarity in multidimensional so- lutions across age groups could have been forced because the same characters were used for all three grades. A more convincing test of space similarity would come from a comparison of two spaces in which the characters represented the unique viewing habits of the specific age group. It is not even totally unreasonable to envision a study in which primary TV characters would not have to be sampled. There are only slightly over 60 such characters on the air at this time. By using separate samples of children to complete different portions of the necessary paired comparisons, similar analyses would then be able to examine population parameters. While it is still possible that these characters have yielded dimensions unique only to themselves, it is also a possibility that the values are unrepresentative because of the sample of children. Other studies should base the character ratings on a more hetro- geneous group of children that vary on qualities other than age and sex (e.g. socioeconomic status, cross-cultural differences, etc.). Ohter unidimensional attributes should also be tested for their relation with the dimensions in the space. As previously mentioned, 99 other attributes which are either positively or negatively related to those used in this study may alter the conceptualization of some dimensions . The use of perceptual dimensions to predict media effects was limited in this study. Although the dimensions were highly associated with children's desire to be like and do like TV characters, these measures represented a small sample of those possible. It would be useful to know if the dimensions children use to distinguish char- acters are related to exposure, behavioral measures of imitation, learning of sex roles from TV, identification with characters, etc. Finally, a greater range of age groups should be studied. It is possible that by third grade children have already been socialized to uniformly evaluate television and TV characters. Ideally, children as young as age three should be included. They have already had one years experience with the medium and have no doubt developed system- atic conceptions of favorite portrayals. This inclusion, however, implies the development of simpler, less time consuming methods of reliably measuring children's judgments of television. Experimental studies could be used to further test developmental and sex differences in the content and use of these dimensions. Chil- dren shown identical portrayals in controlled settings could be sub- sequently tested on both perceived differences between characters and the applicability of defined attributes. Exposing each child to the same segment of a character's behavior would assure a common referent for their evaluations. In a survey situation these refer- ents can only be assumed. 100 Present Implications of Research Findings This type of research should not only be useful for guiding future empirical studies on media and children. There are also several implications for the production and evaluation of tele- vision for child audiences. The knowledge of which character at- tributes are most responsible for impact on children can greatly increase the deliberate communication of pro-social messages and decrease the effects of anti-social portrayals. Each dimension and its association with the media effects variables in this study is related to both types of impact. First, humor appears to be a neutral attribute in terms of differentiating characters which children model. Producers should not depend on humor to deliver pro-social messages, nor is there reason to believe that funny violence, for example, is any differ- ent from serious violence. Perhaps the Cookie Monster should not be counted on to teach children good table manners. Strength, attractiveness and activity are clearly attributes which producers and parents should be aware of. The more they are perceived to be present, the greater the impact of the portrayal. These findings suggest that strong active males will have the great- est effect on girls. As with the other dimensions, the results can be used to either augment prosocial messages on television or dimin— ish negative consequences of exposure. The regression equations from this research could actually be used to calculate which characters would have the most impact for certain subgroups of children. By multiplying the regression weight 101 for each dimension by the coordinate value for each TV character on that dimension, a value could be calculated which represented TV character ' s likely effect . For example, the regression equation for predicting which TV characters males most wanted to be like is: Y = (-.06)(value for DIM l) + (.89)(value for DIM 2) + (.00) (value for DD! 3) + (.56)(value for DIM h) Y in the formula represents the mean rating on "want to be like" for each character. The equation predicts 75% of the variance in Y. Applying this formula to all fourteen of the TV characters yields the following results. The values for each character should indicate the extent to which males want to be like each character based on a weighted consideration of all four perceptual dimensions. Predicted Value of Y for Males Character Steve Austin .339 Hawkeye .329 Reed .309 Chico .292 Marshal Dillon .222 Fred Sanford .021 J ohn-Boy .020 Archie Bunker -.l8l Samantha -.235 Gilligan -.251+ Fred Flintstone -.255 Fat Albert -.267 Laura «3’43 Mary Tyler Moore «M35 From these values the relative impact of several TV characters can .be found. Content analyses could be done to determine the behaviors of those characters highest on the list. The combined information 102 would indicate which characters and which behaviors were most likely having an impact on children. These lists, however, would be no more accurate for all chil- dren than the regression analyses were in this study. Separate consideration should be given to different age groups and especially to males and females. Applying the equation for females yields a completely different table: Character Predicted Value of Y for Females Semantha .159 Mary Tyler Moore .371 Laura .253 Steve Austin .178 Reed .062 Hawkeye .0h9 Marshal Dillon .002 Chico -.026 J ohn-Boy «0147 Gilligan -.O99 Fred Sanford -.lO6 Archie Bunker -.l73 Fat Albert -.h03 Fred Flintstone -.’+08 It is obvious from a comparison of the tables that children identify most with same sex TV models. The portions of the tables that are similar (Steve Austin through Chico) are, however, based on entirely different dimensions. For the males, these characters are ranked high because they are strong and active. For the fe- males, they rank second to the TV females because they are the most attractive. These "impact values" provide a very precise way of deter- mining which characters to watch most closely. They are not based on content analysis of researcher defined attributes, as most other 103 ratings of TV shows and characters. They are dependent on the character's rating along perceptual dimensions which are maxi- mally weighted to predict children's identification with TV char- acters. As indicated by the multiple correlations, these should be very good predictions. Relation of Results to Past Research Media research had already proceeded beyond simple relation- ships between exposure and effects to the study of mediating cog- nitive processes which predict effects for some children and none for others. Research on the perceived reality of television is a good example of this research. This research attempted to take the study of these intervening processes one step further. The methods were chosen so that dimen- sions would not be forced on children but rather would be generated from data based only on similarity ratings between TV characters. Do the dimensions which were found in this study mean, then, that others do not exist, or that ones used successfully in the past to predict media effects are not operative? The relation of these dimensions to those used in previous re- search is very important. A probable interpretation is that chil- dren are capable of distinguishing characters along several dimensions when provided with a concrete description of the at- tribute. Childrcn could probably very accurately rank TV characters according to length of hair. This ability, however, does not sig- nify that the dimensions are in fact operative or related to the character' 8 impact . 1014 This is the most likely explanation for the inability of the dimensions to predict the perceived age or perceived reality of TV characters. There are good theoretical rationales why both vari- ables should predict media effects. Younger characters should be more desirable because they are likely to engage in behaviors and activities most salient to young children. More realistic char- acters should increase the acceptability of applying TV behavior to real life. While the relation of both these variables to media effects may still be valid, they first must be operative as a perceptual process. At least in this study, this is where they failed. Even if they are found to be highly predictive of media effects, their use by children to differentiate TV characters appears to be minor at best. What do we know about children and television that wasn‘t known before? Very generally, we have some idea of the dimensions children use to discriminate between television characters, and there is evidence that these information processing formats are related to how children apply media portrayals to their own lives. These findings also seem relevant to general communication theories of information processing. When people categorize informa- tion, the information is likely to be systematically differentiated along one or several dimensions. The methods used here have been demonstrated to be a useful and parsimonous means of obtaining these dimensions. These dimensions then should be related to the information's application to real-life situations. 105 The findings here could be generalized to the processing of communication messages other than those from television. For ex- ample, political messages may be categorized according to the di- mensions important in voting decisions. Information about occupations may be processed along dimensions relevant to vocational choices, and interpersonal relationships may be sorted according to their function for the participating individuals. These examples suggest that messages are processed in relation to the information's function. These functions may determine how information is originally interpreted and categorized for later use. Humor may be an important dimension for differentiating TV char- acters because humor is a primary attribute that assists children in deciding which characters to watch; strength may be a dimension because it helps children (at least males) identify apprOpriate TV models. Similar reasoning could apply to dimensions used to differen- tiate communication messages about other concepts. For example, the type of information necessary to decide a vote may determine the dimensions underlying political messages. If people decide votes based on candidates attractiveness, then attractiveness should be a dimension that differentiates political candidates. The idea that dimensions of message effects will predict the dimensions of message processing is the most general implication of the present study. .Although the data.here relate most to the pro- cessing dimensions of television information, there is at least preliminary evidence that these dimensions will predict important. effects of the same content. BIBLIOGRAPHY ,Atkin, Charles and Miller, M. Mark. "The Effects of Television Advertising 0n Children: Experimental Evidence." Paper pre- sented to the Mass Communication Division of the International Communication Association, April, 1975. Bandura, Albert. Principles of Behavior Medification. New York: Holt, Rinehart, and Winston, 1969. Bandura, Albert. "Vicarious Processes: A Case of No-Trial Learning," in Berkowitz (ed.).Advances in Experimental Social Psychology, Vol. II. New York: Academic Press, 1965. Bandura,.Albert; Grusec, J.E.; and Menlove, F.L. "Some Social Determinants of Self-Monitoring Reinforcement Systems," Journal of Personality and Social Psychology, 5, 1968. Bandura,.Albert; Ross, D.; and Ross, S. "Vicarious Reinforcement and Imitative Learning," Journal of Abnormal and Social chhology, 66, 1963 . Beuf, Ann. "Doctor, Lawyer, Household Drudge," Journal of Communication, 23, 197k. Bieri, J.; Atkins,.A.L.; Briar, S.; Leaman, R.L.; Miller, H.; and Tripodi, T. Clinical and Social Judgment. New York: Wiley, 1966. Bruner, J.S. "0n Perceptual Readiness," Psychological Review, éfl, 1957. Bruner, J.S. "The Course of Cognitive Growt ," American Psychologist, l9, 196A. Carroll, J. Douglas and'Wish, Myron. "Models and.Methods for Three-way Multidimensional Scaling," in Contemporary Developments in Mathematical Psychology, (Eds.) Krantz, et. al., l97h. Cooley, W.W. and Lehnes, P.R. Multivariate Procedures for the Behavioral Sciences, New York: Wiley, 1962. Dornbush, S.; Hastorf, A.H.; Richardson, S.A.; Muzzy, R.E.; and Vreeland, R.S. "The Perceiver and the Perceived: Their Relative Influence on the Categories of Interpersonal Per- ception," Journal of Personalityyand Social Psychology, 1, 1965. Feshbach, Seymour. "Reality and.Fantasy in Filmed Violence," in Television and Social Behavior. Vol. III, (ed.) Murray, Rubinstein and Comstock. U.S. Government Printing Office, 1969. Fink, Edward and Walker, Barbara. "Relative Status, Anticipated Interaction, and Social Facilitation as Determinants of Humorous Responses to Embarrassment," Unpublished manuscript, Michigan State University, 1975. Gibson, E.J. "Perceptual Learning," Annual Review of Psychology, 11+, 1963. Greenberg, Bradley. "Children's Reactions to TV Blacks," Journalism Quarterly, E2, 1972. Greenberg, Bradley. "Gratifications of Television Viewing and Their Correlates for British Children," in Annual Review of Communica- tion Research (eds.) Blumler and.Katz, 1975. Greenberg, Bradley and Dervin, Brenda, with Jehn Bowes and Joseph Dominick. The Use of the Mass Media by the Urban Poor. New York: Praeger Publishers, 1970. Greenberg, Bradley and Gordon, Thomas. "Perceptions of Violence in Television Programs: Critics and the Public," in Comstock and Rubinstein (eds.), Television and Social Behavior, Vol. I, Department of Health, Education and Welfare, 1971. Greenberg, Bradley and.Hanneman, Gerhard. "Racial Attitudes and the Impact of the TV Blacks," Michigan State University, Department of Communication, 1969. Greenberg, Bradley and Reeves, Byron. "Children and the Perceived Reality of Television," Journal of Social Issues, in press. Hastorf, A.H.; Richardson, S.A.; and Dornbush, S.M. "The Problem of Relevance in the Study of Person Perception," in Tagiuri and Petrullo (eds.), Person Perception and Interpersonal Behavior, 1958. Helm, Carl E.; Messick, Samuel; and Tucker, L. Psychological Law and Scaling Medels, Educational Testing Service: Princeton, 1959. Kruskal, J.B. "How to Use MDSCAL, A Program to Do Multidimensional Scaling and Multidimensional Unfolding." Unpublished report, Bell Telephone Laboratories, 1968. Livesley, W.J. and Bromley, D.B. Person Perception In Childhood and Adolescence. New York: Wiley, 1973. Lyle, Jack and Hoffman, H.R. "Children's Use of Television and Other Media," in Television and Social Behavior. Vol. IV, (ed.), Rubinstein, Comstock and Murray. U.S. Department of Health, Education and.Welfare, 1971. Maccoby, Eleanor and Jacklin, C. The Psychology of Sex Differences. Stanford University Press, 1975. McLeod, Jack; Atkin, Charles; and Chaffee, Steven. "Adolescents, Parents and Television Use: Adolescent Self-Report Measures from a Maryland and Wisconsin Sample," in Television and Social Behavior. Vol. III, (eds.) Rubinstein, Comstock and Murray. U.S. Department of Health, Education andeelfare, 1971. McNemar, Quinn. Psychological Statistics. Wiley, Inc., 1969. Miller, M. Mark and Reeves, Byron. "Children's Occupational Sex Role Stereotypes: The Linkage Between Television Content and Perception." Paper presented at the annual meeting of the International Communication Association, Chicago, 1975. Olshan, K.M. "The Multidimensional Structure of Person Perception in Children." Unpublished dissertation, Rutgers University, 1971. Peevers, B. and Secord, P.F. "Developmental Changes in Attribution of Descriptive Concepts to Person," Journal of Personality and Social Psychology, 21, 1973. Piaget, Jean. The Origin of Intelligence in the Child. London: Routledge, 1953. Serota, Kim. "Metric Multidimensional Scaling and Communication: Theory and Implementation." Masters Thesis, Michigan State University, Dept. of Communication, l97h. Shepard, R.N. "The Analysis of Proxemities: Multidimensional Scaling with an Unknown Distance Function, I." Psychometrika, 31, 1962. Sherman, Richard. "Individual Differences in Perceived Trait Relationships as a Function of Dimensional Salience," Multivariate Behavioral Research, January, 1972. Skinner, B.F. Science and Human Behavior. New York: MacMillan, 1953. Stotland, E.; Zander, A.; and Narsoulas, T. "The Generalization of Interpersonal Similarity," Journal of Abnormal and Social Psychology, 62, 1961. Torgerson, Warren. Theory and Methods of Scaling. Wiley, 1958. Vialters, R.H. and Parke, R.D. "Influence of Response Consequences to a Social Model on Resistance to Deviation," Journal of Experimental Child Psychology, l, 196M. ‘Walters, R.H.; Leat, M.; and Mezei, L. "Inhibitions and Disinhibi- tions of Responses Through Empathetic Learning," Canadian Journal of Psychology, ll, 1963. Wegner, Daniel M. "The Development and Articulation of.Attributes in Person Perception." Dissertation, Michigan State University, 1975- Woelfel, Joseph and Barnett, George. "A Paradigm for Mass Communica- tion Research." Paper presented at the annual convention of the International Communication Association, New Orleans, l97h. Wohlwill, J.F. "From Perception to Inference: A Dimension of Cognitive Development." In Kessen and Kuhlman (eds.) Thought in the Young Child. Monographs of the Society for Research in Child Development, 21, 1962. Yarrow, M.R. and Campbell, J.D. "Person Perception in Children," Merrill-Palmer Quarterly, 2, 1963. APPENDIX A Pre—Test Questionnaire (Introduce yourself to student) A lot of times people ask adults what they think about the shows and people they watch on TV. Today we are interested in what young people think about TV: ‘ Here are some questions about the people and shows on TV. Please listen to the questions and then tell me what you think. There are no right or wrong answers, we just want to know what you uthink. Q First. we want to know about some of your favorite television shows. rgs—O.-' C“ What are your two most favorite TV shows? -e-W".' A) . a) .._.. -v- Now tell me who are your two most favorite TV characters? . 1) 2), .owo. —” Who eretwo characters you don't like on TV? fl...-~o o-O-fi- 3) 7 ll) . 2,; .. f-.- :3} '3 .‘~ (Use the characters' names as indicated above in asking the following questions.) now is .. (1) different from (30 ? I . ,. \ ' - ., .3.. . . (is there anything else?): How is (2) different from (3) 2._ ~ .nn £18 there anything else?): How is" " '(1)‘ " different from (2) ? (is there anything else?): b- ..nuaw is (a) different f'mni' (it) - ? (is there anything ¢1332)= _,- v a. .._ -; a" - ’ ‘ "1““ How is (2) different from (u) " 1““ ' (is there anything else?): How is (1) different from ('4) ? (. -‘:£:""<'I=' “- ~ (is there anything else?): _. Is there anything that (1) does that you would like to do? (If QQESE What? " " — ,‘A "rim" .'= (v - ' ‘ i J- t. If there anything that (2) does that you would like to do? (If yes):" What?. .» - --- ' " ' Student's name: APPENDIX B Description of Television Characters Used for Similarity Judgments Appendix B lists the TV characters used in the study, the network and time of day of the program they appear on, and a brief description of them and their program. This information is based on listings from an April, 1975, issue of TV Guide. Fred Sanford on "Sanford and Son" (NBC, Saturday, 8:00 p.m. , EST). Fred Fred Sanford is in a junk yard partnership with his 30-year-old son. The two live next to their junk yard. A situation comedy, the program usually focuses on conflicts between the lifestyles of Fred and his son. Both characters are black. Flintstone on "The Flintstones" (Syndicated, Saturday, 10:30 a.m-, EST). "The Flintstones" is a cartoon series which depicts a fantasy version of prehistoric life. Fred Flintstone is the head of a cave family from which most of the characters are taken. Gilligan on "Gilligan's Island" (Syndicated, weekdays, 1+:OO p.m., EST). Gilligan is one of seven vacationers trapped on a tropical island after the wreck of their pleasure yacht. Most episodes comically portray different attempts by the group to escape. Gilligan is generally considered the comical "boob" of the group and he is frequently faulted for their failures to escape. Laura on "Little House on the Prairie" (NBC, Wednesday, 8:00 p.m., Mary Reed EST). Laura is the second oldest of three girls (approximately age 7 or 8) in a Mid-Western prairie family. The family lives on a farm.near a small agricultural town during the late nine- teenth century. Frequent themes of the program are attempts by families in the community to help others through various troubles. Tyler Mbore on "The Mary Tyler Moore Show” (CBS, Saturday, 9:30 puma, EST). Mary Tyler Meore portrays a TV news producer in a small Minneapolis newsroom. The situation comedy usually focuses on events at the newsroom and on others who work at the TV station. on "AdamrlZ" (NBC, Tuesday, 8:00 p.m,, EST). Reed is one of two unifbrmed police officers who patrol various neighborhoods in a squad car. The program.most frequently portrays the more common aspects of law enforcement. The officers code name for assign- ments is "Adam-l2." Samantha on "Bewitched" (Syndicated, weekdays, 7:00 p.m., EST). Samantha is a witch in the person of an attractive woman married to an advertising executive. Most stories involve her relatives' use of witchcraft to get her mortal husband into trouble with advertising clients, and Samantha's use of magic to save him from embarrassment and failure. Hawkeye on "M*A*S*H" (CBS, Tuesday, 8:30 p.ms, EST). Hawkeye is an MD working with a mobile field hospital in the Korean war. Both comical and serious parts of the program deal with conflicts be- tween two drafted doctors who are unsympathetic to the war and patriotic attempts by other staff to promote the war effort. Archie Bunker on "All in the Family" (CBS, Saturday, 8:00 p.m., EST). Archie is the father in a family consisting of his wife, daughter and daughter's husband. Several episodes in the Situation comedy involve conflicts between the conservative attitudes of.Archie and the liberal ideas of his son—in-law. This series was one of the first to deal comically with such topics as racism, sex, and religion. Chico on "Chico and the Man" (NBC, Friday, 8:30 p.m., EST). Chico is a young Chicano working for an elderly man in an urban gas station. Most episodes involve Chico's attempts to help his unwilling boss with personal problems or with fixing up and promoting the gas station. Steve Austin on "The Six Million Dollar Man" (ABC, Friday, 7:30 p.m., EST). Steve.Austin is an astronaut and government agent Who possesses unusual strength as a result of bionic limbs and eye- sight. After a near-fatal flying accident, Austin was given Special artificial body functions to replace those damaged in the accident. Most episodes in this action adventure series deal with Austin's use of his strength to overcome environmental or personal opposition. Fat Albert on "Fat Albert and the Cosby Kids" (CBS, Saturday, 12:30 p.m,, EST). Fat Albert is one of several children that play together in a Saturday morning cartoon series narrated by Bill Cosby. Each story focuses directly on some prosocial theme. Examples include: how to treat new kids in a neighborhood, de- struction of property, and obedience to parents. The children in the series are black. .Iohn-Boy on "The Waltons" (CBS, Thursday, 8:00 p.m., EST). John-Boy is the eldest in a large family living in the Virginia.hills during the depression. Most episodes focus on individual members of the family and attempts by the others to help them.through various problems. Marshal Dillon on "Gunsmoke" (CBS, Monday, 8:30 p.m., EST). Marshal Dillon is the head law enforcement officer in a small nineteenth century western town. The series is the last to focus on "good gw/bad guy" episodes taken from traditional western films. APPENDIX C Group Questionnaire for Similarity Judgments A lot of time people ask adults what they think about the shows and people they see on TV. Today we want to know what young people think about TV. Here are some questions about the people and shows on TV. Please read the questions and then put an §_on the line above what you think. There are no right orwrong answers, we just want to know what you think. The questions ask you to think about two people on TV and then decide if you think they are alike or different. Here is an example. What do you think about Fred Flintstone and Gilligan? very alike I'm not different very much sure different alike Your Name: (7-9) Grade: Room Number: 1. What do you think about Fat Albert and Gilligan? Are they: very alike I'm not' different very much sure different alike A 2. What do you think about Mary Tyler Moore and Archie Bunker? Are they: very alike I'm not different very much sure different alike v— W— vii 3. What do you think about Samantha and Fat Albert? Are they: very . alike I'm not different very much sure ‘ , different alike' ‘4. What do you: think about Archie Bunker and Steve Austin? Are they: very alike I’m not different very much sure different alike '— Vi What do you think about Marshall Dillon and Samantha? Are they: very alike I 'm not different very much sure different alike . 6. What do you think about Steve Austin and Fred Flintstone? alike Are they: very alike I'm not different very much , sure different 'alike What do you think about Reed and Marshall Dillon? Are they: very alike I 'm not different very much sure different Card 1 Col. # 10 11 12 13 l“ 15 16 8. What do you think about Fred Flintstone and Chico? ' Are they: very alike I'm not different very much sure different alike ' 9. What do you think about Hawkeye and Reed? Are they: very ' alike I'm not different very much sure different alike 10. What do you think about Chico and Fred Sanford? Are they: very . alike I'm not different very much sure different alike 11. What do you think about Laura and Hawkeye? Are they: very alike I'm not different very much sure different alike 12. what do you think about Fred Sanford and John-Boy? Are they: ' very alike I'm not different very . much sure different alike 13. What do you think about John-Boy and Laura? Are they: very alike I'm not different very much sure different alike 114. What do you think about Samantha and Gilligan? Are they: very alike I'm not different very much sure different alike Col. # 17 18 19 20 21 22 '23 21. Sanford? Are they: very alike much alike I'm not different sure very 15. What do you think about Mary Tyler Moore and Steve Austin? Are they: ’ very alike I'm not different very much sure different alike 16. What do you think about Marshall Dillon and Fat Albert? Are they: very alike I'm not different very-—— much sure different alike 17. What do you think about Archie Bunker and Fred Flintstone? Are‘they: very alike I'm not different very much sure different alike 18. What do you think about Reed and Samantha? Are they: very alike I'm not different very much sure different alike 19. What do you think about Steve Austin and Chico? Are they: very alike I'm not different very much sure different alike 20. What do you think about Hawkeye and Marshall Dillon? Are they: as?“ m m We: very much sure different alike What do you think about Fred Flintstone and Fred. different J. Col. # 24 25 26 27 28 29 30 Q2. What do you think about Laura and Reed? Are they: very alike I'm not different very much sure different alike ‘ Q3. What do you think about Chico and John-Boy? Are they: very alike I'm not different. very much sure different alike 2n. What do you think about John-Boy and Hawkeye? Are they: very alike I'm not different very much sure different alike 25. What do you think about Fred Sanford and Laura? Are they: ' very alike I'm not different very much sure different alike 26. What do you think about Marshall Dillon and Gilligan? Are they: very alike I'm not different very much sure different alike 27. What do you think about Mary Tyler Moore and Fred Flintstone? Are they: , very - alike I'm not different very much sure different alike 28. What do you think about Reed and Fat Albert? Are they: very alike I'm not different very much sure different alike Col. 31 32 33 3M 35 36 37 fl www v—r—v-wv—‘w— alike 29. What do you think about Archie Bunker and Chico? Are they: very alike I'm not different very much sure different alike 30. What do you think about Hawkeye and-Samantha? Are they: very alike I'm not different very much sure different alike 31. What do you think about Steve Austin and Fred Sanford? Are they: very alike I'm not different very much sure different alike 7 32. What do you think about Laura and Marshall Dillon? Are they: very alike I'm not different very much sure different alike 33. What do you think about Fred Flintstone and John—Boy? Are they: Very alike I'm not different very much. ' sure ' different alike ' 3W. What do you think about John—Boy and Reed? Are they: very alike I'm not different very much sure different alike .35. What do you think about Chico and Laura? Are they: 'GZfiy alike I'm not different very mu 3 sure different Col. # 38 39 N0 #1 U2 1:3 at: Fred Sanford and Hawkeye? a] 1.3"; 36. What do you think about Are they: very alike I'm not different very much. sure different alike 37. .What do you think about Reed and Gilligan? Are they: very alike I'm not different very much sure ' different alike 38. What do you think about yourself and Chico? Are you: very alike I'm not different very much sure different alike 39. What do you think about Hawkeye and Fat Albert? Are they: TEEf: alike I'm not different very much sure different alike' #0. What do you think about Archie Bunker and Fred Sanford? Are they: :Ery alike I'm not different very much sure different alike #1. What do you think about yourSelf and Fred Flintstone? Are ;‘du very alike 'I'm not different very nunfl: sure different alike #2. 'What do you think about Steve Austin and John-Boy? Are they: §:: §“——' 372th _' I'm not different very Inb:n a re ' different Col. # 45 us 1+7 W8 us 50 51 fi— u3. What.do you think about John-Boy and Marshall Dillon? Are they: very alike I'm not different very much sure different alike nu. What do you think about Fred Flintstone and Laura? Are they: very alike I'm not different very much sure different alike #5. What do you think about Fred Sanford and Reed? Are they: very alike I'm not different very 'much sure different alike #6. What do you think about Chico and Hawkeye? Are they: very alike I'm not different very much sure different alike W7. What do you think about Hawkeye and Gilligan? Are they: very alike I'm not different very Inuch sure different alike . 48. What do you think about Mary Tyler Moore and Fred Sanford? Are they: - very alike I'm not ' different very much sure different alike 1:9. What do you think about Laura and Fat Albert? Are they: very alike I'm not different very much sure different alike Col. # 52 53 ‘ SH 55 56 57 58 50. ‘What do you think about Archie Bunker and John-Boy? Are they: very alike }I'm not different very much sure different alike 51. What do you think about yourself and Mary Tyler Moore? Are you: very alike I'm not different very much sure different alike 52. What do you think about Steve Austin and Laura? Are they: ' very alike I'm not different very much sure different alike 53. What do you think about Fred Sanford and Marshall Dillon? Are they: very alike I'm not different very ~ much sure different alike 5““ What do you think about Fred Flintstone and Hawkeye? Are ' they: ‘very alike: I'm not different very lflUCh sure different alike ' 55. What do you think about Chico and Reed? Are they: very alike I'm not different very Inuch sure different alike 56. What do you think about Laura and Gilligan? Are they: very alike I'm not different very nuach sure different alike Col. # 59 60 61 62 63 6Q 65 57. What do you think about Mary Tyler Moore and John-Boy? Are they: ' very Ellie I'm not different very much sure different alike 58. What do you think about John—Boy and Fat Albert? Are they: Cary alike I'm not different very much sure different alike 59. What do you think about Archie Bunker and Laura? Are they: very alike I'm not different very much sure different alike 60. What do you think about Fred Sanford and Samantha? Are they: ‘very' alike_. I'm not different very much sure different alike 61. IRJat do you think about Steve Austin and Hawkeye? Are they: 'vergf _ alike I'm not different very IRUCF] sure different alike 62. Vfixnt do you think about Chico and Marshall Dillon? (Are they: twrry ___ alike I'm not different very_fl. inucfim sure ' different 1.i.< 53, anIt do you think about Fred Flintstone and Reed? Are the)": \wef::m- alike I'm not different very nuic}; ' sure different alike Col. # 66 67 68 6,9 70 71 72 64. What do you think about yourself and Archie Bunker? ‘Are you: very alike I'm not different very much sure different alike 65. What do you think about Mary Tyler Moore and Laura? Are they: very alike I'm not different very much sure different alike 66. What do you think about Fred Sanford and Fat Albert? Are they: very alike I'm not different very Inuch sure different alike 67. What do you think about Archie Bunker and Hawkeye? Are they: very alike I'm not different very much sure different alike 68. What do you think about Chico and Samantha? Are they: ‘very alike I'm not different very rnuch sure different alike 69. \Nhat do you think about Steve Austin and Reed? Are they: ‘very alike I'm not different very tnuch sure different alike 70. What do you think about Fred Flintstone and Marshall Dillon? Are they? ‘very alike I'm not diffefent GEFy Inuch sure different alike Col. 73 7Q 75 76 78 79 # vv—i 71. What do you think about Fred Sanford and Gilligan? Are they: very. alike I'm not diffegent very much sure different alike 72. What do you think about Mary Tyler Moore and Hawkeye? Are they: very alike I'm not different very much sure different alike 73. What do you think about Chico and Fat Albert? Are they: very alike I'm not different very Inuch sure different alike 74. lflhat do you think about Archie Bunker and Reed? Are they: 'very‘ alike I'm not different very rnuch sure different alike 75. ldhat do you think about yourself and Steve Austin? Ikre,you: very W m mm my mulch sure different alike 75, What do you think about Steve Austin and Marshall Dillon? Are they: very alike I—‘m not different very nulCh sure different alike Col. # 80 Card 2 I-||I.\ TL What do you think about yourself and Gilligan? Are you: very alike I'm not diffefent very much sure different alike' 78. What do you think about Chico and Gilligan? Are they: very alike I'm not different very much sure ‘ different alike 79. What do you think about Mary Tyler Moore and Reed? Are they: very alike I'm not different very - much _ sure different alike 80. What do you think about yourself and Fat Albert? Are you: very alike I'm not different very Inuch sure different alike 81. ‘What do you think about Fred Flintstone and Fat Albert? Are they: ‘very' alike I'm not different very ' rnuch sure different alike 82. What do you think about Archie Bunker and Marshall Dillon? Are they: . very alike I'm not different very UHJCh sure different alike 83. What do you think about Steve Austin and Samantha? Are they: very alike I'm not different very much . sure different alike Col. # 10 ll 12 8H. What do you think about yourself and Samantha? Are you: very alike I'm not different very much ‘ sure different _galike 85. What do you think about Fred Flintstone and Gilligan? Are they: very . alike I'm not diffefent very much sure different alike 86. What do you think about Mary Tyler Meore and Marshall Dillon? Are they: . ta?— mtg“ mm mm tag?”— much sure different alike 87. What do you think about yourself and Marshall Dillon? Are you: very alike I'm not different very much sure different alike 88. What do you think about Steve Austin and Fat Albert? Are they: very alike I'm not different very much sure different alike 89. What do you think about Archie Bunker and Samantha? Are they: 3E5§7—__ SIIEE“’ -Ifi;7Efi? diffEFEht very xnuch sure different alike 90. Pflmat do you think about yourself and Reed? Are you: verét alike I'm not different very Inuclx sure different alike Col. # l3 1H 15 16 17 18 l9 9L What do you think about Steve Austin and Gilligan? Are they: very alike I'm not different very much sure. different alike. 92. What do you think about Mary Tyler Moore and Samantha? Are they: very alike I'm not different very much sure different alike 93. What do you think about yourself and Hawkeye? Are you: very alike I'm not different very much sure different alike 94. What do you think about Archie Bunker and Fat Albert? *Are they: ca?“ 3137:? mm mm very much sure different alike 95. What do you think about Archie Bunker and Gilligan? Are they: ‘very alike I'm not different very Inuch sure different alike . 96. What do you think about yourself and Laura? Are you: very 515:? m mm very xnuch sure different alike 97. PMnat do you think about Mary Tyler Moore and Fat Albert? Are they: very much alike alike I'm not 8 ure different very different Col. # 20 21 22 23 21+ 25 26 98. What do you think about yourself and John-Boy? Are you: very alike I'm not different very much sure ‘ different alike 99. What do you think about Mary Tyler Moore and Gilligan? Are they: very alike I'm not different very much sure ' different alike 100. What do you think about yourself and John—Boy? Are you: very alike I'm not different very much sure different alike 101. What do you think about Mary Tyler Moore and Chico? Are they: very alike I'm not different very much sure different alike 102V What do you think about Laura and Samantha? Are they: ' very alike I'm not different very much sure ; - different alike 103. What do you think about Fred Flintstone.and Samantha? ' Are they: I very alike I'm not' different very i much sure . different ' alike ' 104.‘What do you think about John—Boy and Gilligan? Are they: very much alike alike I'm not different very sure different Col. # 27 28 29 30 31 32 33 105. What do you think about John-Boy and Samantha? Are they: very alike I'm not different very much sure different alike 106. What do you think about Fat Albert and Gilligan? Are they: very alike I'm not different very much sure different alike 107. What do you think about Mary Tyler Moore and Archie Bunker? Are they: very alike I'm not different very much sure different alike. 108. How old are you? 109. What date is your birthday on? 110. What grade are you in? 111. boy What is your sex? girl Col. # an 35 36 APPENDIX D Subgroup Distance Matrices for Similarity Ratings The following matrices represent the mean distances between all possible pairs of TV characters used in this study. Distances were calculated separately for each grade and sex. The distances in these matrices were the values in- put into the INDSCAL program. :H mw.m ma 50.: ma.; 52.: :O.m NH m:.: HA ea.e 0m.m 0e.m mm.m OH mm.m 00.m :H.: mH.m mm.m om.m mg was 8.: 2am in: w me.m ma.: H:.: H:.: mH.: mm.: N0.m h em.m me.m mm.m 0a.: :H.m mam :0.m om.: m 40.: 00.: sa.m mm.m mm.m :e.m mm.m 0e.m ea.m m m0.: mm.m me.m 0a.: ee.m a0.m mw.m :m.m Oh.m : mm.: mw.m um.: :m.m 09m mo.: ma.m mm.m Hm.m m 0a0n0.auaaa Hoe xahpaz moqapman mm.: mm.m mm.: H:.« Hm.m mo.m wm.: ::.: 0a.: m:.: 0a.: mm.: 02.: m0.: mm.: 0s.m mm.m mm.m as.m HH.: ww.m mm.m HH.: 00.: mnooz awake hem: nqusm menoh< :Hpm5¢.m>mpm mQOpmpqaamreoph ooano whomedm eohh homuGSOh andwm mhmxzem comm qoaafim Hannah: equeamm ppmpH< new newnaaflo :: m0.: m0: m:.: Hm.: m>.m m:.: ma NH Ha 80 mad mm.m m0.: OH em.m s0.m 0a.: m:.m mm.m >:.m bm.: m>.m ma.: mm.m >m.: w mw.m w:.: 000 em.: :m.: em: Hm.m a mmzm m:.m He.m mm.m ms.m mm.m :m.: o m0.m mm.m m:.m m1: mm.m 8.: 0:.m SA om.m m emzm 0d: mm.m 8.: :00 mm.m 5:.m mm.m 00.m Hm.m : mm.m ma.: m>.m mm.: 8.: 8:0. mm.m m>.m mm.m HO.: m 0020 :62 8m fined: 808.30 0:.: mm.m 0:.: inm mo.m em.m m:.: mm.: mm.m em.: Hm.: e:.: 00.: mo.: :m.: :m.m Hm.m mm.m m>.m wa.: mm.m mm.m 8.: om.m em.m whooz hmahe hem: Hakeem manon< napm:<.m>mpm weepmpmwam emnm oofino cuOMme deem mom- EB. whqu mzmxsam doom GOHHHQ damages moweaawm pumpa< pmm adeHHHU :H em.: 0m.m sm.: 0m.: 0:.m mm.: ma NH HH :m.m om.m mm.m m>.m OH 0m.: m0.m 0:.: m0.m mm.m eo.m Hm.: mm.m em.: mm.m m:.: w mo.m mm.: emzm mm: :0.: ::.: om.m a H>.m oe.m me.m mm.m em.m m0.: m:.: o Hm.m em.m mm.m 0a.: Hm.m mm.m ea.m 0:.m mm.m m mm.m mo.: mm.m :m.: 00.: mm.: om.m om.m He.m mm.m : mm.m :m.: mm.m Hm.: em.m Om.: :m.m Ho.m mm.m as m>.m m edema spnm>mm now Mauve: ooqumwm :H.: m:.m em.: w:.m mo.m 0m.m ma.: :H.: meam :o.: :m.: em.: mo.: h®.m wa.: wm.m 0:.m mm.m mm.m mm.m mm.m mm.: mm.m mm.m whooz Amaze hem: magnum mafion< qums¢ m>mpm weepmpsaah deem Oowno wnowewm cosh .8352. maqu mammamm comm GOHHHQ Hamwnaz anpqwamm pumpa¢ pmh aamaaano :H 8.: aa 03 Nm.: Nb.N 8.: ma NH HA No.2 mm.m om.m Ho.: OH ma.: ::.m mm.: ma.m om.N m:.m em.: mw.m 0m.: Hm.m m:.: m mm.m :m.: m:.: we.: mm.: mm.: mm.m e :w.m Hm.m mm.m mo.: ew.m NH.: NN.: m :e.m mo.: ma.m em.: m:.m 00.: m:.m Hm.m m:.m m mm.m ma.: mm.m 0:.: 00.: m:.: me.m mm.m :m.m mw.m : mm.m >m.: sm.m mm.: Hm.m Hm.: :0.: :m.m mm.m em.m mm.m m mmaaamh HH<.nom Neapwz moqumflm mm.: mm.m Hm.: em.m em.m H:.m mm.: em.: :m.m mm.: 0m.: m:.: Hm.: :0.: mm.: Hm.m m:.m mm.m om.m ma.: em.m mm.m 0m.: :0.: am.m whooz peaks hhmz meqsm manage sapm5¢.m>mpm mQOpmeHHm_eonm OUHSU enomqam Umhm som-qnoe enema whoxsem 0mmm soaaam HaSmnmz anpsaaam pumpa<_pam cama:::0 :H Hm.m me.m :0.: e:.: we.m ::.: MH NH HH mm.m :0.m mw.m :e.m OH mm.m Hm.N NH.: H®.N mN.N o:.m 00.: mm.m m0.: we.m mH.: w mm.m :m.: we.m 0a.: m0.: ma.: m0.m a we.m mm.m mm.m me.m :e.m 00.m mw.m e:.: w mm.m mw.m mw.m :m.m om.m Hw.m mm.m :m.m m:.m m mw.m mm.m :o.m mH.: me.m mm.m wm.m HN.m mm.m Om.N : me.m 00.: mw.m Hm.: ew.m mm.m :0.m we.m mm.m mm.m we.m mmHeE HH¢ new xfiupez moqumfin OH.: we... 9...: e3 H©.m mam MH.: 5.: cam 8.: ON.: ::.: 00.: mm.m 50.: mm.m :m.m mm.m 00.m ww.m :m.m Hm.m :H.: be.m mw.m 09002 anhe >902 meqsm menon¢ cflpms¢_m>mpm oQOPmeHHm deem ooano dHOMme comm som-anoe «Adam ommxsem doom GOHHHQ He£mpaz enpqefimm pnoQHd new camnaafl0 APPENDIX E Group Questionnaire for Unidimensional Measures Here are some questions about the people and shows on TV. Please read the questions and then put an'fi on the line above what you think. I The questions are about what you think of people on TV. Here is an example. How funny do you think Fred Flintstone is? very funny not very . not funny funny funny at all Your Name: (7-9) Grade: Room Number: ‘vv v vfirv va ' V‘v ' HERE ARE SOME QUESTIONS ABOUT GILLIGAN ON "GILLIGAN'S ISLAND." THINK ABOUT THEM AND THEN PUT AN _)_(_ON THE LINE ABOVE WHAT YOU THINK. 1. How often do you watch the TV show "Gilligan's Island?" almost most of some of not at always the time the time . all 2. How funny do you think Gilligan is? very funny not very not funny funny . funny at all 3. How active do you think Gilligan is? very active not very not active active active at all u. How good looking do you think Gilligan is? very good not very not good good looking good looking looking A looking at all 5. How strong do you think Gilligan is? very strong not very not strong strong strong ' at all 6. How much like a real person is Gilligan? very much like a not very not like a like a real much like a real person real person person real person at all 7. How good do you think Gilligan is? very good not very not good good good at all 8. How old do you think Gilligan is? very old young very old young 9. How much do the other people on "Gilligan's Island" like Gilligan? a lot a l ttle not very not at all much l0. How much do you want to be like Gilligan? a lot a little not very' not at much all 11. Are there things that Gilligan does that you would like to do? a lot of some almost nothing things things nothing at all Card 3 Col. # 10 11 12 13 1k 15 16 17 18 19 2O HERE ARE SOME QUESTIONS ABOUT SAMANTHA ON "BEWITCHED." THINK ABOUT THEM AND THEN PUT AN §_ON THE LINE ABOVE WHAT YOU THINK. 1. How often do you watch the TV show "Bewitched?" almost most of some of not at always the time the time all 2. How funny do you think Samantha is? very funny not very not funny funny funny at all 3. How active do you think Samantha is? very active not very not active active active at all u. How good looking do you think Samantha is? very good not very not good good looking good looking looking looking at all 5. How strong do you think Samantha is? very strong ‘ not very not strong strong . strong at all 6. How much like a real person is Samantha? very much like a not very not like a like a real much like a real person real person person real person at all 7. How good do you think Samantha is? very good not very not good good good at all 8. How old do you think Samantha is? very old young very old Young 9. How much do the other people on "Bewitched" like Samantha? a lot a little not very not at much all 10. How much do you want to be like Samantha? a lot a little not very not at much all 11. Are there things that Samantha does that you would like to do? a lot of some almost nothing ' things things nothing at all Col. H 21 22 23 2Q 25 26 27 28 29 3O 31 g HERE ARE SOME QUESTIONS ABOUT MARSHAL]: DILLON 0N "GUNSMOKE." THINK ABOUT THEM AND THEN PUT AN 5 ON THE LINE ABOVE mm 3n YOU THINK. 1. How often do you watch the TV show "Gunsmoke?" almost most of some of not at always the time the time all 2. How funny do you think Marshall Dillon is? very funny not very not funny funny funny at all 3. How active do you think Marshall Dillon is? very active not very not active active active . 4 at all u. How good looking do you think Marshall Dillon is? very good not very not good good , looking good looking looking looking at all 5. How strong do you think Marshall Dillon is? very strong not very not strong strong strong at all 6. How much like a real person is Marshall Dillon? very much like a not very not like a like a real much like a real person real person person real person at all 7. How good do you think Marshall Dillon is? very good not very .not good good good at all 8. How old do you think Marshall Dillon is? very old old , young very young 9. How much do the other people on "Gunsmoke" like Marshall Dillon? - 271—0?- me my 527? much all 10. How much do you want to he like Marshall Dillon? mt— me my me— much all ll. Are there things that Marshall Dillon does that you would like to do? a lot some almost nothing of things things nothing at all Col. # 82 33 35 36 37 38 39 Q0 “1 1.2 HERE ARE SOME QUESTIONS ABOUT HAWKEYE ON "HASH.” THINK ABOUT THEM AND THEN PUT AN.§ ON THE LINE ABOVE WHAT YOU to do? a lot of some almost things things nothing THINK. 1. How often do you watch the TV show "MASH?" almost most of some of not at always the time the time all 2. How funny do you think Hawkeye is? very funny not very not funny funny funny at all 3. How active do you think Hawkeye is? very active not very not active active active at all u. How good looking do you think Hawkeye is? very good not very not good good looking good looking looking looking at all 5. How strong do you think Hawkeye is? very strong not very not strong strong strong at all 6. How much like a real person is Hawkeye? very much like a not very 362 like a like a real much like a real person real person person real person at all 7. How good do you think Hawkeye is? very good not very not good good good at all 8. How old do you think Hawkeye is? very old young very old young 9. How much do the other people on ”MASH" like Hawkeye? a lot a little not very not at much all 10. How much do you want to be like Hawkeye? a lot a little not very not at much all 11. Are there things that Hawkeye does that you would like nothing at all Col. 5 #3 an #5 us #7 #8 “9 SO 51 S2 53 1. THE COSBY KIDS." LINE ABOVE WHAT YOU THINK. HERE ARE SOME QUESTIONS ABOUT FAT ALBERT ON "FAT ALBERT AND THINK ABOUT THEM AND THEN PUT AN M'ON THE How often do you watch the TV show ”Fat Albert and the Cosby Kids?" would like to do? a lot of some almost things things nothing almost most of some of not at always the time. the time all 2. How funny do you think Fat Albert is? very funny not very not funny funny funny at all 3. How active do you think Fat Albert is? very active not very not active active active at all Q. How good looking do you think Fat Albert is? very good not very not good good looking good looking looking looking at all 5. How strong do you think Fat Albert is? very strong not very not strong strong strong at all 6. How much like a real person is Fat Albert? very much like a not very not like a like a real much like a real person real person person real person at all 7. How good do you think Fat Albert is? very good not very not good good good at all 8. How old do you think Fat Albert is? very old young ' very old young 9. How much do the other people on QFat Albert and the Cosby Kids" like Fat Albert? a lot a little not very not at much all 10. How much do you want to be like Fat Albert? E'To?’ E-li—tt—fe my STE—at- much all 11. Are there things that Fat Albert does that you nothing at all Col. # 5M 55 56 57 58 59 60 61 62 63 6M HERE ARE SOME QUESTIONS ABOUT REED ON "ADAM-l2." THINK ABOUT THEM AND THEN PUT AN §_ON THE LINE ABOVE WHAT YOU to do? a lot of some almost things things nothing THINK. 1. How often do you watch the TV show "Adam-12?" almost most of_ some of not at always the time the time all 2. How funny do you think Reed is? very funny not very not funny funny funny at all 3. How active do you think Reed is? very active not very not active active active. at all u. How good looking do you think Reed is? very good not very not good good looking good looking looking looking at all 5. How strong do you think Reed is? very strong not very not strong strong strong at all 6. How much like a real person is Reed? very much like a not very not like a like a real much like a real person real person person real person at all 7. How good do you think Reed is? very good not very not good good good at all 8. How old do you think Reed is? very old young very old young 9. How much do the other people on "Adam-l2" like Reed? a lot a little not very not at much all 10. How much do you want to be like Reed? a lot a little not very not at ‘ - much all“. ll. Are there things that Reed does that you would like nothing at all Col. .65 66 67 68 69 70 71 72 73 7“ 75 fl THE PRAIRIE." LINE ABOVE WHAT YOU THINK. HERE ARE SOME QUESTIONS ABOUT LAURA ON "LITTLE HOUSE ON THINK ABOUT THEM AND THEN PUT AN §_ON THE to do? - a lot of some almost things things nothing 1. How often do you watch the TV show "Little House on the Prairie?" almost most of some of not at always the time the time all 2. How funny do you think Laura is? very funny not very not funny funny funny at all 3. How active do you think Laura is? very active not very not active active active at all u. How good looking do you think Laura is? very _ good not very not good good looking good looking looking looking at all 5. How strong do you think Laura is? very strong not very not strong strong strong at all 6. How much like a real person is Laura? very much like a not very not like a like a real much like a real person real person person real person at all 7. How good do you think Laura is? very good not very not good good good at all 8. How old do you think Laura is? very 'old young very old young 9. How much do the other people on "Little House on the Prairie" like Laura? a lot a little not very not at much all 10. How much do you want to be like Laura? a lot a little not very not at much all 11. Are there things that Laura does that you would like nothing at all Card u Col. # 10 11 12 13 1k 15 16 17 18 19 2O HERE ARE SOME QUESTIONS ABOUT JOHN-BOY IN "THE WALTONS." THINK ABOUT THEM AND THEN PUT AN XION THE LINE ABOVE WHAT YOU THINK. I. How often do you watch the TV show "The Waltons?" almost most of some of not at always the time the time all 2. How funny do you think John-Boy is? very funny not very not funny funny funny at all 3. How active do you think John—Boy is? very active not very not active active active {at all Q. How good looking do you think John-Boy is? very - good not very not good good looking good looking looking looking at all 5. How strong do you think John-Boy is? very strong not very not strong strong strong at all 6. How much like a real person is John-Boy? very much like a not very not like a like a real much like a real person real person person real person at all 7. How good do you think John-Boy is? very good not very not good good good at all 8. How old do you think John-Boy is? very old young very old young 9. How much do the other people on "The Waltons" like John-Boy? a lot a little not very not at much all 10. How much do you want to be like John-Boy? a lot a little not very ' not at much all 11. Are there things that John-Boy does that you would like to do? a lot of some almost things things nothing nothing at all Col. # 21 22 23 2a 25 26 27 28 29 3O 31 HERE ARE SOME QUESTIONS ABOUT FRED SANFORD 0N "SANFORD AND SON?" THINK ABOUT THEM AND THEN PUT AN §_ON THE LINE ABOVE HHAT YOU THINK. - 1. How often do you watch the TV show "Sanford and Son?" almost most of some of not at always the time the time all 2. How funny do you think Fred Sanford is? very funny not very not funny funny funny at all 3. How active do you think Fred Sanford is? I very active not very not active active active at all u. How good looking do you think Fred Sanford is? -very good not very not good good looking good looking looking looking ' at all 5. How strong do you think Fred Sanford is? very strong not very not strong strong strong at all 6. How much like a real person is Fred Sanford? very much like a not very not like a like a real much like a real person real person person real person at all 7. How good do you think Fred Sanford is? very good not very not good good good at all 8. How old do you think Fred Sanford is? very old old young very ' young 9. How much do the other people on "Sanford and Son" like Fred Sanford? a lot a little not very not at much ‘ all 10. How much do you want to be like Fred Sanford? a lot a little not very not at much all 11. Are there things that Fred Sanford does that you would like to do? a lot of some almost nothing things things nothing at all Col. # 32 33 3h 35 36 37 38 39 “0 M1 92 HERE ARE SOME QUESTIONS ABOUT STEVE AUSTIN ON "THE SIX MILLION DOLLAR MAN." THINK ABOUT THEM AND THEN PUT AN §_ON THE LINE ABOVE WHAT YOU THINK. I. How often do you watch the TV show "The Six Million Dollar Man?" almost most of some of not at always the time the time all 2. How funny do you think Steve Austin is? very funny not very not funny funny funny at all 3. How active do you think Steve Austin is? very active not very not active active active at all u. How good looking do you think Steve Austin is? very good not very not good good - looking good looking ‘ looking looking at all 5. How strong do you think Steve Austin is? very strong not very not strong strong ' strong at all-‘ 6. How much like a real person is Steve Austin? very much like a not very ' not like a. like a real much like a real person real person person . real person at all 7. How good do you think Steve Austin is? very good not very not good good good at all 8. How old do you think Steve Austin is? very old young very old young L 9. How much do the other people on "The Six Million Dollar Man" like Steve Austin? a lot a little not very not at much all 10. How much do you want to be like Steve Austin? a lot a little not very not at ‘ much all ll. Are there things that Steve Austin does that you would like to do? a lot of some almost nothing things things nothing at all Col. # M3 NM NS “6 N7 88 N9 50 51 S2 53 0"-“ .flhnid '- .. —-- l . - o It‘nn'v‘l HERE ARE SOME QUESTIONS ABOUT FRED FLINTSTONE ON "THE FLINTSTONES." THINK ABOUT THEM AND THEN PUT AN §_ON THE LINE ABOVE WHAT YOU THINK. 1. How often do you watch the TV show "The Flintstones?" almost most of some of not at always the time the time all 2. How funny do you think Fred Flintstone is? very funny - not very not funny funny funny at all 3. How active do you think Fred Flintstone is? very aetive not very not active active active at all 4. How good looking do you think Fred Flintstone is? very good not very not good good looking good looking looking looking at all 5. How strong do you think Fred Flintstone is? very strong not very not strong strong strong at all 6. How much like a real person is Fred Flintstone? very much like a ’not very not like a like a real much like a real person real person person real person at all 7. How good do you think Fred Flintstone is? very good ' not very not good good good at all 8. How old do you think Fred Flintstone is? very old .young very old young 9. How much do the other people on "The Flintstones" like Fred Flintstone? a lot a little not very not at much all 10. How much do you want to be like Fred Flintstone? 370—?“ me W W much all ll. Are there things that Fred Flintstone does that you would like to do? a lot of some almost nothing things things nothing at all Col. # 5U 55 56 S7 58 $9 60 61 62 63 6“ HERE ARE SOME QUESTIONS ABOUT ARCHIE BUNKER ON "ALL IN THE FAMILY." THINK ABOUT THEM AND THEN PUT AN §_ON THE LINE ABOVE WHAT YOU THINK. 1. How often do you watch the TV show “All in the Family?" almost most of some of not at always the time the time all 2. How funny do you think Archie Bunker is? very funny not very not funny funny .funny at all 3. How active do you think Archie Bunker is? very active not very not active active _ active at all u. How good looking do you think Archie Bunker is? . very good not very not good good looking good looking looking looking at all 5. How strong do you think Archie Bunker is? very strong not very not strong strong strong at all 6. How much like a real person is Archie Bunker ? very much like a 'not very not like a like a real ' much like a real person real person -erson real person at all 7. How good do you think Archie Bunker is? very good not very not good good good at all 8. How old do you think Archie Bunker is? very old young very old young 9. How much do the other' people on "All in the Family" like Archie Bunker? :13;— me W W s~--..A . much all 10. How much do you want to be like Archie Bunker? 3'12?— m. m ass—a? much all ll. Are there things that Archie Bunker does that you would like to do? a lot of some almost nothing things things nothing at all Col. # 65 66 67 68 69 7O 71 72 73 7M 75 10‘ HERE ARE SOME QUESTIONS ABOUT CHICO ON ”CHICO AND THE MAN.” THINK ABOUT THEM AND THEN PUT AN §_ON THE LINE ABOVE WHAT YOU THINK. How often do you watch the TV show "Chico and the Man?“ like to do? a lot of things almost not at always all 2. How funny do you think Chico is? very not funny funny at all 3. How active do you think Chico is? IE537_—— ESE—EEETve active at all u. How good looking do you think Chico is? very not good good looking looking at all 5. How strong do you think Chico is? very not strong strong at all 6. How much like a real person very much not like a like a much like a real person real person real person at all 7. How good do you think Chico ' very not good good at all 8. How old do you think Chico is? very very old young 9. How much do the other people on “Chico and the Man" like Chico? a lot not at all l0. How much do you want to be like Chico? a lot not at all 11. Are there things that Chico does that you would nothing at all Card 5 Col. # 10 11 12 13 1k 15 16 17 18 19 2O . 1. HERE ARE SOME QUESTIONS ABOUT MARY TYLER MOORE ON “THE MARY TYLER MOORE SHOW." THINK ABOUT THEM AND THEN PUT AN 5.0N THE LINE ABOVE WHAT YOU THINK. How often do you watch the TV show "The Mary Tyler Moore Show?" almost most of some of not at always the time the time all 2. How funny do you think Mary Tyler Moore is? very funny not very not funny funny funny at all 3. How active do you think Mary Tyler Moore is? very active not very not active active ‘ active at all u. How good looking do you think Mary Tyler Moore is? very good not very not good good looking good looking looking looking at all 5. How strong do you think Mary Tyler Moore is? very strong not very not strong strong strong at all 6. How much like a real person is Mary Tyler Moore? very much like a not very not like a like a real much like a real person real person person real person at all 7. How good do you think Mary Tyler Moore is? very good not very not gaod good good at all 8. How old do you think Mary Tyler Moore is? very old young very old. young 9. How much do the other people on ”The Mary Tyler Moore Show" like Mary Tyler Moore? a lot a little not very not at much all 10. How much do you want to be like Mary Tyler Moore? a lot a little not very not at much all 11. Are there things that Mary Tyler Moore does that you would like to do? a lot of some almost nothing 'things things nothing at all Col. # 21 22 23 24 25 26 27 28 29 3O 31 HERE ARE SOME QUESTIONS ABOUT YOURSELF. THINK ABOUT THEM AND THEN PUT AN §_ON THE LINE ABOVE WHAT YOU THINK.‘ 1. How funny do you think you are? very funny not very not funny funny ' funny at all 2. How active do you think you are? very active not very not active active active at all 3. How good looking do you think you are? very good not very not good good looking good looking looking looking at all u. How strong do you think you are? very strong not very not strong strong strong at all 5. How much do you think you are like other people? a lot a little not very not at much all 6. How good do you think you are? very good not very not good good good at all 7. How old do you think you are? very old young very old young 8. How much do other people like you? a lot a little not very not at much all 9. How old are you 10. What date is your birthday on? ll. Hhat grade are you in? 12. What is your sex? boy girl Col. # 32 33 3M 35 36 '37 38 39 .oapwofiammo pmofi mos mpsnfimppo esp opooflonfi modad> nmswfln one .: Op H Seem comedy mamas madam Haas m xHQmem¢ wo.m mm.m om.m mm.m Hm.m mH.m mo.m :m.m whooz Lease sees Hm.m Ha.m om.m mo.m mm.m oa.m ms.m :m.m possum manure mo.m sm.m :m.m Hm.m mw.m om.m ms.m :m.m nflpmsa.m>mpm sm.m mw.m mw.m ma.m :m.m mo.m mm.m oa.m mgopmpsflas ewes mm.m ma.m mm.m mm.m mo.m sa.m mm.m :m.m conga ms.m mm.m so.m mm.m mm.m Ha.m os.m 0:.m enoeaem ewes se.m ma.m mm.m mH.m mm.m :m.m mo.m ma.m som-qaos ow.m mm.H mm.m mm.m Ha.m ms.m mo.m wm.m eases m:.m :m.m sm.m ma.m sw.m mm.m :m.m as.m msmxsam sm.m mm.m mm.m mm.m mH.m ma.m mm.m mo.m emmm mm.m mm.m ma.m Ha.m Hm.m oo.m sm.m :s.a aoHHao deemsaz mm.m mm.m om.m Ha.m om.m sa.m Ha.m mm.m eeuqenmm Hw.m ss.s om.m :m.m om.m as.a ms.m mo.m pnmpaa.pem mm.m sm.a sH.m os.m ms.H ::.m sm.m mm.m nmmaaaao opossum mma eooo spflammm someonpm wgflxooq msepoa sense meme coco nopomnono omopsofispp< HdQOHmooSfloch pswflm no mempoonoso nowmfl>oama soopHSOm hoe mohoom new: APPENDIX G Plots of Two-Dimensional Comparisons of Fourteen Television Characters (Total Sample) The plots represent all two-dimensional comparisons not presented in Chapter 3. The INDSCAL program does not compute totally orthogonal dimensions. The dimensions in the following graphs are drawn as orthogonal for clarity of presentation. The correlations between the dimensions are in Table 2. Dimension 1 and Dimension 3 DIM 3 o S amantha 0 Mary Tyler Moore Hawkeye Q Steve Austin 0 Reed Fred . Chico . Laura 0 Sanford. DIM l O ..Archie Marshal Dillon Bunker ' o Gilligan John—Boy Fat Fred 0 9 Albert Flintstone Dimension 1 and Dimension h DIM 1+ Gilligan o John-Boy 0 Laura Chico o o Hawkeye Reed 0 . . Steve Austin Samantha DIM l C Marshal Dillon 0’ Fred Sanford Mary Tyler Moore 0 Archie Fat Bunker . . . Albert Fred Flintstone Dimension 2 and Dimension 3 DIM 2 Steve Austin Marshal 0 Dillon . Reed. . .Hawkeye Chico 0 Fred Sanford Archie Bunker o DIM 3 Fat Albert 0' 0 Fred Flintstone Gilligan 0 John -Boy S amantha Laura Mary Tyler Moore 0 Dimension 2 and Dimension h DIM h Gilligan O O John-Boy Laura . Hawkeye Chico. Samantha . 0 Steve Austin . Reed DIM 2 O Mary Tyler Moore Fred Flintstone(l) Fat " Albert (r) O Marshal Dillon . Fred Sanford ..Archie Bunker Dimension 3 and Dimension 1+ Gilligan <0 0 J ohn-Boy Laura I Ch'co Hawksye Samantha 1 . 0 Steve 0 0 Reed Austin DIM 3 Marshal Dillon 0 Mary Tyler Moore 0 Fred Sanford Fat .. Archie Albert Bunker 0 Fred F lint st one APPENDIX H INDSCAL Coordinates for Fourteen Television Characters on Four Dimensions (Total Sample)a DIM 1 DIM 2 DIM 3 DIM h Gilligan -.238 -.597 -.268 +.h68 Samantha .108 -.308 +.l+86 +.081 Marshal Dillon .3uu +.3oo -.033 -.oh2 Hawkeye -.l9l +.266 +.l96 +.lh5 Fat Albert -.172 -.056 —.u5u -.uo7 Reed .2h7 +.298 +.O68 +.106 Laura .325 -.h85 +.066 +.193 John-Boy .299 -.180 -.263 +.35h Fred Sanford -.372 +.J_'L3 +.051 -.257 Steve.Austin .296 +.353 +.225 +.O77 Fred Flintstone -.255 -.066 -.1+37 -.379 Archie Bunker -.309 +.033 -.068 -.h10 Chico -.263 +.222 +.108 +.11+2 Mary Tyler Moore ' .180 -.h31 +.322 -.073 8The dimensions are normalized so that the sum of the squared coordinates on eaCh dimension equals 1.00. APPENDIX I Zero Order Correlations Between Nine Unidimensional Ratings of Television Characters for Each Grade and Sex Appendix I contains the zero order correlations between the nine unidimensional rating for each grade and sex. There correlations are based on the mean rating of each of fourteen TV characters on nine attributes. The n for the correlations is therefore, feurteen, the number of TV Characters. The mean value of each attribute for each character is, however, collapsed across sub- jects. The attribute value for each Character is then based on the number of subjects in the subroups. '\1 0\ UT 47' 00 Zero Order Correlations for Third Grade Among Nine Unidimensional Ratings of Fourteen Television Characters Funny - Active -.27 - Good looking -.35 ,7h** - Strength —.A3 .68*- .50 - Reality -.27 .07 .20 -.01 - Good -.A6 .75** .87** .5h* .ul - Age .IA -.02 -.12 .23 .01 -.1A - Support -.h8 .06 .2A .21 .19 .29 -.56* Sex (of character) -.23 -.27 .25 -.A3 -.06 .07 -.38 *P