.nll'lllll'l'llnllnf L ‘ F‘ .1 R Y M. gar]. State University This is to certify that the thesis entitled The Dimensions of Persuasion Situations: Implications For Communication Research and Assessments of Taxonomy Constructing Methodologies presented by Michael John Cody has been accepted towards fulfillment of the requirements for Ph. D. degree in Communication AEZ~“h;-G?'(Eiett4evy (:‘*””° Major professor Date November 3) 1977 0-7639 I: d'fli'JMI i fits/W c, THE DIMENSIONS OF PERSUASION SITU TIONS: IMPLICATIONS FOR COMMUNICATION RESEARCH AND ASSESSMENTS OE TAXONOMY CONSTRUCTING METHODOLIGIES By Michael John Cody A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY '- Department 0: Communication 1978 DEDICATION For my Father ii “EST Rim. T THE DIME NSIOi'S OF PEPSUASI ON SITUAMFIO : IMPLICATIONS FOR COMMUNICATIONL YESEARCH AND ASSESSI iENTS OE TAXOIOMY ONSTR-C ING METHOD’LO’GIES The focus of the present study was to inIestiga-e the percept LlO us of persuasion situations and the question of trans-situational consisten- cy of behavior. Four emp1r1cal approaches to the queStion of transitu- O Q ati onal consistency of behavio were identified and critic we (a) the ’ V predictive utilit ' of trait measureS' tb) tne "sources of variance” rara- - ’ 1‘ LL L 14 igm;( o) the moderator variable approach; and, i ) the behavior observa— tic n approach. It was argued that prediction of behavior will be enhanced only under circumstances where prior knowledge of how individuals per- ceive the behavioral requ1rements of situations are obtained. Thus, re— search concerning taxonomies of situations was provided and an explora— tory study was conducted in order to identify the maior characteristics of the perceptions of persuasion situations. A study was conducted in order to provide a taxonomy of situations. Several methodological questions concerning procedures used in derivin tzxonomies were identified: (1) ouestions oertainin to 51m 1. ar- e - ies between factor analysis and multidimensional scaling; (2) similar- tias amo n various multidimensional scaling techn'tues- and (3) sim- .J 9 P- t« larities between cluster analysis and multidimensional scaling. In order (.1. to rovide assessments of the aeneralit‘ of imensions comparisons were D 9 q made between faCtor analusis azid mult idiirensional scaling solutions and Michael John Cody between multidimensional scaling solutions and cluster analysis solu- tions. Another set of methodological questions anus addressed: differ- ences in scale usage and the replication of factors. Pretest data was used in order to construct a population of situations from which random samples of situations were drawn. Three sets of situations were random— ly sampled and three sets of respondents were provided with a paired- comparison questionnaire with an interval scale for rating dissimilari- ty. A fourth group of respondents received one of the sets of situa— tions with a ratio scale. Thus, a comparison of the rating scale was made. All respondents rating situations in their respective sample on 23 attributes. The factor analysis solution provided six factors: formal vs. in- formal, superficial vs. intimate, influencibility, I benefit vs. I do not benefit, easy to persuade vs. difficult, and rights vs. favors. The multidimensional scaling solutions for each set of data indicated that four dimensional solutions were sufficient to account for major portions of the respective variances. The §Z§E_and GALILEO solutions were simi- lar, but the INDSCAL solutions failed to correspond to the other con- figurations. Several assumptions of the INDSCAL model were not support- ed. Three dimensions were replicated across the KYST solutions: in- tense feelings, superficial vs. intimate and easy to persuade vs. dif— ficult. Other dimensions obtained in the multidimensional scaling an- alyses included: subordinate vs. social affiliation, importance of goal, benefits, and familiarity. Only minor differences were obtained in the Michael John Cody interpretation of the solutions for the interval and ratio scaled data. The cluster analysis solutions were found to be compatible with the KYST interpretations. A number of divergent points of view in the perceptions of situations were obtained that could not be attributed to a sex differ- ence. A discussion of the outcomes of jointly employing factor analytic and multidimensional scaling techniques was provided. A discussion of the limitations of employing the INDSCAL model was also provided. Sev- eral avenues for future research were discussed. Accepted by the faculty of the Department of Communication, College of Communication Arts, Michigan State University, in partial fulfillment of the require- ments for the Doctor of PhilosoPhy degree. {,5an yaw. Director of Dissertation ‘ 6 t I. I "7 . x ‘v '16”) 4. 1.. r ‘.v’/ C. r L[MC/t W 'U/ -‘- ‘4 7 Guidance Committee: , Chairman Th1] Maccmp / €112,444 JALMGHSCWL ACNKOWLEDGMENTS The list of all of the individuals who contributed to this volume is simply too long to provide. All of my students who shared their ex- periences and strategies with me should be noted. Thank you very much. Special thanks must go to Dr. Miller and to Frank Boster and Ed Kaminski for work reported in Chapter 1. Thanks must also go to Drs. Parks and Marlier for their input into this project when it was a young and inno— cent, little study. I must also thank my two roommates at the farmhouse for the type of social support all doctoral candidates need during the long-drawn out process of continuous frustration and during the prolonged inhumane mara— thon of writing: Carter Clary and Steven Thomas. The two of you will probably never read this note, but I want everyone to know of my grati- tude to you both. If only you would have kept the upstairs bathroom clean- er, we would have had a perfect world. Of course, many of the doctoral candidates at Michigan State Uni- versity also deserve to be acknowledged: Keith Adler, Gordon Dahnke, Nadyne Edison, Jim DinKelacKer, Elyse Werner, Tom Mwanika, Mary Hines, Betty Schweitzer, Carol Stein, Nicky Stoyohoff, Michael Sunnafrank, Laura Henderson, and Steve McDermott. Hopefully, I have not missed anyone with this general acknowledgement. In general, I should thank everyone in that academic arena who had to put up with my apparent anti-social ten- dencies during the data collection period and during those casual moments when I simply factored too many matrices, conducted just too many multiple iii regression runs and simply wrote too many pages of text. Special thanks goes to Darlyne at Odessa College in Texas and to Dr. Jordan at Texas Tech for their encouragment and continued support. ' When I think of what this dissertation might have been like with- out the aid and infinite amount of support offered by Kathy Clyde, I simply panic. Kathy Clyde did another excellent job with all of the ta- bles and figures and helped edit the text. I can quite truthfully say that this dissertation would not be a good one (if, indeed, it is) with- out the quality of Kathy's help, not to mention her sheer endurance. Also, Kathy Sherry helped to correct errors in the text, typed nearly all of the text, corrected poor English when she found it (which was not sel- dom) before Dr. Bettinghaus had a chance to do his editing work. I thank both of you. Dr. Bettinghaus is one of the most patient individuals I know (aside from my father). There is simply no way of recalling how many times he read and re-read Chapters 1 and 3, nor is it possible to thank him for all the direction he gave and all of the ideas he provided that were in- corporated throughout the dissertation. His work support and encourage- ment, his social support and the financial support he extended during my stay at Michigan State University, especially during the last year, sim- ply made this whole project possible. Drs. Simmons and Woelfel also made considerable inputs into this work, as well as into so much of my work in general. t it were not for my interaction with these two individuals over the last several years, this dissertation would have focused on far less general issues and would have entailed implications far more limited in scope. Drs. Atkins and Tompson also played valuable roles. I thank Dr. Thompson in particular, iv as well as Dr. Phillips. for help in interpreting the most seemingly un— interpretable dimensions, and for trying to interpret dimensions that really were uninterpretable. 'Finally, I need to thank an individual who taught me a lot of neat tricks and, hopefully, will continue to teach me new ones. Unfortunately, Rick Holmes did not want his name to appear in this acknowledgement. Therefore, I am afraid to erase the above note which clearly identified Rick as one of my clos- est friends and which clearly identified him as an individual whose con- stant help, advice, and encouragement provided me with the ability to go to battle with the computer "scenter." After all of the dinners, lunches, breakfasts, snacks, brunches, empty bottles of wine and all-nighters-— after all of those equation-filled napkins bit the dust--after all of this, Rick, I can finally, finally say it: "My time has come." TABLE OF CONTENTS Chapter Page 1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . 1 1.0 Statement of Purpose . . . . 4 1.1 Heuristic Value and Pragmatic Research Utility. 6 1.2 Trans- situational Consistenct Across What Types of Situations, for What Types of Modes or Re- sponses, for What Types of Persons? . . . . . . 11 1.2.1 Overview . . . . . . . . . . . . . . . . . . 11 1.2.2 The "Sources of Variance" Paradigm . . . . . 14 1.2.3 The "Moderator Variable" Approach, the "Behav- ioral Observation" Approach, and the Idiosyn- cratic Behavioral Patterns . . . . . . . . . . 26 1.2.3.1 Overview . . . . . 26 1.2.3.2 The "Moderator Variable" Approach and the "Behavior Observation" Approach . . . . . . 29 1.2.3 3 Mischel's Social Learning Theory Reconcept- ualization . . . . . . . . . . . . . . . . . 38 1.3 Summary . . . . . . . . . . . . . 48 1.4 Perceptions of Situations'. A Literature Review . 49 1.5 Summary . . . 67 1.6 Exploratory Examination of the Perceptions of Persuasion Situations . . . . . . . . . . . . . 69 1.6.1 Initial List of Situations . . . . . . . . . . 7 1.6.2 Construction of Situations . . . . . . . . . . 70 1.6.2 1 Materials and Respondents . . . . . . . . . 72 1.6.2.2 Procedures . . . . . . . . . . . . . . . . . 72 1.6.3 Results . . . . . . . . . . . . . . . . . 72 1.6.3.1 Interview Results . . . . . . . . . . . . . 79 1.6.3.2 Cluster Analysis Results . . . . . . . . . . 80 1.6.3 3 Multidimensional Scaling Results . . . . . . 85 1.6.4 Discussion . . . . . . 87 1.7 Criterion for Assessing "Generality" of Dimen- sions Obtained by Multidimensional Scaling Techniques . . . . . . . . . . . . . . . 89 1.7.1 Effects of Aggregation . . . . . . . . . . . . 93 1. 8 Conclusions and Implications . . . . . . . . . . 96 NOTES . . . . . . . . . . . . . . . . . . . . . . . . . 98 2 2.1 Pretest for Nominating Situations . . . . . . . 99 2.2 Selection of Situations . . . . . . . . . . . . 100 2.3 Procedures Employed in the Study . . . . . . . . 102 2.3.1 Materials . . . . . . . . . . . . . . . . . . 102 2.3.2 Procedures . . . . . . . . . . . . . . . . . . 105 2.3.3 Respondents . . . . . . . . . . . . . . 106 2.4 Analysis of Respondents' Reactions to the Study. 106 vi Chapter Page 3 RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . 112 Overview . . . . . . . . 112 3.1 Factor Analysis of Unidimensional Attributes . . 114 3. 2 An Assessment of the Stability of Situations in the Multidimensional Configurations . . . . . . 124 3.2.1 Reliability of the Paired Comparison Data: Assessing Reliability Between Male—Female Configurations . . . . . . . . 137 3. 2. 2 Comparison of Form 3 and Form 4 Solutions . . . 152 3. 3 The Number of Dimensions Problem Resolved . . . . 153 3. 4 Dimensional Structure: Results of the Four Dim- ensional (KYST) Solutions . . . . . . . . . . . 168 3.5 The Hierarchical Structure of Situations: Re- sults of Johnson' 8 (1967) Hierarchical Cluster— ing of Situations . . . . . . . . 195 3.6 Dimensional Structure. Results of INDSCAL Analysis . . . . . . . . . . . . 208 3. 7 Investigation of Individual Differences . . . . . 227 Summary . . . . . . . . . . . . . . . . . . . . . . . 239 NOTES . . . . . . . . . . . . . . . . . . . . . . . . . . 244 4 DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . 245 4.1 Methodological Importance . . . . . . . . . . . . 245 4.2 Theoretical Significance . . . . . . . 248 4.3 Summary and Implications of the Dimensions of Situations . . . . . . . . . . . . . . . . . . 251 4.4 Future Research . . . . . . . . . . . . . . . . . 252 4.4 Summary . . . . . . . . . . . . . . . . . . . . . 2S4 APPENDICES Appendix A Initial List of Situations . . . . . . . . . 256 Appendix B Instructions for Sorting Task Pretest . . . . . . 261 Appendix C List of Nominated Situations . . . . . . . . . . 262 Appendix D Instructions for Assumed Interval Scale . . . . . 267 Appendix E Instructions for Assumed Ratio Scale . . . . . . 269 Appendix F Order by Which Attributes Were Presented . . . . 272 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 vii Table 10. 11. 12. 13. 14. 15. 16. 17. LIST OF TABLES Forty Situations Used in Sorting Task Pretest . Attributes Used to Interpret Dimensions Obtained by Wish, Deutsch, and Kaplan (1976), and by Forgas (1976). Results of Factor Analysis of Unidimensional Attrib- utes (n = 263) . . . . . . . . . . . . . . . . . Distances Between Situations in Split-Half Comparisons. Column Correlations for Split-Half Comparisons for Eight Dimensions . . . . . . . . . . . Row Correlations for Split-Half Comparisons . Distances Between Situations in Male-Female Comparisons . Column Correlations for Male-Female Comparisons for Eight Dimensions . . . . . . . . . . . . Row Correlations for Male-Female Comparisons Stress Values for Dimensions One to Five for Each of the Four Questionnaire Types (KYST Analysis) . Correlations Between Input Data and Predicted Scores and Percent of Variance Accounted for by Two through Six Dimensional Solutions (INDSCAL Anlysis) Percent of Variance Accounted for by Each of Eight Dimensions (GALILEO Analysis) . Coordinates of 12 Situations (Form 1) in Four Dimensions from KYST Analysis (Stress = 4.65%) Standardized Regression Weights for Four (KYST) Dimens- ions Used to Predict Mean Ratings on Unidimensional Attributes (Form 1; n = 64) . . . . . . . . . . . . Coordinates of 12 Situations (Form 2) in Four Dimens- ions from KYST Analysis (Stress = 3.30) . Standardized Regression Weights for Four (KYST) Dimen— sions Used to Predict Mean Ratings on Unidimensional Attributes (Form 2; n = 67) . . . . . . . . . . Coordinates of 12 Situations (Form 3) in Four Dimens- ions from KYST Analysis (Stress = 1.48%) viii Page 73 107 116 127 128 130 139 141 142 155 164 169 170 176 Table 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. Standardized Regression Weights for Four (KYST) Dimen- sions Used to Predict Mean Ratings on Unidimensional Attributes (Form 3; n = 66) Coordinates of 12 Situations (Form 4) in Four Dimen- sions from KYST Analysis (Stress = 2.73%) Standardized Regression Weights for Four (KYST) Di- mensions Used to Predict Mean Ratings on Unidimension- al Attributes (Form 4; n = 66) . . . . . . . . . Situations Included in Clusters Generated from John— son's (1967) Hierarchical Clustering Program (Form 1 Data) . . . . . . . . . . . . . . . . . . Situations Included in Clusters Generated from John- son's (1967) Hierarchical Clustering Program (Form 2 Data) . . . . . . . . . . . . . . . . . Situations Included in Clusters Generated from John- son's (1967) Hierarchical Clustering Program (Form 3 Data) . . . . . . . . . . . . . . . . . . . . . Situations Included in Clusters Generated from John- son's (1967) Hierarchical Clustering Program (Form 4 Data) . . . . . . . . . . . . . . . . . . . . . INDSCAL Coordinates for Form 1 (Pour Dimensions) Standardized Regression Weights for Four (INDSCAL) Dimensions Used to Predict Mean Ratings on Unidimen- sional Attributes (Form 1; n = 64) . . . . . . . INDSCAL Coordinates for Form 2 (Pour Dimensions) Standardized Regression Weights for Four (INDSCAL) Dimensions Used to Predict Mean Ratings on Unidimen— sional Attributes (Form 2; n = 67) . . . . INDSCAL Coordinated for Form 3 (Four Dimensions) Standardized Regression Weights for Four (INDSCAL) Dimensions Used to Predict Mean Ratings on Unidimen- sional Attributes (Form 3; n = 66) . . INDSCAL Coordinates for Form 4 (Four Dimensions) Standardized Regression Weights for Four (INDSCAL) Dimensions Used to Predict Mean Ratings on Unidimen- sional Attributes (Form 4; n = 66) Page 182 186 187 197 203 210 211 213 214 217 218 221 222 Table 33. 34. 35. 36. Coordinates of 12 Situations for the Fourth Cluster of Respondents (Form 3) in Four Dimensional KYST Solution . . . . . . . . . . . . Standardized Regression Weights for Four (KYST) Di- mensions Used to Predict Median Ratings on Unidimen- sional Attributes (Form 3, Cluster 4) Coordinates of 12 Situations for the Sixth Cluster of Respondents (Form 3) in Four Dimensional KYST Solution . . . . . . . . . . . . . . . . . . . . . . Standardized Regression Weights for Four (KYST) Dimensions Used to Predict Median Ratings on Uni- dimensional Attributes (Form 3, Cluster 6) . Page 228 229 231 232 10. 11. 12. 13. 14. 15. 16. LIST OF FIGURES Tree Diagram of Results of Johnson's (1967) Hierarchi- cal Clustering of 40 Persuasion Situations . . Three-Dimensional Representation of 12 Situations for Split-Half Comparison of Form 100 Data . . . . . . Three-Dimensinoal Representation of 12 Situations for Split-Bald Comparison of Form 200 Data . . . . . . Three-Dimensional Representation of 12 Situations for Split-Half Comparison of Form 300 Data . Three-Dimensional Representation of 12 Situations for Split-Half Comparison of Form 400 Data . Three-Dimensional Representation of 12 Situations for Male-Female Comparison of Form 100 Data . . . . . . Three-Dimensional Representation of 12 Situations for Male-Female Comparison of Form 200 Data Three-Dimensional Representation of 12 Situations for Male-Female Comparison of Form 300 Data . . . Three-Dimensional Representation of 12 Situations for Male-Female Comparison of Form 400 Data . . . . . Plot of Stress Values for Dimensions One to Five for Each of the Questionnaire Forms (KYST Analysis) . . Plot of Percent of Variance Accounted for by (INDSCAL) Dimensional Solutions Two through Six . . . . . . . . Plot of Percent of Variance Accounted for by Each Dimension (GALILEO Analysis) . . . . . . . . . . . . Tree Diagram of Results of Johnson's (1967) Hierarchi- cal Clustering of Form 1 Data . . . . . . . . . . Tree Diagram of Results of Johnson's (1967) Hierarchi- cal Clustering of Form 1 Data . . . . . . . . . . . . Tree Diagram of Results of Johnson's (1967) Hierarchi- cal Clustering of Form 3 Data . . . . . . . . . . Tree Diagram of Results of Johnson's (1967) Hierarchi- cal Clustering of Form 4 Data . . . . xi Page 81 126 131 134 135 138 147 149 157 161 165 196 199 CHAPTER I INTRODUCTION Persuasion is pervasive in our social environments. In everyday experience, individuals attempt to persuade others who to vote for, which contraceptive is best, to stop smoking, to improve grades, which compu- ter analysis is appropriate, to purchase every item known to man, 22.227 finitum. Theories of attitude changes have been deveIOped based on notions of balance (Osgood and Tannebaum, 1955; Festinger, 1957), innoculation (McGuire, 1964), attitudinal anchors (Sheriff and Hovland, 1961), and many others. There exists a plethora of research on the nature of at- titude change and the effectiveness of various manipulations on obtained attitude change (i.e., credibility, physical attraction, latitude of re- jection, etc.) which attests to a thriving, active field of scientific enquiry. However, a problem stems from the context within which nearly all communication research on persuasion phenomena is conducted. Most stud- ies have assessed attitude change in the one-to—many context, often a message attributed to an anonymous source, sources varied in perceived credibility or physical attractiveness, or some variant of such a proce- dure. The problem is simply that persuasion phenomena arg_pervasive in our daily lives and past research lends no knowledge or insight whatso- ever into the nature of daily, dyadic persuasion situations, how indi— viduals perceive such situations and how they construct persuasion strat— egies and formulate messages based on perceptions of such situations. 1 2 In essence, the general problem is concerned with how appropriate or inappropriate it is to generalize the research results from the one- to-many context to the personal, dyadic persuasion situations. Without a doubt, hypotheses stemming from general theories of attitude change are applicable to the individual's everyday experience. However, in order to understand and predict behavior, it is essential to investigate the main features of how daily situations are perceived in order to as- sess the strategies daily employed, and their success and failure. The approach taken in this paper should not be considered radical in and of itself. As Miller (1976) noted in the foreward of a recent book on communication research: . . .the 'times' reinforced the shift in emphasis from public platform to private doalogue: encounter groups and sensitivity sessions stressed communica- tion purposes other than persuading and informing large heterogeneous audiences, and students them- selves began to demand answers about how to relate communication with their acquaintances, close friends and romantic partners. (pp. 9, 10) The emphasis placed in this paper on assessing the perceptions of common, relevant persuasion situations is representative of a current trend to- wards assessment of individuals interacting on a more private level. An analysis of how individuals categorize persuasion situations is of value because it provides an understanding of individuals' daily per- suasion existence and suggests limits on the generalizability of research results. However, the more theoretically important reason for assess- ing the perceptions of situations is concerned with the question of how such perceptions influence behavior. Of late, a considerable amount of interest has focused on the question of "consistency" of behavior across situations. While procedures for enhancing the predictive utility of trait measures may be devised, I argue that the search for consistency 3 in behavior across situations is a misguided quest. It is unfortunate that Mischel's (1968) claim that a trait meas- ure cannot predict behavior better than .33 is often interpreted as meaning that trait information is relatively unimportant and that "situ- ations," in some way, are determininats of behavior. Trait information is of value in many instances (i.e., reliably differentiating individu— als on a global orientation, assigning of individuals to treatments, testing of hypotheses in experiments [controlled situations], etc.). Further, "situations" are not determinants of behavior; how people as- sign meanings to situations may have such an influence. Given that people assign differential meanings to situations, it is not "consistency" or "inconsistency" across situations communication scholars ought to inves- tigate: people are at least adaptable in their environments and how they adapt to their perceptions of situations ought to be our central concern. The individual judged to be "inconsistent" does not behave irrationally; he behaves in accordance with different perceptions of the situation and different perceptions of the behavioral requirements of the situation. I submit that predictions of behvaior in naturalistic settings cannot be accomplished without a thorough examination of the individual's (a) behavior repertoire (ability to utilize any response pattern from a set of diverse response patterns); (b) the individual's perception of the situation and ability to discriminate among types of situations; (c) the ability of the individual to select, on the basis of such perceptions and discrimination abilities, a response pattern appropriate to the suc- cessful completion of a goal; and (d) the degree to which the individual values the expected outcomes. Finally, an individual's behavior at any point in time may be influenced by a self-regulatory strategy. For u instance, a domineering individual will be submissive in many instances in order to obtain a delayed goal or dominance at some future time. These five person variables have been offered by Mischel (1973) as a challenge to construct a new conceptualization of personality. I will show that this challenge should be given serious consideration. 1.0 Statement of Purpose Why is it important to assess the dimensional structure of the perceptions of persuasion situations? I will discuss three reasons: (1) the heuristic value of such knowledge; (2) pragmatic research utili- ty for delimiting generalizability of research results and implications for assessing, and generating a taxonomy of strategies individuals use; and (3) implications concerning the degree to which ”situations" influ- ence behavior. The first two reasons are discussed in Section 1.1. However, dis- cussing the implications concerning the degree to which "situations" in- fluence behavior requires considerable development. In Section 1.2, literature pertaining to the issue of trans-situational consistency of behavior will be reviewed, including research utilizing the "sources of variance" paradigm (see Endler and Magnusson, 1976a), in Section 1.2.2. The most severe problems in this research are interpreting interaction effects, problems in sampling situations and inflated interaction terms. Two alternative procedures for enhancing the prediction of behavior are also critiqued: the moderator variable approach and the behavioral ob- servation approach (Section 1.2.3.1 and 1.2.3.2). These approaches are shown to have very limited utility in improving predictions of modes of responses . Mischel's social learning theory reconceptualization of person- ality is presented (Section 1.2.3.3.) in order to highlight the various idiosyncratic reasons for why trait measures do not appear to predict behavior, even in situations which ought to enhance the emergence of in- dividual differences in responses. As noted in the summary of these various reviews (Section 1.3), one of these idiosyncratic reasons will serve as the focus of the present inquiry: the individual's perception of the behavioral requirements of the situations. The research investigating the perceptions of situations is re- viewed in Section 1.4. Since this literature is found to be quite impov- erished, an exploratory study is presented in Section 1.6 From this ex- ploratory study, some of the major parameters of the perceptions of per- ceptions of persuasion situations will be identified. Sampling problems identified in the exploratory study prohibit generalizing to any larger category of situations. Section 1.7 details the appropriate procedures for correctly arguing for generalizable dimensions. Chapter 2 of this paper provides a discussion of how the above procedures were implemented in the present study. Chapter 3 provides the results of various taxonomy constructing methodologies: factor an- alysis, multidimensional scaling analysis (GALILEO, KZSI, and INDSCAL) and cluster analysis (Johnson's [1967] hierarchical clustering). Chap- ter 4 summarizes the results and discusses various avenues for future research. 6 1.1 Heuristic Value and Pragmatic Research Utility Briefly, there exists an heuristic value in knowing the dimensions of perceived similarity of persuasion situations. The dimensional ar- rays of objects found in previous multidimensional scaling research have been found to correlate with independent judgments of communication be- haviors (Jones and Young, 1972) and with the emotional intensity expressed in facial expressions and implicit in trait adjectives (see Section 4). That is, construct validity of multidimensional scaling solutions has been frequently supported, and the obtained dimensions of situations would relate to the cognitions and the behavioral expectations the respondents perceive in the situations. This information is of value in and of it- self because it provides knowledge of the respondent's daily persuasion exsitence. Further, it will be argued later that the domain of situations comprising the domain of persuasion situations is a very large sub- category of the domain of "social behavior" situations (see Section 2). To the extent that this is true, there ought to be strong commonalities between the dimensional structures of situations sampled from either of the two general domains. Thus, an analysis of the dimensions of persua- sion situations not only informs us of the respondents' daily persuasion existence, but also suggests how one might attempt to assess the larger domain of social behavioral situations. Attempts at assessing such situ- ations are long overdue. for they provide answers to two legitimate questions concerning generalizabilty: (1) to what additional situations can these research results be generalized?, and (2) in what other situ- ations do other processes take precedence? For instance, in assessing the stochastic process of relational 7 communication in marital dyads, Parks, Farace, Rogers, Albrecht, and Abbott (1976) conclude that their findings: . .imply that most couples maintain relational defi— nition in on—going interaction by means of control- neutralization and submission or acceptance. While these are definitely control strategies, they are not reflect- ive of the blatant game of 'one—upmanship' often attri— buted to marital interaction. Prolonged struggles be- tween the participants over such issues as who is to con- trol interaction on a moment to moment basis were apparent- ly infrequent.. . . .It had been hypothesized that rela- tional communication patterns would vary as a function of how satisfied the participants were with their communica- tion, length of marriage and the level of role discrepancy in the relationship. In no case did the difference in structure between groups constructed according to the values of these variables reach statistical significance. As empha- sized above, a remarkable similarity in structure and pat- tern of relational communication was observed across all groups. (p. 17) Parks et al., analyzed relational communication patterns for the follow- ing four discussion topics: 1) how the couple met and decided to marry, 2) preparation of and sharing of a home, 3) career versus family interests (for the wife), and 4) plans for protection during a tornado. All discussion topics not only deal with family topics but cooperative tasks in a family setting (situation 3, however, may deal more with mu- tual compromise). It is to this class of situations that research re- sults can be generalized. It is of no surprise that one-upmanship should be so infrequent in this set of situations. Nor would it be surprising to find that the hypothesized relationships dealing with length of marriage, etc., were to be supported if the situations sampled were family, uncooperative (or even non—cooperative) tasks. In relation to research within the domain of persuasion situations, an assessment of the structure of situations will not only help address 8 the questions of generalizability but can also help facilitate research on two general questions concerned with persuasion strategies: (1) when will a particular strategy be employed? and (2) what are the strategies which individuals use in their daily persuasion experiences? In terms of generalizability, Kaminski, McDermitt, and Boster (1977) did the only study on compliance-gaining strategies that included an analysis of the sampled situations. Kaminski et al., conducted a pre- test on 40 persuasion situations and selected three situations entailing non-interpersonal relationships and three situations entailing interper- sonal relationships. Respondents completed the Machiavellianism measure (Christie and Geis, 1970b) and indicated likelihood of use and preference of use for 16 compliance-gaining strategies offered by Marwell and Schmitt (1967). Kaminski et al., supported a model of strategy selection proposed by Boster (1977): the sixteen compliance-gaining strategies can be or— dered on a continuum from "negative" to "positive” strategies. Further, "negative" strategies were found to be used less often than more "posi- tive" strategies. Finally, the authors contend that Machiavellianism and type of situation were not useful predictions of strategy selection. Indeed, the authors conclude by questioning the utility of the Machia- vellianism measure in differentiating individuals on the basis of strat- egy use and preference. Note that one is equally justified in concluding that the type of situation sampled had little utility in differentiating high and low Machiavellians on strategy use and preference. Nonetheless, the speci- fic conclusion of the Kaminski et al., study is as follows: no differ- ence exists in strategy use and preference of use between high and low 9 Machiavellians for "interpersonal relationship" and "non-interpersonal relationship" situations. To what population of situations, persons, and modes of responses can this conclusion be generalized? While it is likely that "interpersonal relationships vs. non-interpersonal relation- ships" is a very important attribute, "situations" also vary in many other ways: face-to-face, unstructured, who benefits from the persua- sion, etc. Unless a complete analysis of these types of situations is conducted, this question of generalizability to situations will remain open. To what modes of response can the result be generalized? High Machiavellians are "manipulative"; they lie better than low Machs (Ex- line, Thibaut, Hickey, and Gumpert, 1970); devise innovative manipula- tions (Geis, Christie, and Nelson, 1970); and do better than low Machs in competitive bargaining (Geis, Weinheimer, and Berger, 1970). Do the sixteen compliance-gaining strategies offered by Marwell and Schmitt (1967) include "manipulative" response modes? This question immediate- ly implies the more general question: are the compliance-gaining strat- egies proposed by Marwell and Schmitt an exhaustive list of strategies (let alone relevant to a class of individuals and situations)? By puta- tively sampling situations, or pre-calibrating situations along a single attribure (or a few attributes), one is not assured that the more pro- nounced, salient features which (from the respondent's perspective) dif— ferentiate sets of situations from each other will be uncovered. Once a complete analysis of situations is conducted, then one can return to the question of relevant modes of response in an analysis which requires in- dividuals to report their strategies along precalibrated situations. Aside from the strategies suggested by Marwell and Schmitt (1967), people 10 pout, cry, control the persuadee's information environment, employ "foot- in-the door" tactics, etc. Constructing such a taxonomy of relevant strategies is of prime importance for communication. This is not to say that the strategies offered by Marwell and Schmitt have no utility. Wood, Weinstein, and Parker (1967) have found that the use of threat ("negative sanctions") is rare among children (kindergarten through third grade) when dealing with mothers, teachers, and peers. However, Wood et al., as well as Kaminski et al., analyzed the first attempt persuasion strategy. Still employing the strategies, it would be of interest to construct models for how certain strategies come into preferred usage based on various person and situational influ- ences. For example, when will threat be used? Tentatively, I hypothe— size that usage of threat is a function of (a) one's level of latent hostility; (b) the extent to which one has initially used more positive strategies and failed; (c) the extent to which one perceives one's rights have been infringed upon; (d) the extent to which one perceives the goal to be achieved as important; and (e) the extent to which one has "means control"--ability to exercise some threat. Other hypotheses can be con- structed. In sum, there are several distinct reasons for assessing the dimen- sional structure of the perceptions of persuasion situations: (1) there is an heuristic value in knowing the dimensions of the respondent's daily persuasion existence; (2) the analysis suggests how other domains may be similarly assessed, and some of the attributes which may charac- terize other domains of situations; (3) the analysis clarifies problems of generalizability (to situations); and (4) it facilitates construction of a taxonomy of relevant strategies. 11 Aside from these issues, the more theoretically interesting reason for the assessment of situations deals with the question of cross- situational consistency of behavior. This question has recently been the focus of considerable attention by psychologists (Mischel, 1968) and more recently in communication (Hewes, Haight, and Szalay, 1976). A number of different perspectives have focused on this question and I shall review these perspectives in Section 1.2. 1.2 Trans-situational consistency across what types of situations, for what types of modes or responses, for what types of persons? 1.2.1 Overview Hewes, Haight, and Szalay (1976) have recently questioned the pre- dictive utility of several personality constructs. Specifically, they assessed the ability of measures of reticence, predisposition toward verbal behavior, anxiety, self-concept, and extraversion to predict each of six communicative behaviors: 1) duration of conversation within a small group, 2) length of an extemporaneous Speech, 3) number of persons contacted in unstructured class time, 4) choices of written versus oral presentation of an assignment, 5) duration of response to a free association task, and 6) latency of response in a free association task. The authors found little predictive utility for four of the five meas- ures and the fifth measure (predisposition toward verbal behavior) had a slight predictive ability on measures of duration only. The authors were careful to note, however, that the results of their study ought not to imply a rejection of personality variables in 12 predicting communication behaviors. Instead, Hewes et al., present a number of alternative implications concerning future theorizing and meth— odology. The Hewes et al., paper will undoubtedly catapult the general ques- tions of cross-situational consistency of behavior from the discipline of psychology into the communication field. This paper reviews this current debate in order to obtain information relevant to future communi- cation theory and research. It should be noted at the onset of this paper that the question of cross-situational consistency of behavior (or the predictive utility of trait measures) is not a new issue. According to Bem and Allen (1974), the "controversy" existed in the 1930's (Hartshorne and May, 1928), when Allport (1937) and Stagner (1937) defended modified trait conceptions of personality against a number of studies questioning the consistency of behaviors across situations. The debate lasted for nearly a decade be- fore receding into the background just prior to World War II. Mischel (1968) has revived the debate by arguing that the commonly observed +.33 ceiling on cross-situational correlations probably reflects true behavioral variability rather than imperfect methodology. The de- bate is an active one, and the implications of the debate and the research generated by the debate to the field of communication are plentiful. Since "situations" form the background of all behavior, this debate should not be taken too lightly by communication theorists (c.f., Hewes et al.). The goal of this section of the paper is to provide clarification concerning the various arguments provided by advocates of the trait, in- teractionist and Situationist positions. Since providing this 13 clarification is no easy task, the goal of this overview is to provide a structure for the organization of the clarifications. In Section 1.2.2, the research utilizing the "sources of variance" paradigm which provide the persons X situations interactions will be criticized on the grounds that: (a) they provide no substantive inter- pretation of the interaction effects; (b) interactions may be spuriously inflated; (c) interactions have not been demonstrated to have predictive utility; (d) insufficient attention has been paid to the problems of sampling situations; and (e) failure to specify why the particular mode of response or behavior is worthy of study. An application of generali- zability theory (Cronbach, Gleser, Nanda, and Rajaratnam, 1972) would effectively avoid the problems encountered in (b) but does not address other problems. As Golding (1976) noted, only under the circumstances when indices of both persons and situations have been obtained can the interaction effects be interpreted in any substantive fashion. Another type of analysis (reaction mode analysis) provides an assessment of sit- uations (factors of situations), and these studies will be reviewed in Section 1.4. Aside from studies conducted in the "sources of variance" paradigm, several alternative procedures have been offered for assessing, and en- hancing, predictions of behavioral performance. First, the "moderator variable" approach has been offered as a means to differentiate indi- viduals on some criteria and to obtain differential predictions for the several subgroups. Unfortunately, while the differentiation of indivi- duals on a criterion variable is done a priori, research utilizing the moderator variable approach will likely lead to variable success until such time as the selection of situations which yield the differential 11+ predictions can also be done a priori. That is, the approach will be of value only if it ultimately specifies why the "high-variable" individual is very "friendly" in some situations and less "friendly" in others; and similarly for the "low-variable" individual. The "behavior observation" approach also suffers from the same lim- itation: prediction of behavior from one observation to another will be of little value until an analysis of how the individuals perceive the behavioral requirements of situations, and act accordingly, is conduct- ed. That is, the success of the behavior observation method rests sole- ly on the respondent's implicit notion of functional equivalence of sit- uations. Both the "moderator variable" approach and the "behavior obser- vation" approach will be discussed in Section 1.2.3.1. 1.2.2. The "Sources of Variance” Paradigm Recent emphasis has been given to the question of an individual's behavioral consistency across situations (Mischel, 1968, 1973; Alker, 1973; Bem, 1972; Bem and Allen, 1974; Hewes, Haight, and Szalay, 1976). There are three positions which must be considered when discussing this question: (1) the traditional trait psychology position; (2) the situa- tionist position (see Bowers, 1973), and the interactionist position (Argyle and Little, 1972; Endler, 1973; Ekehammar, 1974). As noted by Endler and Magnusson (1976b), the trait model assumes that there are various continuous dimensions with individual differences on each of these dimensions. The trait theorists emphasize traits as the prime determinants of behavior, where the factors determining beha- vior are withjn the person himself (i.e., B = f(P)). From the perspec- tive of the trait model, the major interest lies in the relation between responses in different situations and between responses and the latent 15 dispositions for which the responses are supposed to be indicators. For the trait model, traits are the main sources of behavioral differences and the rank order of individuals with respect to a certain behavior is assumed to be consistent across different situations. Situationism is usually considered to be the antithesis of the trait model. The general Situationist model regards factors (or compo- nents of the situation) as the main determinant of individual behavior. This model is essentially a stimulus-response (S-R) model, and its advo- cates are typically sociologists (Goffman, 1975) who argue that individ- uals assume a role on the basis of implicit knowledge of its assignment-— based on their perceptions of the situation. In general, assessment of the Situationist model is done by manipulating or selecting situations that vary in some aspect and assessing the main effects on some dependent variable. Recently, however, Bowers (1973) noted that the interaction effects obtained by advocates of the interactionist position are "grist for the Situationist mill" (p. 151). Apparently, some individuals would tend to be content with interactions between persons and situations, as opposed to holding a hard—line Situationist perspective that situations are the main source of variance. According to the interactionist model, . . .actual behavior is the result of an indispensable, continuous interaction between the person and the situa- tions he encounters (i.e., B = f(P,S)). This implies that the individual's behavior is influenced by signifi— cant features of the situations, but furthermore the in- dividual chooses the situations in which he performs and selects significant situational aspects in these situa- tions. He subsequently affects the character of these situations. (Endler and Magnusson, 1976b, p. 3) Of late, the interactionist position has received overwhelming support, but this support should be viewed skeptically. Based on research using 16 \ the "sources of variance" paradigm introduced by Endler (Endler, 1973; Endler and Hunt, 1966, 1968, 1969), a number of studies have found that a substantial amount of variance is accounted for by the interaction of persons and situations, more than by either persons or situations inde- pendently. Unfortunately, the interpretation and implications of the obtained interaction effects are not clear. Golding (1975) noted that: .one must bear in mind that the currently avail- able data are far too fragmentary, method specific, or confounded to allow for strong theoretical inferences. Most studies rely on self-report behavior in imagined situations (e.g., Endler and Hunt, 1966) or, more rarely, on behavioral observation of questionable generality or importance (e.g., frequency of head nods, leg movement, scratching, or smoking in Moos, 1969). This data base hardly qualifies as representative of the behavioral do- main in which we would like to fashion our theories of human behavior. (P. 283) Golding further notes that: Progress in the development of theories concerning the joint influence of individual differences and situations has been hampered by a lack of attention to methodologic- al and logical rigor. Much of the debate has been waged without adequate specification of what is meant either by individual differences or by situations. For example, it is clear that all conceivable individual differences are not of equal importance in the theoretical construction of concepts of personality (Goldberg, 1971). Addition- ally, with few exceptions (Argyle and Little, 1972; Gold- berg, 1972a) little attention has been given to various models of individual difference-situation interactions that could be subjected to competitive hypothesis test- ing. Moreover, as long as situations are arbitrarily defined without an attempt to assess their psychologi- cal differences and similarities (Magnusson, 1971; Moos, 1973), little progress in the development of comprehen- sive, noncircular theories about human behavior is pos- sible. As Bowers (1973) has pointed out, it is current— ly common practice to reason, quite circularly, that if individuals behave differently, then the situations were "different," and if they behave similarly, then the sit- uations were "not different." One can logically draw conclusions about the singular and joint influence of individual differences and situations only to the extent 17 that each is explicitly and noncircularly defined and independently measurable by techniques that approach currently acceptable levels and definitions of relia— bility and validity. (p. 286) Finally, Mischel (1973), Bowers (1973), and Golding (1975) have noted the extremely problematic questions dealing with the sampling of individuals and with the sampling of situations. For one, without know- ledge of how persons or situations are sampled, it becomes impossible to generalize the results of any particular study to any domain of individ- uals or domain of situations. As noted above, without such knowledge, interpretation of such interaction effects are hazardous, study specific, and are not generalizable. Further, without such knowledge, it becomes impossible to place any value on the obtained interaction because the experimenter may have (unknowingly) sampled an extremely heterogeneous set of individuals and/or homeogeneous set of situations. If this is the case, then the results ought to support the trait position. Further, the experimenter may have (unknowingly) sampled an extremely homogeneous set of individuals and a heterogeneous set of situations, in which case some support for the Situationist position would be obtained. In essence, studies in the "sources of variance" paradigm can be designed to demonstrate almost any outcome. The problem of sampling "putatively different experimental situations" becomes a gravely distur- bing problem when one notes that any situation may be perceived very differently by the respondents, contrasted to the perspective of the experimenter who sampled the situations. Thus, it is of critical impor- tance to assess the respondents' perceptions of the situations. A brief review of several of the studies conducted in the "sources of variance" paradigm would quickly inform the reader of the essential reasons for Golding's criticisms. Moos (1969--replicating Moos' 1968 18 study) had sixteen psychiatric patients describe their reactions to six psychiatric ward settings. Their behavior was also observed twice in each of six settings that were representative of regular ward activities. Pa— tients were selected as respondents on the basis that: (a) their conversation had to be easily understandable, (b) they had to be able to cooperate in wearing a wire- less radio transmitter and also be willing to answer a questionnaire about their reactions in each setting, (c) they had to be on the ward long enough for system- atic behavioral change to occur. (p. 406) Items selected for use in the questionnaire were taken partially from Nowlis' (1965) Mood Adjective Check List (anxiety, depression, vigor, and pleasantness); from Moos (1968--perceived self worth and perceived therapeutic benefits); and from Barker and Cump (1964--affiliation, part- icipation, leadership). Each of the above nine scales was labeled a "dimension." Two observers rated behaviors across the settings: hand and arm movement, foot and leg movement, scratch, pick, rub, general movement, nodding yes, smiling, talking, and smoking. Inter-rater reliability ranged from .77 to .99. Results indicated that for each item on the questionnaire and for most behaviors observed, the Person X Setting interaction accounted for a larger portion of the variance than the amount of variance accounted for by either persons or situations. Smoking was found to be due to persons (some people consistently smoked more than others), and most of the variance of talking and nodding "yes" was due to setting. Trickett and Moos (1970) selected twelve upper middle class high school students who were ". . .less representative of high school sopho- mores in general than those with well above average intelligence coming 19 from upper middle class families" (p. 400). For "situations" Trickett and Moos selected the four classes the twelve students had in common: mathematics, English, biology, and government. The individuals rated each class on thirteen feeling and mood adjective pairs. Five factors emerged from a factor analysis of the ratings. The results of the an- alysis of variance indicated that for each of the five factors, labeled "paying attention," "participation,' "anxious uncertainty," "satisfac- tion," and "seating," the Person X Setting interaction accounted for a larger portion of the variance than the amount of variance accounted for by either persons or situations. Note, however, that it is impossible to generalize these results to any set of individuals and across only a very small number of situa~ tions. Further, note how Trickett and Moos described individual differ- ences: Subject #7 obtained a higher mean satisfaction score than Subject #10 in all four classes. Also, Subject #7 differentiated very little between classes whereas Subject #10 viewed Government as far more satisfying than his other three classes. (p. 403) There are two observations to be made in relation to the Moos (1968, 1969) and Trickett and Moos (1970) studies. First, without knowledge of the perceptions of situations and without knowledge of individual differ- ences, uncovering the nature of the process becomes circular descriptions: respondents who report high satisfaction across all four classes do not "differentiate" between the classes. Further, no explanation is given for why subject #10 is more satisfied in government. Second, for a Person X Situation interaction to be informative and of value, the exact nature of the interaction must be articulated, and specific interaction effects are often very elusive (see Goldberg, 1972b). 20 The Trickett and Moos study ought to be contrasted to the types of ap- proaches parsimoniously utilized by advocates of the trait position; namely, tests of the hypotheses such as "individuals who are high in need to achieve will express more satisfaction across courses than individuals low in need to achieve" (thus arguing that levels of satisfaction rest primarily in the person), or a test of the hypothesis, "individuals high in authoritarianism will express more satisfaction in courses where instructors are authoritarian" (thus arguing that levels of satisfaction rest partly in the person and partly in some element of the situations). Further, what is it about the situations which influence satisfaction? Instructional format, instructor credibility, quizzes, lectures, diffi- culty, etc.? Trickett and Moos leave all of the important questions un- answered. The main body of Person X Situation interaction studies have been conducted by Endler and his colleagues (Endler and Hunt, 1966, 1968, 1969; Endler, 1973). In these studies, the respondents are presented with a variety of potentially anxiety-arousing situations or hostility-arousing situations, and are required to indicate the perceived likelihood of their response in certain ways during each situation. Endler and Hunt (1969) will be used as an illustrative study. Endler and Hunt provided six forms of the S-R Inventory of Anxiousness (in which from eleven to eighteen situations were provided with ten modes of responses indicative of anxiousness) for the respondents' use in rat- ing their behavior in each situation. The results strongly supported the interaction effects. Further, Endler and Hunt noted that the vari- ance due to situations was larger for females (7.78 percent) than for males (3.95 percent), which is not a very large difference. Differences due to age were also noted, but were also small. 21 A number of other studies utilizing the "sources of variance" para— digm attest to the attractiveness of the paradigm. Such studies include work in anxiety, leisure time behavior (Bishop and Witt, 1970), prefer— ence for drinking various beverages (Sandell, 1968), and moral behavior (Nelson, Grinder, and Mutterer, 1969). In all of these studies, no in— terpretation of the interaction effects was provided. In all fairness to the trait position, it should be noted that the classic psychometric and/or trait theorists position does not ignore sit- uational influences on behavior. Goldberg (1972a) notes that: . .the classic psychometric position has been that situations "constrain" individual difference--that they profoundly affect both the mean and the variance of these differences, though the rank order of individuals on the "trait" should remain relatively invariant across those situations which permit sizable trait variations to occur. (p. 550) The interactionists contend that the rank order of persons is not invari- ant. A considerable amount of attention has been given, on the part of the interactionists, to the fact that the interactions account for so much of the behavioral variance that they (interactions) are somehow "better" than a main effect for persons (trait measures). It seems legit- imate, then, to ask a very simple question: does utilization of inter- actions enhance the prediction of behavior across situations? Goldberg (1972b), in a thorough study of interaction hypotheses in college instruction (utilizing criteria such as course satisfaction, course achievement, and non—grading [extra-curricular] readings) tested the predictive utility of the best general predictor of each criterion (separately for males and females) and the most significant interaction effects (separately for males and females). Goldberg found that in no case (male-female for each criterion) did the significant interaction 22 effects result in predictions that were any more valid than simply using a single general predictor. About twice as much of the criterion vari- ance was predictable by the "best" of the general predictors than by the "best" of the differential predictors. Golding concluded that: These poignant findings, when coupled with those con- cerning the fate of attempts to construct new empiri- ca; interaction scales, suggest that the significant interactions discovered in this project-~even if rep- licated at the very same strength in future studies-- are unlikely to lead to differential predictions whivh are more valid than those achievable by general predictors alone. (p. 200) Admittedly, I have over—exxagerated the claims of the interaction— ists; they do not explicitly argue that interaction effects can better predict behavior-—only that persons (traits) and situations do not. I have included the comments of Goldberg only to point out that interaction effects, or differential predictors, have not been shown to have predic- tive utility. Mischel (1973) similarly criticized the interactionist research on this count: the interaction studies conducted by Endler and Hunt and by Moos have left the ultimate question of prediction unanswered. Finally, Cartwright (1975) questioned how an inventory (the S-R inventory in particular--Endler, Hunt, and Rosenstein, 1962) can have re- liability coefficients of .97 and .95 and yet account for only 5 percent of the variance due to persons. He argued that the inventory as an "inventory" be differentiated for the inventory as an "experiment"; and that as an "experiment" there exist a number of undesirable features, not the least of which is the question that modes of response do not form a single factor but two: anxiousness of anxiety and anxiousness of pleasurable anticipation. Both factors are utilized as a single "condi— tion" in the analysis of variance, and Cartwright correctly argued that: . . .this factor should not be allowed to contribute 23 variance to the total against which percentages of variance from other sources are evaluated. It is doubtful whether it should be entered as a factor at all. (p. 410) A similar argument applied to "situations." Cartwright's results confirmed his expectations that the inclusion of modes of response (and situations) as single conceptual categories is improper: We are now in a position to explain why the variance due to traits (individual differences) is so small when the original S-R Inventory is analyzed as an ex- periment (5%), whereas it is so large when the inven- tory is analyzed as a test (95%). The answer seems to be that as an experimental design, the inventory allows large quantities of variance from unjustified sources to enter into the total variance against which the individual differences component is eval- uated. By summing across modes of response, the pro- portion of variance due to individual differences becomes relatively larger (33% to 59%), since varia- tion due to the modes factor is eliminated from the total. By summing across situations (as in computing a total test score on the inventory) variation due to situation is also eliminated from the total. As a test, the Inventory's results do not appear to "doom omnibus inventories" at all. (p. 414). The results of Cartwright's study are applicable to all of the "sources of variance" studies, though it is not certain to what degree these ef- fects operate in other sets of data. A distinction must be made between the set of studies which have employed the "sources of variance" paradigm and the set of studies that utilized a reaction mode analysis (see Section 1.4). The criticisms leveled at the "sources of variance" paradigm do not necessarily apply to the reaction mode abalysis studies. Concerning the former, I should state my argument succintly: the analysis using the "sources of variance" paradigm which provides the Persons X Situations interactions has little value. The primary reason for concluding this is the fact that no 24 substantive interpretation of the interaction is or can be provided. Other considerations taken into account in making this statement in- clude: (a) the fact that interpretations may be spuriously inflated (Golding, 1975, p. 285); (b) interactions have not been demonstrated to have predictive utility; (c) insufficient attention given to the prob- lems of sampling situations; and (d) failure to specify why the behav- ior or mode of response is worthy of study (Moos, 1968, 1969). Finally, Cartwright's observation must also be dealt with. The reacion mode analysis entails factoring 13mg response ratings over persons in order to provide an assessment of the factors of situa- tions. Sometimes, when multiple response categories are employed, fac- tors of situations and factors of responses can be obtained. When cou- pled with analysis that enables homogeneous subclusters of individuals to be identified (on the basis of similarities of responses), substantive interpretaion of interactions can be made. For example, Ekehammar, Magnusson, and Ricklander (1974) obtained three factors of situations, "threat of punishment," "anticipation fear," and "inanimate fear," and two interpretable factors of anxiety responses, "psychic anxiety" and "somatic anxiety." (For additional details of this study, see Section 2.) These factor analyses were done by (a) factoring ratings of anxiety responses for seventeen situations for eighteen modes of responses and by (b) factoring eighteen modes of responses for the seventeen situations. Magnusson and Ekehammer (1975) reported analyses of the individuals who served as respondents in the Ekehammar et al., study. In order to isolate homogeneous subgroups of individuals, a method of latent profile analysis was employed. The results of the latent profile analysis 25 indicated that there were three homogeneous subclusters for both males and females. Females were found to be more anxious, for both types of anxiety and across all three types of situations. Females elicited sub- stantially higher levels of psychic anxiety than males, particularly in "threat of punishment" situations and "inanimate fear" situations, while males reported far less anxiety in "inanimate fear" situations than in "threat of punishment" situations. While the profiles of anxiety responses for males and females were non-parallel (there were strong interactions), Magnusson and Ekehammar found that 57 percent of the females were trans-situationally consistent in their reported responses, and an additional 17 percent were compara- tively high in consistency across situation type. For males, 53 percent were judged as being trans-situationally consistent, and an additional 36 percent were judged as comparatively high in consistency. Unfortunate— ly, no independent assessment of persons (trait measures) were utilized and these conclusions are post hoc. I have discussed these two studies at this point in time only to point out that an analysis different than the "sources of variance" an- alysis, but done with the same type of input data, can yield valuable information: many of the individuals (both male and female) were fairly consistent in their responses--but the ordering of mean anxiety responses was different between males and females. Further, the pattern of anxiety responses for "inconsistent" individuals (of both sexes) indicated that the individuals elicited much higher levels of both types of anxiety for "threat of punishment" situations, somewhat higher anxiety responses for "inanimate fear" situations, and as low (or lower than) "anticipation fear" situations than the individuals who were judged as consistent. 26 In order to analyze the impact of situations, it is a basic con- cern to obtain an analysis of the situations themselves. This is true even when the analysis employs the generalizability theory offered by Cronbach, Gleser, Nanda, and Rajaratnam (1972). Utilization of the gen- eralizability coefficients helps to avoid spuriously inflated interaction terms. However, Golding nonetheless recommended forming indices of sit- uations (or providing some assessment of how situations vary) in order to provide the basic information for interpreting interaction effects. 1.2.3 The "Moderator Variable" Approach, the "Behavioral Observa- tion" Approach, and the Idiosyncratic Behavioral Patterns of Individuals 1.2.3.1 Overview. It should be noted that the interaction- ists claim that trait psychologists have never addressed the question of situational influences is incorrect. Christie and Geis (1970b) go to great length to discuss the interaction of Machiavellianism and the en- vironment. Indeed, one can measure a variable such as "expresses con- trol" (or "assertiveness"), vary level of Machiavellianism and sample situations that: (a) are ambiguous or highly structures, and (b) allow for improvisation or do not allow for improvisation. In a factorial design, a strong interaction effect will (theoretically) be obtained. Analysis using the "sources of variance" paradigm will provide a large Persons X Situations effect that will not be interpretable. Correlations between "social desirability" modes of responses and "need for approval" are also moderated by a situational variable: "in- terpersonal evaluation." Individuals who are rated as high on "need for approval" will elicit more "socially desirable" responses than low "need for approval" individuals only if they perceive the situation as 27 entailing an evaluation. Other examples can be given, but they are in— cidental to the purpose of this section of the paper. No one disputes the claim that behavior is a function of both the individual and the environment. Further, no one disputes the claim that trait measures have the general utility of reliably differentiating in- dividuals in a relevant, yet global, fashion. Further, no one disputes the fact that trait measures have specific utility in testing hypotheses concnering communication phenomena (i.e., dogmatism in persuasion [see Miller and Rokeach, 1968]; and reticence in identifying individuals who are high on speech anxiety [Phillips, 1968]; to cite only two examples). What is in dispute is the answer to the question: why is it not possible to predict from a trait measure to a specific behavior elicited in a specific instance? It must be noted at the onset of this section of the paper that the theorist must specify a relationship between four classes of vari- ables: (a) why a mode of response is relevant to a person measure; (b) why a set of situations will differentially elicit such a mode of re- sponse; (c) why the respondents ought to agree with the experimenter's differentiating the situations into classifications of high and low, in terms of eliciting the response; and (d) why the respondents ought to agree among themselves that the experimenter is essentially correct in classifying the situations. To the reader unfamiliar to the literature on consistency of behavior across situations, (c) and (d) may sound a bit unfamiliar. However, they are the crux of the whole matter, at least in terms of predicting behavior in a specific instance. Most research in this area begins with the logic as follows: (1) Theorist T claims that Measurement M differentiates (at a global 28 level) people on the basis of behavior B; (2) Situations S S S S 1’ 2’ 3’ 4’ 85’ and S are perceived such that they ought to elicit behavior 8' 6 (Closely related to or the same thing as B); (3) a study is done. Alas, at no point in time is r greater than .33. Therefore, either (a) the BB' study was poorly done (meaning almost anything); (b) M was a poor measure, go back to the blackboard; (c) M and the whole world of personality, plus parts of social psychology and communication are doomed to ultimate, un- equivocal failure; or (d) people's response patterns are too idiosyncratic to permit such prediction. It seems to me, and perhaps to many others as well, that (2) ought to be considerd as one of the culprits that usurps the predictive utility of our measures. Further, Theorist T ought not to be blamed, for the "ought to elicit" can be as tied, related, and relevant to the "M measure- ment of B" rationale as possible, yet the respondents may see something totally different in (any of, but not generally all of) the situations. Thus, (d) is the correct answer. If it is possible to eliminate any future reference to the research conducted in the "sources of variance" paradigm, then few alternatives remain. In this section of the paper I shall argue that there are only two alternative procedures that enable (or, at least, enhance) the pre- diction of behavior in a specific instance: define the situation for all respondents (thus ensuring the experimenter's conception of operative behavioral expectancies of the situation will be made known to the res- ‘pondents) or assess how the respondents themselves perceive the behavior- al expectancies of the situation. However, before discussing these two alternatives, I shall explain why two other alternatives will lead to more adequate levels of prediction, but are less than totally desirable: 29 the "behavior observation method" and the "moderator variable approach." 1.2.3.2 The "moderator variable" approach and the ”behavior observation" approach. When the consistency question thrived in the 1930's, Allport (1937) contrasted the nomothetic view of personality with an idiographic approach. The nomothetic view upholds an assumption that a trait (or set of traits) are universally applicable to all persons and that individual differences can be described in terms of different loca- tions on these traits. For example, Hartshorne and May (1928) assumed that an "honesty-dishonesty" dimension could be used to describe all of the sample of children (i.e., people vary only_in degree of "honesty- dishonesty"). Allport (1937) simply noted that a person's performance in any single instance may be overridden by other concerns. That is, a person may hold that being honest is simply less important than being loyal, etc. A child may lie to a teacher in order to maintain loyalty to friends or perhaps to avoid hurting someone's feelings. Similarly, a moderate- ly high-anxiety individual's anxiety about delivering a speech may be overridden by the need to get good grades, a particularly high interest, etc. Allport's idiographic view emphasized that individuals vary not only in locations on traits, but also in terms of which traits are rele- vant. Bem (1972) employed the term "equivalence classes" to denote the case where a sample of individuals share the same behavioral response pat- terns for the same types of situations. For instance, for some individ- uals "socially desirable responses" form an equivalence class of beha- viors, and "interpersonal evaluation situations" form an equivalence class 30 of behaviors, and "interpersonal evaluation situations" form an equiva- lence class of situations. Other individuals do not share in these equivalence classes of responses and situations. Bem and Allen (1974) extended Bem's arguments and identified a key problem of uncovering (constructing?) trans-situational consistency of behavior from the nomothetic conception: . . .the traditional trait-based research study will yield evidence of cross—situational consistency only if the individuals in the research sample agree with the investigator's a priori claim that the sampled behaviors and situations belong in a common equiva— lence class 32d only if the individuals agree among themselves on how to scale those behaviors and situa- tions. (p. 510-—emphasis theirs) Thus, in any sample of individuals from a population, only a subset of individuals will share 4———-—> ————) 6-—-—) 7——> e-———> 9————> 1o——> Interpersonal - Commitment b = Interpersonal - No Commitment Non-interpersonal - Commitment Non-interpersonal ~- No Commitment c d: * Situations are listed in Table 1. a “ “AA.— 82 also appear in Table 1) has a dotted line that goes downward. Situa- tions that have merged are depicted by the shaded bars. Thus, during the clustering process, situations 35 and 36 merged together first, then situations 5 and 6, etc. (see Figure 1). All situations eventually merge together. For the sake of clarity, I have placed situations into 10 numbered clusters. Cluster one contains a combination of situations from three of the four Miller and Steinberg categories. The first two situations entail retrieving one's property from a close friend and from a next door neighbor. Situations three and four entail obtaining money that is due the persuader. In general, these four situations entail a right of the persuader to request or retrieve money or tangible goods, regardless of the type of relationship involved. Situations 5, 6, 7, and 8 are all non-interpersonal—no commitment to future interaction situations. Each of these situations entails an infringement of the persuader's rights by companies or "professionals" (insurance companies, record companies, newspapers, landlord). Clusters 1 and 2, then, comprise clusters of situations which involve the infringe- ment of the persuader's rights in general; and a distinction is apparently made between money, etc., due them and the general inconveniences some- times involved in dealing with professionals. In all of these situa- tions, the persuader has a right to make the request. Cluster 3 consists of situations 9, 10, 11, and 12. While situa- tions 11 and 12 are non-interpersonal relationship-commitment to future interaction, these four situations appear as if they commonly deal with a certain type of relationship in which the other person has let them down (professor didn't give a raise, gives poor grade, dog barks, and 83 car-pooler is always late). Generally, these situations entail a fail- ure on the other individual's part to uphold what the persuader probab- ly perceives as an implicit agreement. Cluster 4 consists of situations 13, 14, 15, and 16. All of these situations deal with persuading someone to stop an action which irritates the persuader, as well as failing to uphold an implicit rule (failure to work on a group project, be quiet in the library, stop smoking). In general, situations in cluster 3 and 4 deal with persuading others to acknowledge their responsibility and to stop an action which is irritat- ing. A Note that at a global level all situations in clusters 1 through 4 deal with the individual's rights. However, situations in clusters 1 and 2 deal with exercising specific rights, while situations in clusters 3 and 4 deal with the other individual's responsibility to the persuader. Situations in clusters 5 and 6 will be discussed together. Aside from situation 17, all of the situations in these clusters are non- interpersonal-no commitment to future interaction situations. The dis- tinguishing feature of situations in cluster 5 is the fact that they en- tail bargaining; the persuader attempts to bargain down the price of a rocking chair, get out of a speeding ticket, reinstate the phone without an additional deposit, and to obtain permission to return to the hotel for a wallet. Cluster 6 situations deal with requesting special favors (a ride, buy beer for the persuader, and permission to see the doctor next). Cluster 7 consists of situations 25, 26, 27, and 28, all of which are interpersonal relationship—commitment to future interaction (except situation 28, a non-interpersonal—commitment to future interaction 84 situation). These situations involve persuading a friend to do something especially for the persuader's benefit (move off campus, permit the per- suader to visit with an old friend instead of going to the movies, go to the movies instead of dancing, spend the evening alone with the per— suader). Cluster 8 situations are all interpersonal relationship-no commit- ment to future interaction situations (except situation 29, interpersonal- commitment situation). These situations (29 through 32) perhaps deal with more important matters than other situations. Aside from this, these situations appear as if they deal with more intense interactions and possibly overcome resistance (take a job in the immediate vicinity, break off a long-term relationship, persuade a job competitor to take a less attractive offer, break a deadlocked jury). Situations in both clusters 7 and 8 deal with personal benefits, but situations in cluster 8 may be more important, more intense or occur with less frequency than those in cluster 7. All of the situations in cluster 9 consist of interpersonal rela- tionship-no commitment to future interaction situations (expcept situ- ation 33, an interpersonal-commitment situation). These situations (33 through 36) deal with persuading a person to do something for the persuad— er's benefit and lead to further interactions (possible becoming more ac- quainted with each other). These situations include persuading a person to have pre-marital sex, going out to breakfast in the morning, and hav- ing dinner at the end of the flight. Cluster 10 consists of situations which are all interpersonal rela- tionships-commitment to future interaction situations (except situation 40, a non-interpersonal-no commitment to future interaction situation). 85 These four situations (37 through 40) are situations entailing helping others (persuade a friend to seek professional help, a son to get bet- ter grades, borrow the family car to help out a friend, and purchase a set of encyclopedias for a person's children). In general, the results of the cluster analysis indicate the fol- lowing breakdown of the four-part typology: 1) Interpersonal-commitment to future interaction a) personal benefits b) help others 2) Interpersonal-no commitment to future interaction a) personal benefits--casual b) personal benefits--intense, important 3) Non-interpersonal-commitment to future interaction a) infringements of rights b) irritation--failure to uphold responsibility 4) Non-interpersonal-no commitment to future interaction a) baraginaing (compromise) b) requesting special favors) 1.6.3.3 Multidimensional Scaling Results6 The percent of variance accounted for by each dimension, for the first six dimensions, are as follows: 23.29, 13.27, 10.02, 6.39, 5.98, 5.14. It would appear that a change in the slope of the scree line occurs between the fourth and fifth dimensions. A four-dimensional solution ac- counts for 52.97 percent of the variance. The first dimension can be interpreted as an "interpersonal rela- tionship vs. non-interpersonal relationship" dimension with situations 28, 33, 29, 26, 27, 34, 21, 30, and 36 all having high positive loadings. Non-interpersonal relationship situations (14, 6, 7, 5, 11, 13, 12, 8, 16, etc.) have high negative loadings on this dimension. 86 The second dimension more or less represents a split among the non- interpersonal relationship situations. Situations with high positive loadings include situations 12, 13, 15, 11, 14, 16, and 7. These are non-interpersonal relationships entailing an infringement of rights. The situations that load negatively on dimensions (situations 20, 23, 24, 40, 19, 21, 18, and 17) are those situations which entail non-interpersonal relationships and subsuming both bargaining and special favors situations. Thsi dimension can be labeled a "non-interpersonal relationship-infringe- ments vs. non-interpersonal negotiation" dimension. Situations that load positively on the third dimension include situ- ations 3, 1, 5, 9, 10, 2, 4, 6, and 8. Situations that load negatively on this dimension include situations 14, 16, 21, 22, 23, and 24. The positive loading situations correspond to an infringement of one's rights, while the negative loading situations correspond (in general) to the set of non-interpersonal relationship situations; more specifically to the subset of situations entailing a request for special favors. This dimen— sion, then, can be labeled an "infringements vs. requests" dimension. Situations with positive loadings on the fourth dimension include situations 35, 36, 34, s, 6, 8, and 17. These situations deal with per- suading someone to do something for the persuader's benefit. Situations with negative loadings include situations 38, 37, 39, and 9. All of these situations entail persuading someone to do something for their own benefit (i.e., help others). This dimension can be labeled a "per- sonal benefit vs. helping others" dimension. In sum, the four-dimensional solution can be labeled as follows: (1) "interpersonal relationship vs. non-interpersonal relationship"; 87 (2) "non-interpersonal—infringements vs. non-interpersonal negotiating"; (3) "infringements vs. requests"; and (4) "personal benefits vs. helping others." 1.6.4 Discussion There are three main topics to be discussed in relation to the re- sults of the exploratory study: (1) the classification of situations on the basis of being perceived as "functionally equivalent" situations; (2) the potential commonality between factors of situations across the domains of social behavior situations and persuasion situations; and (3) generalizability of the factors (or dimensions) and clusters of situa— tions to a larger population of situations. I concluded the discussion of the question of behavioral (personal- ity) consistency by noting that one can conclude that the individual be- haves "inconsistently" only when the individual perceives several situa- tions as functionally equivalent and yet behaves differently across the situations. While no behavior data was collected in the exploratory study, some observations can be made regarding the functional equiva- lence in the perceptions of situations. The situations provided the respondents were constructed in such a way as to vary along two criteria: "interpersonal vs. non-interpersonal relationships" and "commitment to future interaction vs. no commitment to future interaction." However, the respondents obviously perceived a. large number of additional differences between the situations. ,In con- structing the situations, no attempt was made to vary the situations in terms of who benefits, if one's rights were involved, whether the inter- action involved negotiating (bargaining) or requests for special favors, or even the intensity of the interaction. However, such concerns are 88 perceived to be important differences in the perceptions of the respon- dents. While a sorting task makes it difficult to uncover reliable indiv- idual differences in the perceptions of the situations, it would be of interest to conjecture about differences in behaviors and strategies employed across the clusters and dimensions of situations. Indeed, Bos- ter (personal communication) has noted that somewhat higher levels of threat (in first-attempt strategies) may be used in situations where the persuader is involved as persuading a close individual (interpersonal re- lationship) to do something for their own benefit. This can be explained by the fact that the negative consequences of generally using threat as a first attempt strategy are attenuated in such a situation: the relation- ship will not end or suffer other consequences that would (usually) re- sult by emlpoying threat for the persuader's benefit, etc. In general, however, I suspect that different strategies would be employed in special favors situations as opposed to helping others, and infringements of rights as opposed to negotiating. One can be more asser- tive or demanding when one's rights are involved; but one needs to be more tactful in negotiating. Simple requests are likely to be used in special favors (or, indeed, ingratiation), while strategies employed in helping others are likely to be quite varied. In regard to the commonality between the global social behavior domain of situations and persuasion situations, it would appear that there does exist a relationship factor (interpersonal vs. non-interperson— al relationship). Further, several factors relate to behaviors or feel- ings. Specifically, situations which deal with non-interpersonal in- fringements of fights are likely to be perceived as less friendly and 89 potentially more hostile than non-interpersonal negotiating. Also, sit- uations which form the general category of "infringements of rights" are likely to be perceived as more intense than the situations comprising the general category of special requests. Finally, the fourth dimension of persuasion situations deals with the goals or consequences of event-- personal benefits vs. helping others. This last dimension is undoubted- ly specific to the domain of persuasion situations. Finally, are these dimensions and clusters of persuasion situations generalizable to any larger sample of situations, such as to the domain of persuasion situations? Unfortunately, they cannot be generalized as such for the simple reason that neither cross-validation nor proper sam- pling from a population were employed in the study. The only warranted conclusion is that the obtained clusters and dimensions of situations obtained in this study describe the similarities and dissimilarities a— mong the situations specifically employed in this analysis. The problem of generalizability of factors of situations to a lrager sample is a tOpic I shall discuss in detail. 1.7 Criterion for Assessing "Generality" of Dimensions Obtained by Multidimensional Scalipg Techniques Several comments should be made concerning the general use of multi- dimensional scaling techniques. Early investigators were content to find that multidimensional scaling techniques yield several dimensions that were interpretable (Abelson, 1955; Jackson, Messick, and Solley, 1957). As Stewart (1974) noted, these findings were important in fostering later work in multidimensional scaling research; however, there are any number of meaningful and interpretable dimensions that could describe a set of 9O complex stimuli. If this were not true, then there would be no reason to employ multidimensional scaling. Therefore, a criterion must be sought, aside from interpretability, to assess whether or not dimensions obtained from multidimensional scaling are general-~are true dimensions and not artifactual. Three procedures are usually recommended for assessing such "gen- erality" of dimensions. One procedure for assessing generality of dim- ensions is to assess the degree to which a spatial representation can be used to account for behavior in a variety of independent tasks or sit- uations. This procedure is synonymous to construct validation. Usually, the procedure entails deriving dimension by a multidimensional scaling technique and then applying some model or rule to the resultant configu- ration to predict behavior in one or two tasks. Cliff (1972) found that an additive model could be used to predict the respondents' ratings of emotional connotation of adjectives. Cliff and Young (1968) found that the distance from a hypothetical origin in a configuration could be used to predict the respondents' ratings of the intensity of the emotions ex- pressed in photographs of facial expressions. Jones and Young (1972) found that an individual's socialization into an academic department could be observed by the implementation of data collection procedures over time. Jones and Young further found it possible to predict rele- vant communication phenomena from the obtained dimensions (i.e., inter- action frequency and selection of commitee membership from among the faculty). Cliff (1969), however, found weak results for predicting liking judgments from a multidimensional configuration of 20 academic fields. 91 It may be the case that some academic titles may be fairly unfamiliar to some individuals, whereas trait adjectives, emotions and familiar faculty members are objects within specific domains for which more reliable dim- ensions can be obtained (see below for a discussion of effects of aggre- gation). Nonetheless, many of these "construct" validation studies yield support for the claim that the obtained dimensions are meaningful. A second approach, used by Magnusson and Ekamn (1970), AIken and Brown (1969), and Wiggins and Hoffman (1968) consists of deriving the multidimensional scaling configurations in different ways, and then com- paring two configurations on the basis of similarity. The studies re- -iewed in Section 2, which compared the reaction mode analysis and per- ception approaches, fit into this category. For the specific samples of stimulus objects, these studies illustrate a good deal of similarity in solutions across procedures. Stewart (1974) argued that a third procedure be adopted. Stewart obtained a five-factor solution from factor analytic procedures, which were labeled Sociability, Ability to Supervise Others, Athletic Ability, Accept as an Intimate Friend, and Extremity. The factor analysis was done on data on four basic cues (traits) of 81 profiles. Stewart select- ed 15 profiles to be used in a multidimensional scaling analysis. This analysis provided a five-dimensional configuration and the multiple cor- relations between the five dimensions and the scale values of the 15 stimulus profiles on the four basic cues were .968 (excitable), .986 (cautious), .994 (sophisticated), and .987 (talkative). The two struc- tures were rotated such that four dimensions would correlate maximally with each of the four cues. This rotation enabled Stewart to identify the extremity dimension in the multidimensional configuration. Thus, 92 both the results of the factor analysis procedures and the multidimen- sional scaling procedures were quite compatible, and such compatability obviously enhances the confidence in the solutions of both techniques. Stewart notes, however, that such compatability may not often be obtained. First a factor analytic space may have one more dimension than the solution obtained when employing the multidimensional scaling procedures. In this case, Stewart argued that: Since it did not appear in the multidimensional scaling, its generality might be subject to some question and an investigation of the set of judg- ments would be indicated to determine if the spe- cific combination of judgments included could have caused such a factor to arise artificially. If not, then the generality of the multidimensional scaling dimensions would have been shown to be lim— ited. Given the multidimensional scaling alone, such a dimension would have been missed. Given the factor analysis alone, the dimension might have been accepted as general without further examina- tion. (p. 516) Second, a multidimensional scaling solution may have one more dimension than the factor analytic solution and contain the factor analysis as a subspace. Stewart noted that: In this case, the extra dimension would be singled out for study with regard to the following questions: Has some important component been omitted from the set of judgments used for factor analysis? Is this dimension introduced because of a specific character- istic of the type of psychological distance measure employed? (p. 516) The answers to these two questions hinge on the immediate replication of the study, with an inclusion cuiscale which should tap the additional dimension. To this list of procedures, I shall include a fourth. If it is the goal of a miltidimensional study to obtain a dimensional analysis of a set of objects that one can generalize the obtained dimensions to 93 a wider set of objects, then it seems legitimate to argue that a superior assessment of the generality of the multidimensional techniques lies in randomly sampling various sets of objects from a population or universe of objects, and then comparing the various solutions for the generality of dimensions. The degree to which various dimensions re—occur across random samples of objects enhances the confidence one might have for con- cluding that such dimensions are part of the respondents' perceptions of entire class of objects. Such a study also possesses a self-contained replication. 1.7.1 Effects of Aggregation Of the various criteria mentioned above, the second criterion de- serves some consideration. There are several reasons why a particular multidimensional scaling solution may be different using similar data and different techniques. One of the main reasons deals with the effects of aggregation. Anderson (1970) utilized the non-metric INDSCAL program to assess the dimensional structure of 12 adjectives. The average cor- relation between each individual's solution and the aggregate was .74. Thus, Anderson concluded that a configuration for the group did not dif- fer greatly from that of the individual. Anderson (1973) retracted this conclusion when the average correlation between each individual's config- uration for "public figures" and that of the aggregate was only about .38. Anderson recommended the utility of the INDSCAL model to assess the relationship between individual and aggregate. The INDSCAL model assumes that the aggregate solution is a solution common to all respondents and that different respondents will have dif- ferent dimensional salience weights, which are indicative of the emphasis 94 individuals placed on the dimension. To the author's knowledge, the only studies that have explicitly tested this assumption is a study by Sherman (1972) using trait adjectives, and a study by Carrol and Chang (1970) using tea preference data. Woelfel and Danes (1977), however, note that the INDSCAL solutions are not general if the stimulus structure is not invariant across in- dividuals. That is, if two individuals rank three objects in different orders, the solution will be a two-dimensional solution, as opposed to the one-dimensional solution obtained for the two individuals separately. This suggests that the generality of the INDSCAL solutions is a function of how many various points of view are sampled and the degree to which individuals comprising the points of view hold diverse viewpoints (name- ly, the extent to which objects are ranked differently across the points of view). For some sets of data (i.e., adjectives, colors, etc.), the diversity of points of view is not great and the dimensional solutions will be general. However, when the objects are complex (public figures, interaction situations, etc.), effects of aggregation may be more pro- nounced. In order to assess diversity (and ultimately the effects of aggregation), attention must be given to individual differences in per- ception. At this point in time, it is difficult to report on the extent of different points of view in the perceptions of situations. Pervin (1976) undoubtedly uncovered what might be called diverse points of view. How- ever, Magnusson (1971) and Magnusson and Ekehammar (1973) found individ- ual differences in the perceptions of academic situations to be trivial. Further, Wish, Deutsch, and Kaplan (1976) found that the indiVidual dif- ferences in the perceptions of role relations to be adequate handled o 95 within the INDSCAL assumptions, and Wish (1975) unfortunately is mute on the topic. Finally, Magnusson and Ekehammar (1975) found three homogen- eous samples of males and three samples of homogeneous females in terms of similarities of anxiety responses. In general, it would appear safe to conclude that either the INDSCAL model can adequately represent the extent of individual differences or that some small number of clusters of individuals can be identified. In sum, these disucssions concerning generality of factors and ef- fects of aggregation suggest, first, that several methods be employed in order to assess "generality" of dimensions. Of the various criteria, the investigation of the similarity between factor analysis and multidimen- sional scaling solutions would be the most informative. Second, sim- ilarities between solutions offered by several multidimensional scaling techniques would also be informative and help to identify any dimension or factor that a particular technique may yield artifactually. Further, it is possible to employ the INDSCAL model as one of the techniques to be compared and help to identify individual differences at the same time. Of the other criteria, it should be noted that the "construct validity" of the dimensions has been assumed throughout this paper. That is, I have assumed that the obtained dimensions can be related to differential strategies and behaviors. The commonalities between the response analyt- ic and perception approaches discussed above support the claim that the perceptions of situations yield factors which correspond to behaviors. Obviously, this assumption leads directly to the implications of future research for daily persuasion situations (Section 1.8). Further, attention should be paid to the question of generalizability of dimensions to situations which form a distinct domain of situations 96 (i.e., persuasion situations). Unless such generalizability can be dem- onstrated, one can only conclude that the obtained dimensions of situa- tions are specific to the sampled situations. Since it is usually the case that generalizability is desirable (indeed, if not an implicit as- sumption), it would seem appropriate to select a sample of relevant (to the respondents) situations inclusive of the domain of interest and drawn random samples of situations from this "population" of situations. In combination with the suggestion made above, one can eliminate problems of artifactual dimensions while generalizing "general" dimen- sions to the domain of situations. With these issues resolved, it be- comes possible to assess strategies and behaviors and outcome-expectancy hypotheses individuals find applicable according to pre-calibrated situ— ations. 1.8 Conclusions and Implications I have argued that the proper procedure for assessing behavioral consistency across situations must begin with a careful analysis of how individuals categorize events which comprise their social environments. This is necessary in order to understand what aspects of the meaning as- signed to situations influence the individual's selection of an "approp- riate" behavioral response. Understanding the emergence of a behavior- al response in a naturalistic situation can only be achieved by under- standing what the individual perceives as a "situational requirement." A review of the literature on the perceptions led to the conclusion that three classes of dimensions have been obtained; yet information con- cerning the generality of these dimensions across domains of situation- al types (let alone generality within a domain of situations) is sev- erely lacking. Further, information concerning individual differences-- 97 what accounts for them, how they are manifested in the perceptions of situations, and the degree or the scope of individual differences--is also severely lacking. Finally, I have discussed two methodological issues which need to be resolved concerning the "generality" of dimen- sions and the generalizability of dimensions. Thus, there are a number of general questions concerning how individuals categorize events which need to be considered in order to provide a complete, unbiased under- standing of the perceptions of naturalistic situations. Implications of this research on the perceptions of situations are enumerable. I will suggest only a few of the more important implications. First, it is informative in the sense of uncovering what the dimensions of these situations are, their generality and the extent of individual dif- ferences. Second, possessing the set (or various sets) of situations pre-calibrated along meaningfully interpreted, non-artifactual dimensions makes it possible to assess what strategies individuals select differen- tially according to a meaning system. "Strategies," of course, are what Mischel (1973) referred to as "behavior-outcome eXpectancies" and relate to onw way in which individuals vary in selecting "appropriate" modes of response. There are other variables that remain to be explored, such as one's "abilities" (Wallace, 1966) or construction competencies (Mischel, 1973). Such analyses as these entail differentiating individuals on the basis of ability to conceptualize a number of various strategies approp- riate to the pre-calibrated situations which are likely to be successful, and on the abilities of executing any number of such devised alternatives. Furhter, in order to assess the "consistency" of behavior, one would also have to assess the values individuals place on the outcomes of various situations. 98 NOTES I wish to acknowledge the assistance of Frank Boster and Eileen Thomp- son for collection of this data. I wish to acknowledge Ed Kaminski for assistance in this task. Several of the situations dealing with "friends" and "roommates" in the list of situations in Appendix A were switched to "close friends," "acquaintance," or "neighbor" for this task. I wish to acknowledge the assistance of Dr. G. R. Miller, Mr. Ed Kaminski, and Mr. Frank Boster in writing these situations. Because of space requirements in Figure 1, the entry levels are not included. The entry levels can be obtained from the author. Write to the author at the Department of Speech Communication, Texas Tech University, Lubbock, Texas 79409. Because the presentation of a 4 x 40 matrix is somewhat difficult to present here, I have taken the liberty of not presenting the solu- tion. The printout is available to those interested. Write to the author at the Department of Speech Communication, Texas Tech Univer- sity, Lubbock, Texas 79409. CHAPTER 2 Four topics will be discussed in this chapter. First, a pretest for nominating situations was conducted in order to select situations relevant to the social environments of the individuals. Second, after a set of rel- evant situations was selected, three sets of twelve situations were random- ly sampled (with replacement) to be analyzed. Third, I shall discuss the procedures and materials used in the study. Fourth, an analysis of the respondents' reactions to the study will be presented. 2.1 Pretest for Nominating Situations Since any one individual could have nominated a situation to be in— cluded in the list of situations in Appendix A, the relevancy of any situ— ation is not guaranteed. Therefore, a pretest was conducted in order to assess the relevancy of the situations. In the Fall of 1976, 38 individuals --15 males and 23 females--volunteered to complete a questionnaire which presented 93 of the situations from Appendix A and required the respondents to circle +1 if the situation "definitely applies to them," 0 if the situa- tion "possible applies to them," and -1 if the situation "does not apply to them." All respondents, who were enrolled in either Communication 199 or 205, received course credit for their participation. The situations were grouped on the basis of "parents," "roommates," "friends," "people you know," "professionals," and "strangers." After each of the sets of situ- ations, respondents were asked to list four additional situations which were relevant to them. Two lists of situations were constructed: a list of the "most frequently nominated" situations and a list of the "least objectional" 99 100 situations. The first list was based on the frequency by which the re- spondents reported the situations as applicable to them. The second list was based on the frequency by which the respondents reported the situations as not applicable to them. There was very little difference between the two lists of situations, with the least objectionable set of situations containing 12 more situations than the most frequently nominated situa- tions. The least objectionable set of situations will be used as the start- ing basis for the main study. Selection of this list, over the most fre- quently nominated list of situations, helps ensure that both very familiar and unfamiliar situations would be provided the respondents, so that the familiarity distinction will not be severely attenuated. The list of selected situations is provided in Appendix C. The fre- quency of nomination is listed behind each situation, followed by the fre- quency of rejection. Both are presented as percentages. Several of the situations included in Appendix C were included from the free response nominations given by the respondents in this pretest. These were includ- ed on the basis that at least two (if not more) individuals freely nominat- ed the same situation (or one similar to it). This list contains 87 situ— ations. 2.2 Selection of Situations To assess the robustness and replicability of the dimensions ob- tained in the perceptions of these situations, a random number generator program was utilized to provide 12 numbers at random within the limits of 1 and 87, inclusive. Three sets of situations were selected. Situations in set one included: M mummcw 12. Situations (II-FCOMH HOLDmQO) 1 1 . 12. Situations 1. M 0:014:00 \l 9. 10. 11. 12. 101 Persuade your roommate(s) to leave phone messages for you while you are out. Persuade a professor to let you enroll in an already crowded class. Persuade a group of people to stop teasing a friend of yours. Persuade your boy/girlfriend to meet your relatives. Persuade a friend not to swear so much. Persuade your friend to let you stay out late. Persuade your boy/girlfriend not to date another person. Persuade a group of people to order a pizza because you feel like eating one. Persuade a friend to take a class with you. Persuade your father to take more time off from work and find something relaxing to do. Persuade a bouncer to let you into a bar even though you have forgotten your I.D. Persuade your parents to give you money. in set two include: Persuade your boss to let you leave work early. Persuade a group of people to go to the bar with you. Persuade a professor that an answer on a test is incorrect. Persuade a friend to go shopping with you. Persuade your boy/girlfriend to go to a party where there is mostly your friends. ' Persuade your boy/girlfriend not to date another person. Persuade a friend to give you a ride. Persuade a sister/brother to go to a four—year university. Persuade a boy/girlfriend to study more. Persuade a person to ask you out. Persuade a group of people to participate in an experiment for your group project. Persuade parents that marijuana is not bad. in set three include: Persuade a boy/girlfriend that there is nothing wrong in occasionally dating another person. Persuade your roommate to keep quiet when you are trying to sleep. ' Persuade your boy/girlfriend to confide in you more. Persuade a professor that a test item was ambiguous. Persuade a professor that his grading scale is too hard. Persuade a person you know in class to help you with your hcmework. Persuade your landlord to fix the plumbing in your apartment. Persuade an old boy/girlfriend that the relationship is worth saving. Persuade your father into getting a check-up. Persuade a police officer not to give you a ticket. Persuade a professor to increase your grade. Persuade your mother to go back to college and get a degree. 102 It should be noted that this procedure entails a random sampling of situations from a population of situations (Appendix C) with replace- ment. Therefore, the obtained dimensions of situations can be generalized to the population of situations. 2.3 Procedures Employed in the Study The study was conducted in the Winter of 1977. Two hundred sixty- three individuals enrolled in Communication 100 and 101 classes volun- teered to participate. All received courses for their participation. 2.3.1 Materials All respondents received two questionnaires. The first question- naire included instructions on the use of paired comparison scales and presented all non-redundant pairs of 12 situations. The instructions of an assumed interval scale are presented in Appendix D. Woelfel (1974) has argued that a ratio scale will provide more precise measurement of multi- dimensional scaling solutions. For this reason, an additional group of individuals was used and was provided an aesumed ratio scale. Instruc- tions for the assumed ratio scale are included in Appendix E. Four different types of paired comparison questionnaires were con- structed. Form 1 utilized an assumed interval scale and included situa- tions in set one. Form 2 utilized an assumed interval scale and includ- ed situations from set two. Form 3 utilized an assumed interval scale and included situations from set three. Form 4 utilized the assumed ratio scale and included situations from set three. Joelfel, Cody, Gilham, and Holmes (1977) have noted that the first reponse on a paired comparison questionnaire is usually the least reliable. Therefore, a dummy first pair was included and not used in the analyses. The pairs of situations 103 were not randomized, but presented row-wise. The first questionnaire also contained questions concerning demographic information (i.e., sex, age, race, major, and year in school). A second questionnaire included instructions on the use of the at- tribute scales and 23 attributes on which each of the 12 situations were to be rated. Some discussion of how these 23 attributes were selected in in order. In order to assess the perceived level of friendliness of a situa- tion, the following set of attributes was selected: I. Friendly vs. unfriendly friendly-unfriendly easy to resolve conflicts-difficult to resolve conflicts confortable-uncomfortable frank and open~reserved and cautious Since the type of relationship varies considerably across these sampled situations, several attributes were employed to assess perceived levels of intimacy: II. Superficial vs. intimate superficial-intimate impersonal-personal and several attributes were employed to assess perceived levels of status and/or formality of the situation: III. Formal vs. informal formal-informal I have relatively low status-high status one totally dominates-treat each other as equals However, since the disctibution of social power may be of a nature that is only partly related to status or intimacy, two attributes were included to assess general influencibility: 104 IV. Influencibility s/he has a major effect on my life-does not have a major effect I have a major effect on his/her life-I do not have a major effect Two attributes were employed to assess perceived levels of differential benefits: V. I benefit vs. I do not benefit I benefit-I do not benefit persuade him/her to do something for his/her benefit-not to do something for his/her benefit and two attributes were employed to assess familiarity: VI. Familiarity I have been in this situation many times before-I have never been in this situation_before I know how to behave-I do not know how to behave Since the situations under investigation deal with persuasion, sev- eral attributes were included in order to assess perceived levels of diffi- culty in persuasion: VII. Easy to persuade vs. difficult to persuade easy to persuade-difficult to persuade resistant to persuasion-not resistant to persuasion I must be assertive-I need not be assertive and one attribute was used to assess the general type of strategy prefer- ence: "I would use a short term strategy-long term strategy." Several additional attributes were also employed. One attribute was employed to assess whether or not the situation involved a favor or involved a right of the respondent. With the Miller and Steinberg (1975) categories in mind, the attribute "no commitment to future interaction- commitment to future interaction" was included. Further, the attribute "emotional-unemotional interaction" was included to assess perceived lev- els of emotional feelings. Finally, the attribute, "s/he will think less 105 of me-will not think less of me” was included in order to assess whether or note the respondents perceive that some situations entail a negative altercasting or a blemish of their perceived character. ‘ In relation to the attributes found by Forgas (1976) and Wish, Deutsch, and Kaplan (1976) to correspond to dimensions of situations (see Table 2), I have included two sets of attributes which could be subsumed into the set of "friendly vs. unfriendly" attributes. Further, I have subsumed the status and formal vs. informal set of attributes into one category. Further, for the present study, I have eliminated attributes dealing with work and task-oriented projects, but have added attributes which deal with the perceptions of difficulty of persuasion, influencibil- ity, benefits and have included attributes which should assess familiar- ity (see Forgas, 1976). The 23 attributes were re-ordered enni many of the end points re- versed in order to avoid response bias. The order by which attributes were presented is provided in Appendix F. Instead of randomizing the at- tributes for each situation, the same order of presentation was employed for each situation. Since the second questionnaire was a lengthy quest~ ionnaire, requiring 276 responses, the order by which situations were presented was reversed for one-half of the respondents. That is, one- half of the respondents in each of the four groups received the situations in the order as they are listed above, and one-half of the respondents received situation 12 first, followed by 11, etc. This was done in order to counterbalance fatigue effects. 2.3.2 Procedures Data was collected during four test sessions. Respondents were pre- sented in the first questionnaire, instructed to read the instructions to 106 themselves and to take as long as they needed in order to complete the questionnaire. When the respondents returned the first questionnaire, they were given the second questionnaire. Respondents were assigned to one of the four groups on the basis of their arrival at the test session. Therefore, individuals were randomly assigned to one of the groups during each of the test sessions. Every other individual who completed a paired comparison questionnaire of a particular type received the second questionnaire of a different order of presentation of situations. 2.3.3 Respondents Two hundred sixty-three individuals participated in the study: 64 were assigned to Form 1 (29 males and 35 females); 67 were assigned to Form 2 (21 males and 46 females); 66 to Form 3 (17 males and 49 females); and 66 to Form 4 (23 males and 43 females). Most respondents were Cauca- sion. There were 10 non-Caucasians assigned to Forms 1 and 2, seven to Form 3, and nine to Form 4. The mean ages of respondents, for the four groups, are 20.3, 19.3, 19.2, and 19.5, respectively. 2.4 Analysis of Respondents' Reactions to the Study Three open-ended questions were presented to the respondents at the end of the second questionnaire which allowed them to (a) suggest addition- al attributes not included in the study; (b) to report whether they consid- ered their participation in the study as laborious or interesting, and (c) report which, if any, of the situations presented in their questionnaire did not apply to them. Thus, an analysis of the open-ended questions pro- vides an indivation of the exhaustiveness of attributes provided the re- spondents, the feasibility of boredom (or fatigue), and a further indication 107 TABLE 2 Attributes Used to Interpret Dimensions Obtained by Wish, Deutsch,emuiKaplan (1976), and by Forgas (1976) Wish et al.: . Forgas: FRIENDLY vs. UNFRIENDLY harmonious—clashing friendly-unfriendly cooperative-competitive pleasant-unpleasant friendly-unfriendly informal-formal compatible goals-incompatible intimate-superficial productive-destructive at ease-ill at ease easy-difficult to resolve conflicts altruistic-selfish fair-unfair relaxed-tense INVOLVED-UNINVOLVED active-inactive involved-uninvolved intense-superficial interaction intimate-superficial intense-superficial feelings interesting-dull STATUS equal-unequal status similar-different roles FORMAL-INFORMAL important-unimportant to society pleasure-work oriented formal-informal FAMILIARITY know how to behave- do not know at ease 108 if any situation is not a common, relevant situation. Sixty-four respondents completed questionnaire Form 1. Thirteen respondents left the question concerning additional attributes blank. Of the remaining respondents, 15 males and 17 females reported that they could offer no remaining attributes. The remaining individuals reported a wide range of attributes. Two males and two females reported that the "impor- tance" of the persuasion was a distinction they made. One male suggest- ed a difference between public persuasion and more private persuasion set— tings. One male suggested a difference between being sincere or insincere. Of the 67 individuals in the Form 2 group, 14 left the question blank, 14 of the 21 males and 28 of the 46 females indicated that they could not suggest additional attributes. Two individuals noted that in some situations, damage to future communication was eminent. Three in- dividuals stated that all situations were unique. Of the 66 individuals in the Form 3 group, eight individuals left the question blank, and seven of the 17 males and 29 of the 49 females in- dicated that they could not suggest additional attributes. Two females suggested that the emotional strength required to engage in the situation was an important consideration for them. Two females suggested that the importance of the persuasion was an important consideration. One male and one female suggested that whether the persuasion had short term ef- fects on the relationship or long term effects on the relationship was an important consideration. One male indicated that the likelihood of fre- quency of situation occurrence was worthy of consideration, and two females noted that each situation is unique. Of the 66 individuals in the Form 4 group, nine individuals left the question blank, and 14 of the 23 males and 26 of the 43 females 109 reported that they could not suggest additional attributes. Two females and one male noted that the impersonal-personal distinction was important, one male responded that who benefits was an important consideration, and one male and one female noted that a high regard for the other person's opinion was an important consideration. One individual suggested that perceptions of situations vary more as a funtion of the other interactant than the nature of the goal of the persuasion, particularly situations deal- ing with professors. In sum, most of the individuals felt that the attributes employed in the present study were quite exhaustive. Many of the suggested attri- butes were already included in the study (i.e., benefits, impersonal- personal, emotional-unemotional, etc.). However, two attributes, per- ceived degree of importance and the degree to which the situation affects the relationship, are two attributes worthy of future consideration in studies of this type. Of the 66 individuals in the Form 1 group, eight males and 11 fe- males perceived their participation in the study as monotonous, five males and six females as laborious, six males and 12 females as easy, and eight males and 18 females perceived the study as interesting. For the Form 2 individuals, nine males and 10 females perceived the study as monotonous, eight males and 17 females as easy, three males and one female as labori- ous, and eight males and 19 females as interesting. For the Form 3 individuals, six males and nine females perceived the study as monotonous, seven males and 14 females as easy, one male and four females as laborious, and seven males and 22 females as intersting. For the Form 4 group, eight males and 11 females perceived the study as monotonous, three males and 13 females as easy, six females as laborious, 110 and 10 males and 23 females as interesting. It would appear that the respondents, by and large, perceived their experience in this study as an interesting one. Very few individuals per- ceived the task as laborious; most perceived it as interesting and ap- proaching the point of monotony. In general, the results of the analysis of the third open-ended question confimrs the results of the Pretest for Nominating Situations. Nearly all of the situations sampled are common, relevant situations for the majority of individuals. For the Form 1 individuals, seven females and three males reported that they would never be in the situation "persuade a boy/girlfriend not to date another person." This means that less than 10 percent of the respondents reject this situation as being relevant. On- ly seven respondents left this open-ended question blank. For the Form 2 group, eight individuals left the question blank, and only two situations were rejected by a number of individuals. The situation "persuade your parents that marijuana is not bad" was rejected by 19 (of the 46) females and by eight (of the 23 males). The situation "persuade a boy/girlfriend that there is nothing wrong in occasionally dating another person" was rejected by 10 females and four males. For the Form 3 group, eight individuals left the question blank, eight males and five females rejected the situation "persuade your mother to go back to college and earn a degree," seven females and four males rejected the situation "persuade a police officer not to give you a tick- et," 10 females and one male rejected the situation "persuade a profes- sor that his grading scale is too hard," and five females and three males rejected the situation "persuade a boy/girlfriend that there is nothing wrong in occasionally dating another person." These situations were 111 rejected by more or less the same number of individuals in the Form 4 group. After respondents completed the questionnaires, they were given a debriefing document to take with them which described the goals of the ex- periment. CHAPTER 3 RESULTS Overview In order to assess the interrelationships among the attributes, a factor analysis of the unidimensional attributes was done. The results of this analysis are presented in Section 3.1. The stability of situations in the multidimensional spaces was done in two ways. First, configurations for random split-halves of the respondents for each of the four groups were compared. Second, configur- ations for males and females for each of the four groups were compared. A metric multidimensional scaling program (GALILEO), which has the capa- city of performing least-squares Procrustean rotation was used in these comparisons (see Cliff, 1966; Schonemann, 1966; Woelfel, Saltiel, McPhee, Danes, Cody, Barnett,and Serota, 1975, for a discussion of this rotation procedure). The results concerning stability of the locations of situa- tions in the multidimensional spaces will be presented in Section 3.2. The paired comparison data for each of the four sets were analyzed using three multidimensional scaling models (GALILEO, KYST [with TORSCA Option] and INDSCAL) and Johnson's (1967) hierarchical clustering program. Results from the utilization of all three multidimensional scaling pro- grams which have a bearing on determining the number of usable dimensions will be presented in Section 3.3. It will be argued in Section 3.3 that there are four usable dimen- sions. Zero order correlations between the four dimensional (HIRE) solutions and the first four dimensions obtained by utilizing the metric 113 (GALILEO) program are as follows: -for Form 1: —.99, .99, .99, and .98; -for Form 2: .99, -.95, .97, and -.75; —-for Form 3: -.93, .95, .97, and .97; and, —-for Form 4: -l.00, .99, -.93, and .93. A rotation of the KYST configurations to that of the GALILEO configura- tions would do little to the configurations except to reverse the signs on the appropriate dimensions. It would appear that the major features of the perceptions of situations are equally represented in the GALILEO and KY§I_solutions. Therefore, only the results of using the KY§T_model will be presented in this paper. The GALILEO program will be employed, when appropriate, for comparing various solutions. Thus, three sections of this paper deal with the psychological interpretation of the dimensional structures of clusters obtained from the other three analytic tools. Section 3.4 presents the psychological interpretation of the four dimensional (§Y§E) solutions. In Section 3.5, I will discuss the results of hierarchically clustering the situations in relation to the interpretation of dimensions presented in Section 3.4. In Section 3.6, I shall discuss (briefly) the interpretation f the INDSCAL solutions, the dimensional salience weights for all respondents and a clustering of the respondents on the basis of correlations among dimensional salience weights. Note that in Section 3.2 I discuss the fact that some situations are perceived differently by males and females. To assess these sex differ- ences, coordinate loadings for situations for males were regressed onto the mean ratings (for males) on attributes. The same was done fer the females in each group (the coordinate loadings for situations from a 114 configuration which was rotated to the male configuration was used). For all practical purposes, there was essentially no difference in the inter- pretation of the first four dimensions. Apparently, there is not a male point of view, or a female point of view. For pragmatic reasons, this regression analysis will not be presented in this paper. Other points of view will be analyzed. Repeatedly throughout this section of the paper I will make refer- ence to the Operation of various points of view. No less than eight distinct clusters of individuals were obtained in Section 3.6. Respond- ents representative of two clusters for one of the sets of data will be identified, and, in Section 3.7, I will compare the multidimensional con- figurations for these individuals. In Chapter 4 of this paper, I shall discuss the implications of the obtained interpreted (aggregated) dimensions and the implications concern- ing the points of view that are obtained concerning future research on (a) the perceptions of situations, and (b) the influence of perceptions of situations on behavior. 3.1 Factor analysis of unidimensional attributes Recall that attributes were chosen in such a way as to assess the dimensions that were obtained by Wish, Deutsch and Kaplan (1976) and by Forgas (1976). Further, several attributes were included from the exploratory study and from other literature. However, neither the Forgas or Wish, Deutsch and Kaplan studies reported an assessment of the interrelations of the attributes separate from the obtained multidimensional scaling solutions. An assessment of the interrelationships among the 23 unidimensional attributes is of 115 interest for two reasons. First, such an assessment would be of benefit to future research in (a) constructing indices of attributes, (b) select- ing attributes which may be better suited to measure a single construct or factor of perceptions, and (c) eliminate attributes which do not measure a single factor (i.e., have high cross loadings between multiple factors). Thus, a factor analysis of the unidimensional attributes was performed, and I shall make some recommendations for future research. The second reason for such analysis relates directly to the paper at hand. With the knowledge of how attributes form into factors, a discus- sion of how each dimensional representation relates to the attributes can be simplified. That is, instead of discussing how each dimension corres- ponds to each of the 23 unidimensional attributes, it would be more con- venient and parsimonious to discuss how each dimension corresponds to each factor of attributes. Further, the degree to which commonalities exist between the factor analytic and multidimensional scaling solutions is one criteria for assessing the "generality" of the dimensions. In order to assess the interrelationships among the attributes, a factor analysis (principal components with interactions; varimax rota- tion) was performed. Table 3 presents the results of this factor analy- sis (n = 263). Six factors had eigenvalues greater than unity. A six factor solution accounts for 59.2 per cent of the variance. Factor one contains four attributes which load highly only on one factor: "formal-infermal" loads .608; "one totally dominates-treat each other equally" loads .619; "s/he will think less of me-will not think less of me" loads .642; and "I have relatively low status-high status" loads .315. Further, the following attributes have cross loadings with other factors: (with factor five) "friendly-unfriendly", "difficult to 116 TABLE 3 Results of Factor Analysis of Unidimensional Attributes (n=263) Factors: Attributes l 2 3 4 5 6 friendly-unfriendly -.435 .025 .002 .083 .417 .150 difficult-easy to resolve conflicts .441 .173 .003 .088 -.373 -.075 short term- long term strategy .019 .124 .014 .028 .186 .096 comfortable- uncomfibrtable -.536 .039 -.O65 .110 .622 .155 resistant-not resistant to persuasion .288 .121 -.O7l .312 -.587 .101 easy to persuade- difficult to persuade -.l48 -.075 .159 .051 .849 -.009 I benefit- I do not benefit .022 .033 .105 .578 .006 -.091 I have been-I have never been in this situation .085 -.030 .319 .253 .310 252 involves a favor— involves a right -.O47 -.O74 .031 .106 .179 .639 do something for his/her benefit-not do something far his/her benefit .122 -.00’4 .261 .358 .187 -.202 I have relatively low status-high status .305 .027 -.006 .140 -.133 .140 fbrmal-informal .607 .136 -.072 .104 -.079 -.119 one totally dominates- treat each other equally .619 .124 -.043 -.021 —.133 .003 TABLE 3 (cont'd. )' 117 Factors: Attributes 1 2 3 4 5 _ 6 superficial-intimate relationship .337 .740 -.192 .080 -.172 -.l79 no commitment to future interaction-commitment .041 .457 -.152 -.075 .129 .087 frank and open- rgsgrvgd and cautious -.420 -.063 -.034 .203 .499 .089 impersonal- personal relationship .193 .871 -.087 -.018 -.l29 -.169 emot ional -unemot ional situation .313 -.229 .243 .329 -.058 - 121 I know-I do not know how to behave -.182 .020 .181 .266 .594 .062 he/she will think less of “will not think less .642 .128 .111 .036 -.036 124 he/she has a major effect- does not have a major effect on my lifb .034 —.l94 .792 .120 -.001 .026 - I have a major effect- do not have a major - effect on his/her life -.078 -.170 .829 .061 .195 .007 I must be assertive- I need not be assertive .079 -.060 -.022 .293 -.041 .220 118 resolve conflicts-easy to resolve conflicts", "comfortable-uncomfortable", and "frank and Open-reserved and cautious". It would appear, however, that this factor contains only the four attributes which load cleanly on this factor. This factor is a "formal-informal" factor. Factor two consists of three attributes with clean loadings: "superficial-intimate" loads .740; "no commitment-commitment to future interaction" loads .457; and, "impersonal-personal" loads .871. This factor is a "superficial-intimate" factor. Factor three consists of two attributes: "s/he has a major effect on my life-does not have a major effect on my life" loads .792; and, "I have a major effect on his/her life-I do not have a major effect" loads .829. Thus, the two items which were included in order to assess overall influencability among the participants appears to be fairly independent of both the more formal or informal situations and fairly independent of the degree of intimacy of the relationship. Because of problems of cross loading attributes, discussion of the fourth factor is a bit more problematic. There are two attributes which have their highest loadings on the fourth factor: "I benefit-I do not benefit" loads .578; and, "do something for his/her benefit-not do some- thing for his/her benefit" loads .358. One attribute has cross loadings with the fourth and fifth factor: "resistant to persuasion-not resistant to persuasion". Further, the following two attributes have small load- ings across a number of factors, but load highest on the fourth factor: "emotional-unemotional" and "I must be assertive-I need not be assertive". However, I will not attempt to categorize attributes with high cross loadings as belonging to a distinct factor. It should be sufficient to conclude that the two items which attempt to assess who benefits from the 119 persuasion form a distinct unit and represent the fourth factor. As noted above, a number of unidimensional attributes have high cross loadings between the fifth factor and other factors. Attributes which have fairly pure loadings on the fifth factor include: "easy to persuade-difficult to persuade" loads .849, and "I know how to behave-I do not know how to behave" loads .594. Of the attributes which cross load with factors one and five, the "comfortable-uncomfortable" attribute loads highest on the fifth factor (.622) and the "frank and open-reserved and cautious" attribute also loads highest on the fifth factor (.499). Of the attributes which load on both factors four and five, the "resistant-not resistant to persuasion" attribute loads highest on the fifth factor (-.587). This factor appears to be an "easy to persuade-difficult to persuade" factor and I shall include in this factor the following four attributes because I feel that these attributes jointly assess perceived resistance: (1) "easy to persuade-difficult to persuade", (2) "I know how to behave-I do not know how to behave", (3) "comfortable-uncomfortable", and (4) "resistant-not resistant to persuasion". While I have labeled this factor an "easy to persuade-difficult to persuade factor", it is interesting to note that the "I know how to behave-I do not know how to behave" attribute has more to do with this factor than with the formal-informal, intimate-superficial or influenca- bility factors. Evidently, perceiving difficulty in persuading another has a lot to do with not knowing how to behave in the face of resistance and feeling undomfortable during the persuasion. Further, these rela- tionships are somewhat (according to the factor analysis) independent of the intimacy and influencability of the relationship. These attributes 120 however, may be somewhat related to the perceived formality of the situa- tion, given the high cross loadings of such attributes as "friendly- unfriendly", etc., across factors one and five. The sixth factor includes only one attribute. It is the "favor- rights" factor, since the attribute "involves a favor-involves a right" loads .639 on this factor and does not cross load with any other factor. To summarize, the following six factors have been obtained: I. Formal-informal l. formal-informal 2. one totally dominates-treat each other equally 3. s/he will think less of me-will not think less 4. I have relative low status-high status II. Superficial-intimate l. impersonal-personal 2. superficial-intimate 3. no commitment to future interaction-commitment III. No major effect-influencability 1. I have a major effect on his/her life-I do not 2. s/he has a major effect on my life-does not IV. I benefit-I do not benefit 1. I benefit-I do not benefit 2. do something for his/her benefit-not do something for his/her benefit V. Easy to persuade-difficult to persuade . easy to persuade—difficult to persuade I know how to behave-I do not know how to behave comfortable-uncomfortable resistant-not resistant to persuasion $500M?" VI. Favors vs. rights 1. involves a favor-involves a right Further, there are several attributes which fail to be categorized into any factor: (1) "friendly-unfriendly" relates to both the formal and perceived resistance factors; (2) "difficult to resolve conflicts- 121 easy to resolve conflicts" relates to both the formal and perceived resistance factors; (3) "emotional-unemotional" relates to both the for- mal and perceived resistance factors; and (4) "frank and open-reserved and cautious" relates to both the formal and perceived resistance factors. Three attributes do not relate to any factor: "I have been-I have never been in this situation"; "short term-long term strategy"; and "I must be assertive-I need not be assertive". Therefore, in the discussion below concerning the psychological interpretation of the dimensions of situations, reference will be made to these seven attributes. Recall that Wish et a1. obtained a four dimensional structure (by regressing the dimensional coordinates onto the unidimensional attri- butes) which were labeled as follows: (1) "friendly-unfriendly"; (2) "involved-uninvolved"; (3) "status"; and, (4) "fOrmal-informal". At first glance, it may appear that the six factor structure I have obtained suggests some major discrepancies from the structure obtained by Wish et a1. However, there exists a major methodological difference in how these two structures were obtained. In factor analysis, the goal is to reduce a number of attributes onto a m_matrix of coordinate values based on the zero order correlations among the attributes. There is nothing (as McLaughlin [1975] and Cody, Marlier,anxiWoelfel [1975] have noted) to ensure that the m_factors obtained by factor analysis are relevant to the perceptions of objects rated, nor is it possible to discover what distinctions exist among the objects rated aside from the factors (formed by the attributes) given the respondents for their use. 122 Multidimensional scaling requires paired comparison data, in which respondents rate levels of perceived similarity (or dissimilarity) among the objects. From a matrix of interpoint mean distances, orthogonal axes are obtained which correspond to the major distinctions the respondents perceive among the objects. To interpret the dimensional solution, it is possible to regress the coordinate loadings for r_number of dimensions onto the median ratings (if the sample size is small) or the mean ratings (if the sample size is large enough to believe the mean is robust). Obviously, one would expect an immediate resemblance between the factor structure obtained by factor analysis and that obtained by regres- sion of coordinate loadings onto attributes. However, such a resemblance may be substantially weaker than one would expect. While Stewart (1974) found a fairly strong resemblance between a factor analytic solution and a multidimensional scaling solution for trait terms, there is no guaran- tee that the two types of procedures will provide similar solutions in the present study. A high degree of similarity may be obtained by com- paring the factor analytic solution for each of the three sets of situa- tions with each respective multidimensional scaling solution, but an overall factor analysis of the ratings across all three sets of situa- tions would be based on the degree to which attributes consistently cor- related across the various sets of situations. Since it is very likely that some subset of attributes will correlate less with another subset of attributes in some set of situations than in another set of situations, the degree of correspondence between each of the four multidimensional scaling solutions and the overall factor analytic solution will be atten- uated. Nonetheless, the overall factor analytic solution is useful in providing a convenient means for interpreting the multidimensional 123 scaling solutions and some comparisons between the overall factor analy- tic solution and the four multidimensional scaling solutions will be noted below (see below). Before discussing the dimensional configurations of situations, a final comment is in order. The factor analysis of the unidimensional attributes was done in order to facilitate a discussion of the psycholo- gical interpretation of the dimensional configurations. It was also done in order to discuss ways in which to improve the factor analysis in future research. Obviously, nothing has been gained, from a factor analysis point of view, by including the attributes which did not fit into any of the factors. However, the factor analysis of unidimensional scales is a fragile procedure and results are strongly influenced by the inclusion (or deletion) of additional items. For example, it was origi- nally felt that the two familiarity items ("I have been-I have never been in this situation" and "I know how to behave-I do not know how to behave") should have loaded highly on the same factor. The inclusion of several additional items, such as "occasional-regular", may enhance the emergence of such a factor. Second, several items would be needed to better assess the "rights vs. favors" factor, such as "I have a moral obligation to " etc. Several addi- make this request-I do not have a moral obligation, .tional items such as these would be necessary if indices were to be con- structed. Further, it is discouraging to note that attributes such as "I must be assertive-I need not be assertive" are not included in a factor. From a factor analysis point of View, such items would be eliminated from a second wave of data collections (if "pure" factors were to be obtained). This is unfortunate, from the perspective of the dimensional 124 analysis, because the attributes which have high cross loadings prove to be extremely helpful in discussing individual differences in perceptions. 3.2 A El] I 1.1-I E cl I. o I] _ lloio - . J E. gmzationsl To assess reliability of the aggregated paired comparison ratings, data from respondents in each of the four groups were randomly split into two halves. Each split-half was factored and one of the split-half configurations was rotated onto the respective second split-half config- uration. A least-squares Procrustean rotation was used (Cliff, 1966; Schoneman, 1966; see Woelfel and Danes, 1977). Correlations were com- puted between the loadings of the situations for respective columns of the factor matrices. These column correlations indicate the degree of stability between dimensions. Correlations were also computed between respective rows of the factor matrices. These row correlations indicate the degree of stability of each of the situations across the split-half configurations. In computing these correlations, loadings of situations across all of the real dimensions were used (usually eight or nine of the dimensions were real -accounting for 98 percent or 99.percent of the variance). Finally, the distance between each situation's location in the first split-half configuration and the second split- half configuration was obtained. Before discussing the stability information, it should be noted that departures from stability can either suggest unreliability in the respon- dent's distance estimates or suggest the operation of varying "points of view". It is impossible to distinguish the two effects. In utilizing paired comparison data, averaging across divergent points of view would 125 yield a solution that may represent only the more salient aspects of a particular point of view. Thus, if unstability is observed, the prOper course of action is to question the random assignment of respondents to each split-half and to suggest that there are various points of view that are operative. If the points of view cannot be parsimoniously explained, then one must conclude that the data are unreliable. I shall discuss the question of points of view in detail in Section 3.7. Figure 2 presents the three dimensional representation of the loca- tions of the 12 situations in the split-half comparison of Form 1 data. It would appear that situations 2, 7, 8,5nu110 exhibit some change. Such a conclusion, however, is unwarranted because there is only one view of the comparison in Figure 2 (one angle by which the viewer perceives the three dimensional configuration) and only three dimensions are represented. Table 4 presents the distance between the location of each situation in the first and second split-half comparisons. For the split-half com- parison of Form 1 data, the mean distance between situations is 1.955 (see the first column of Table 4). The situation that exhibits the most change is situation six ("persuade your parents to let you stay out late") which "moved" a distance of 4.009 units. Further, the seventh situation ("persuade your boy/girlfriend not to date another person"), which moved a distance of 2.759 units, exhibits some change. All other situations in this comparison appear to be quite stable. Column correlations for the split-half comparisons are presented in Table 5. For the split-half comparison of the Form 1 data, the first three dimensions are very stable (see the first column of Table 5). The stability of the fourth dimension drops (r = .756), but the stability of dimensions five through seven are fairly high (.814-.871). Finally, the 126 FIGURE 2 Three-Dimensional Representation of 12 Situations for Split-Half Comparison of Form 100 Data X-Y PLflNE \ + { X-l PLRNE if Y-Z m 127 TABLE 4 . . . . . a Distances Between Situations in Split-Half Comparisons Split-half Split-half Split—half Split-half Situation FORM 1 FORM 2 FORM 3 FORM 4 (n=32, 32) (n=34, 33) (n=33, 33) (n=33, 33) 1 1.545 4.664 1.431 22.147 2 2.154 2.545 1.569 33.087 3 1.006 4.283 2.264 20.408 4 1.220 1.581 2.293 21.729 5 1.789 .881 2.018 26.380 6 4.009 4.425 .773 29.127 7 2.759 3.104 .701 22.180 8 1.359 6.706 1.870 23.385 9 1.980 5.254 1.194 27.697 10 2.380 1.281 .872 18.578 11 1.293 2.607 1.949 14.109 12 1.914 3.832 1.369 23.185 Mean distance i 1.955 3.283 1.525 23.501 a For Forms 1, 2 and 3, an assumed (eleven-point) scale was used. For Form 4, the "distance" specified for the criterion pair was 100 units. TAB LE 5 Column Correlations for Split-Half Comparisons for Bight Dimensionsa Split—half Split-half Split-half Split-half Column FORM 100 FORM 200 FORM 300 FORM 400 (n=32; 32) (n=34; 33) (n=33; 33) (n=33; 33) .975 .982 .996 .971 12.82 10.77 5.31 13.87 .975 .248 .964 .953 12.85 75.59 15.34 17.71 .908 .627 .978 .959 24.84 51.19 11.95 16.39 .756 .819 .879 .802 40.87 35.02 28.48 36.70 .857 .678 .947 .742 31.00 47.30 18.78 42.08 .814 .230 .826 .904 35.55 76.72 34.27 25.24 .871 .305 .883 .742 29.44 72.23 27.96 42.12 .720 .603 .535 .934 43.97 52.90 57.65 20.86 a The first number is the correlation; the second is the angle. 129 stability of the eighth dimension drops (r = .720), but not sufficiently to exclude it from further analysis. Row correlations for all of the split-half comparisons are presented in Table 6. For the split-half comparison of Form 1 data, only situation six ("persuade your parents to let you stay out late") correlates poorly across the two configurations (r = .326). All other situations correlate .823 or better. Thus, for the split-half of Form 1 data, there appear to be eight fairly stable dimensions and one situation that is not stably located in the space. Note that while some situations (namely, situation seven) appeared to "move" between the two configurations, this movement is an expansion of the vector representing the situation further out into the space (or contracting towards the centroid). I say this because the situation moved, yet the vectors representing the locations of the situa- tion in the two configurations correlate highly (situation seven, for example, correlates .823 across the two configurations). Figure 3 presents a three dimensional representation of the loca- tions of the twelve situations in a split-half of Form 2 data. Evidently there are some serious problems in fitting the second split-half config- uration to that of the first. The mean distance between situations is 3.283 (see second column of Table 4), and the situations that exhibit change in this comparison include the following: situation eight ("persuade a sister/brother to go to a four-year University") moved 6.706 units; situation nine ("persuade a boy/girlfriend to study more") moved 5.254 units; situation one ("persuade your boss to let you leave work early") moved 4.283 units; situation six ("persuade your boy/girlfriend not to date another person") moved 4.425 units; and situation three ("persuade a professor that an answer on a test is incorrect") moved 130 TABLE 6 Row Correlations for Split-Half Comparisonsa’b Split-half Split-half Split-half Split-half Situations FORM 1 FORM 2 FORM 3 FORM 4 (n=32; 32) (n=34; 33) (n=33; 33) (n=33; 33) 1 .954 .598 .934 .915 17.43 53.35 20.84 23.73 2 .881 .835 .929 .917 28.23 33.40 21.68 23.47 3 .953 .642 .861 .945 17.61 50.08 30.50 19.15 4 .924 .874 .854 .924 22.94 29.03 31.31 22.51 5 .917 .905 .870 .901 23.43 25.19 29.56 25.75 6 .326 .609 .952 .850 70.95 52.47 17.67 31.78 7 .823 .741 .987 .894 34.56 42.16 9.38 26.55 8 .936 .166 .768 .937 32.98 80.45 39.86 20.38 9 .839 .293 .950 .901 32.98 72.97 18.12 25.69 10 .889 .930 .979 .882 27.30 21.59 11.76 28.06 11 .953 .655 .842 .963 17.60 49.07 32.64 15.62 12 .856 .778 .959 .918 31.12 38.94 16.41 23.38 a The first number is the correlation; the second is the angle. b All real dimensions were used in calculating correlations and angles. 131 FIGURE 3 Three-Dimensional Representation of 12 Situations for Split-Half Comparison of Form 200 Data X-Y PLflNE _____, ‘1 Y—l PLRIE X-l PLEUE 132 4.283 units. Column correlations for this comparison indicate that only dimensions one and four reach acceptable levels of stability (see second column of Table 5). Evidently, the instability of the above five situa- tions is sufficient to alter the factor structure of the Form 2 data. Row correlations indicate that six situations do not correlate well across the split—half comparison configurations (see Table 6): (1) sit- uation eight correlated only .166; (2) situation nine correlated .293; (3) situation one correlated .598; (4) situation six correlated .609; (5) situation three correlated .642; and (6) situation eleven ("persuade a group of peOple to participate in an experiment for your group pro- ject") correlated .655. Six of the twelve situations in the Form 2 data are not stable. Thus, either there exists substantive unreliability in the distance estimates, or this particular random split-half split the sample into two different points of view. Sex differences may play a role in under- standing the differences observed in this comparison and I shall discuss sex differences in a following subsection of this paper. Alternatively, various "points of view" may account for this apparent lack of stability. However, at this point in time, I believe that it is permissable to present the results of a second random split- half of the Form 2 data. Row correlations for this new analysis are as follows: situation one, r = .800; situation two, r = .875; situation three, r = .783; situation four, r = .890; situation five, r = .832; situation six, r = .569; situation seven, r = .920; situation eight, r = .487; situation nine, r = .486; situation ten, r = .899; situation eleven, r = .744; and situation twelve, r = .837. Thus, situations which are unstable include situations six, eight and nine. 133 Column correlations for this second split-half comparison indicate a substantial increase in the stability of the dimensional structure: dimension one, r = .973; dimension two, r = .739; dimension three, r = .794; dimension four, r = .943; dimension five, r = .820; dimension six, r = .789; dimension seven, r = .526; and dimension eight, r = .668. Thus, it would appear that there are probably two (or more) very strong points of view. Figure 4 presents the three dimensional representation of the twelve situations for the split-half of Form 3 data. There appears to be an extremely good fit between the two configurations. The mean distance between each situation and its counterpart is 1.525 units (see third column of Table 4). The three situations that exhibit the most change are as follows: situation four ("persuade a professor that a test item was ambiguous") moved a distance of 2.293 units; situation three ("persuade a boy/girlfriend to confide in you more") moved 2.264 units; and situation five ("persuade a professor that his grading scale is too hard") moved 2.018 units. Column correlations indicate that the first seven dimensions are stable (see third column of Table 5). Dimension eight correlates .535. Row correlations indicate that all of the situations are stable in this comparison (see third column of Table 6). If any situation is unstable, it is situation eight ("persuade an old boy/girlfriend that the relation- ship is worth saving"), which correlates .768. Figure 5 presents the three dimensional representation of the loca- tions of the twelve situations in the split-half of Form 4 data. While it appears that several of the situations exhibit movement, it also appears that this movement is in the form of expansions (or contractions) 134 FIGURE 4 Three-Dimensional Representation of 12 Situations for Split-Half Comparison of Form 300 Data XHYIWJHE «(.1 ’1‘ m m Yellflflflt 135 FIGURE 5 Three-Dimensional Representation of 12 Situations for Split-Half Comparison of Form 400 Data X-Y RLRNE / x-z rum: 0 Y-ZJNJNE 136 along the vector which identifies the situation's location between the two configurations. Thus, the situations could exhibit some change in their locations in the two configurations, but correlate highly. The mean distance between situations in this comparison is 23.501 units (see the fourth column of Table 4). The situations that exhibit movement are as follows: (1) situation two ("persuade your roommate to keep quiet when you are trying to sleep") moved 33.087 units; (2) situa- tion six ("persuade a person you know in class to help you with your homework") moved 29.127 units; and (3) situation nine ("persuade your father into getting a checkup") moved 27.697 units. Column correlations for this comparison indicate that the first four dimensions are stable (see column four of Table 5). Stability of the dimensional structure drops for the fifth dimension (r = .742) and for the seventh dimension (r = .742), but dimensions six and eight are‘ extremely stable (.802-.971). Row correlations indicate that all of the situations are stable (see column four of Table 6). The smallest row correlation is for situation six ("persuade a person you know in class to help you with your homework"), which correlates .850. In general, the multidimensional configurations generated by randomly splitting the data for each of the four groups yield extremely stable dimensions. This was true for both the assumed interval dissimi- larity data (Forms 1-3) and for the assumed ratio dissimilarity data (Form 4). The analysis of the first random split of Form 2 data was both very disappointing as well as informative. The fact that a second random split improved the stability of the dimensional structure obviously indi- cates that the stability of any of the comparisons is dependent upon choosing the sample in such a way as to obtain either the right point of 137 view or some mixture of various points of view. I shall return to the question of points of view in Section 3.7 of the paper. For the present, I will discuss the similarity of male-female configurations. 3.2.1 Reliability of the paired comparison data: assessing_ reliability between male-female configurations. In order to assess sex differences in the perceptions of situations, data for all male respondents in each group was factored, data for all female respondents in each group was factored and the female configura- tion was rotated (least-squares Procrustean; see Cliff, 1966; Schonemann, 1966; and Woelfel. et al., 1975) into congruence with the male configura- tion. The coOrdinates for these solutions will not be provided. Pre- sently, I wish to discuss the stability of the dimensions and situations across the various male-female comparisons. Figure 6 presents the three dimensional representations of twelve situations for the male-female comparison of Form 1 data. As illustrated in the figure, several situations exhibit some change in their locations between the two configurations. Table 7 presents the amount of distance moved by a situation across the male-female configurations. The mean distance moved by situations i Form 1 is 2.645 units (see column one of Table 7). The situations which exhibit the most change are as follows: situation six ("persuade your parents to let you stay out late") moved 3.431 units; situation ten ("persuade your father to take more time off from work and find something relaxing to do") moved 3.44 units; situation eight ("persuade a group of people to order a pizza because you feel like eating one") moved 3.295 units; and situation one ("persuade your roommates to leave phone 138 FIGUP“ 6 Three-Dimensional Representation of 12 Situations for Male-Female Comparison of Form 100 Data XHYIWJNE \K- "‘"‘ \\ CL) .\ /\. X#ZINJM£ €//\ .\ YellflflflE 139 TABLE 7 O I O O O a Distances Between Situations in Male-Female Comparisons Male-Female Male-Female Male-Female Male-Female FORM 1 FORM 2 FORM 3 FORM 4 Situation (male n=29; (male n=21; (male n=l7; (male n=23; female n=35) female n=46) female n=49) female n=43) trimmed 1 3.032 3.265 1.128 30.883 2 1.397 1.329 2.852 19.992 3 1.721 1.870 2.547 33.135 4 1.912 1.422 1.715 25.455 5 2.892 1.617 1.863 19.828 6 3.431 4.937 2.137 28.039 7 2.789 1.608 2.105 25.481 8 3.295 5.157 2.155 29.832 9 2.677 2.931 2.617 33.074 10 3.344 2.917 1.427 25.507 11 2.286 2.418 1.112 24.510 12 2.940 2.680 2.721 33.706 Mean distance 2.645 2.458 2.036 27.454 a For Forms 1, 2 and 3, an assumed (eleven-point) scale was used. For Form 4, the "distance" specified fer the criterion pair was 100 units. 140 messages for you while you are out") moved 3.032 units. The factor structure stability appears to be somewhat lower for the male-female comparison than for the split-half comparison. For the male- female comparison, the first three dimensions are fairly stable, but stability is lower for dimensions four to eight (see column one of Table 8). In contrast with the split-half comparison (see Table 5), the corre- lation for the sixth dimension drops appreciably (r = .814 for the split- half comparison; r = .508 for the male-female comparison). Thus, there are some changes in dimensional structure in the male-female comparison, but this change is extremely small -the first three dimensions are very similar and they account for approximately 60-65 per cent of the variance. If the dimensional structure is somewhat different between males and females, the differences exist in the smaller dimensions in which case it would be hard to differentiate true differences from unreliability. Row correlations (see first column of Table 9) indicate that the following situations are located differently across the male-female con- figurations: situation six ("persuade your parents to let you stay out late") correlated .556; situation nine ("persuade a friend to take a class with you") correlated .730; and situation eight ("persuade a group of people to order a pizza because you feel like eating one") correlated .747. Several other situations correlated less well than in the split- half comparison (see first column of Table 6). However, it would appear that sex differences are not very strong. In general, the dimensional structure stability decreases from correlations which range from .720 to .975 (column one of Table 5) to correlations which range from .508 to .924; and row correlations indicate that only a few situations are 141 TABLE 8 Column Correlations fer Male-Female Comparisons for Eight Dimensionsa FORM 1 FORM 2 FORM 3 FORM 4 Column (male n=29; (male n=2l; (male n=l7; (male n=23; female n=35), female n=46) female n=49) female n=43) .893 26.69 .924 22.45 .904 25.24 .733 42.87 .720 43.97 0 508 59.47 .757 40.75 .698 45.69 .982 10.77 .754 41.04 .459 62.67 .834 33.46 .869 29.70 .821 34.80 .317 71.53 .851 31.64 .991 7.83 .980 11.36 .913 24.01 .846 32.17 .829 34.03 0 50” 59.76 .701 45.51 .752 41.25 .909 24.56 .890 27.10 .847 32.15 .780 38.69 .822 34.65 .737 42.53 .332 70.59 .069 86.05 a The first number is the correlation; the second number is the angle. 142 TABLE 9 Row Correlations for Male-Female Comparisonsa’ FORM 1 FORM 2 FORM 3 FORM 4 Situation (male n=29; (male n=2l; (male n=l7; (male n=23; female n=35) female n=46) female n=49) female n=43) 1 .824 .815 .972 .756 34.49 35.38 13.64 40.92 2 .951 .881 .831 .854 18.04 28.26 33.78 31.39 3 .934 .895 .808 .682 20.95 26.49 36.10 47.01 4 .862 .840 .865 .844 30.44 32.87 30.08 32.42 5 .774 .740 .827 .928 39.27 43.09 34.15 21.82 6 .556 .488 .849 .791 56.19 60.81 31.85 37.75 7 .811 .915 .889 .753 35.84 23.84 27.26 41.16 8 .747 .501 .883 .724 41.71 59.91 27.97 43.60 9 .730 .699 .851 .814 43.14 45.67 31.72 35.48 10 .793 .720 .949 .776 37.49 43.95 18.35 39.06 11 .867 .775 .955 .718 29.91 39.20 17.33 44.09 12 .787 .865 .872 .761 38.11 30.11 29.11 40.43 a The first number is the correlation; the second number is the angle. All real dimensions were used in calculating these correlations and angles. 143 located in the female configuration at small angles from the situation's locations in the male configurations. Figure 7 presents the three dimensional configuration of twelve situations for the male-female configuration of Form 2 data. Again, several situations exhibit some changes, most notably situations eight and six. The mean distance moved between situations is 2.458 (see second column of Table 7). Situations that exhibit the most change are as fol- lows: situation eight ("persuade your brother/sister to go to a four- year University") moved 5.157 units; situation six ("persuade your boy/ girlfriend not to date another person") moved 4.937 units; and situation one ("persuade your boss to let you leave work early") moved 3.256 units. Column correlations for this comparison indicate very fragile dimen- sional stability across the male-female configurations (see second column of Table 8). The first dimension correlated very highly (.982), but stability is somewhat low for dimension two (.754) and (particularly) dimension three (.459). Dimensional stability is moderately high for dimensions 4, 5, 6 and 8, but extremely low for dimension 7 (.317). Row correlations for this comparison indicate that the following situations are located differently in the two configurations (see second column of Table 9): situation six correlated .488; situation eight cor— related .501; situation nine ("persuade a boy/girlfriend to study more") correlated .699; situation ten ("persuade a person to ask you out") cor- related .720; and situation five ("persuade a boy/girlfriend to go to a party where there is mostly your friends") correlated .730. All other situations correlate very well (.775-.915). Thus, there are five situa- tions which do not optimally correlate across the sexes. 144 FIGURE 7 Three-Dimensional Representation of 12 Situations for Male-Female Comparison of Form 200 Data X€YIWJNE [ fi/T\\ X4IIWSNE ,4 S?” \\ 145 It would appear that sex differences are much more prevalent in the male-female comparison of Form 2 data. These differences affect some of the larger dimensions (dimensions two and three, which account for 20% and 15% of the variance, respectively). NOte that with the exception of situation eight ("persuade a brother/sister to go to a four-year Univer- sity") all of the situations that differ between males and females are situations that deal with boy/girlfriend relationships. Further, such sex differences cannot explain the instability obtained in the first random split-half comparison of the Form 2 data. In that comparison, situations eight ("persuade a sister/brother to go to a four-year University"), nine ("persuade a boy/girlfriend to study more"), one ("persuade your boss to let you leave work early"), three ("persuade a professor that an answer on a test is incorrect"), five ("persuade your boy/girlfriend to go to a party where there is mostly your friends"), and eleven ("persuade a group of people to participate in an experiment for your group project") were found to be instable. Of these, several which deal with boy/girlfriend relationships may_be attributable to sex differ- ences. However, the majority of situations found instable in the first random split-half of the Form 2 data are not the same situations observed to be instable in the sex difference comparisons. In general, the situations found to be instable in the first random split-half dealt with one (or more) of the following: (1) persuading an individual who has higher relative status (situations one and three); (2) a brother/sister relationship (situation eight); (3) a group (situa- tion eleven); or (4) a boy/girlfriend relationship (situations five and nine). Thus, it would seem reasonable to suspect a different perspec- tive, separate from the sex differences, may be operative. Such a point 146 of view may place more emphasis on the level of status of the persuadee (as in situations one and three), or perhaps a subset of individuals locate situations they have never been in at a different location in the multidimensional configuration (given that some set of people may have neither sister nor brother to persuade, nor have they ever persuaded a group of people to participate in an experiment for their group project). Thus, an analysis of individual differences (points of view) will be a paramount aspect of the present inquiry (see Section 3.7). That is, it seems clear that in order to construct a taxonomy of situations of this type (and perhaps any type of situations), points of view must be taken into account. Figure 8 represents the three dimensional representation of the twelve situations in the male-female configuration of Form 3 data. As illustrated in the figure, there appears to be little difference between situations across the two configurations. The mean distance between situations in this comparison is 2.036 units (see third column of Table 7). The three situations which exhibit the most change include: situation two ("persuade your roommates to keep quiet when you are trying to sleep") moved 2.852 units; situation nine ("persuade your father into getting a checkup") moved 2.617 units; and situation three ("Persuade your boy/girlfriend to confide in you more") moved 2.547 units. Column correlations for this comparison indicate that the first three dimensions are extremely stable (.991-.913), and that dimensions four and five correlate moderately high. The sixth dimension did not correlate highly (r = .504), but this dimension is a fairly small dimen- sion -accounting for five per cent of the variance. Finally, dimensions 147 FIGURE 8 Three-Dimensional Representation of 12 Situations for Male-Female Comparison of Form 300 Data XJYlWflNE + x-z mm /a\ f’ Z \ y YBIIWJUE 148 seven and eight correlate moderately well (r = .701 and .7 2, respectively). Row correlations for this comparison indicate that all situations correlate highly across the male-female configurations (see column three of Table 9). The lowest correlation is .808 (situation three, "persuade your boy/girlfriend to confide in you more"). Figure 9 presents the three dimensional representation of the loca- tion of twelve situations in the male-female comparison of Form 4 data. It would appear that a number of situations are not stably located across the male-female configurations. The mean distance between situations in this comparison is 27.454 units (see column four of Table 7). Situations which exhibit the most change are: situation twelve ("persuade your mother to go back to college and get a degree") moved 33.706 units; situation nine ("persuade your father into getting a checkup") moved 33.074 units; situation three ("persuade your boy/girlfriend to confide in you more") moved 33.135 units; situation one ("persuade a boy/girl- friend that there is nothing wrong in occasionally dating another per- son") moved 30.883 units; and situation eight ("persuade an old boy/girl- friend that the relationship is worth saving") moved 29.832 units. Column correlations for this comparison indicate the first three dimensions are highly correlated (.909-.847) and that dimensions four through six correlate moderately high (.822-.737). Dimensions seven and eight do not correlate well, but they account for only a small portion of the variance. Row correlations for this comparison indicate that the following situations do not correlate highly across the male-female configurations (see column four of Table 8): situation three (r = .682); situation 149 FIGURE 9 Three—Dimensional Representation of 12 Situations for Male-Female Comparison of Form 400 Data X‘YIWRNE _,k *‘1. \ Y-llnjfll XJZIWJNE 150 eleven ("persuade a professor to increase your grade", r = .718); situation eight (r = .724); situation seven ("persuade your landlord to fix the plumbing in your apartment", r = .753); situation one (r = .756); situation twelve (r = .761); and situation ten ("persuade a police offi- cer not to give you a ticket", r = .776). All other situations correlate .791 or better. By way of concluding these assessments of stability, several obser- vations should be noted. First, the level of stability obtained in the random split-half comparisons is apparently not affected as much by the type of scale used to collect the data (assumed interval or assumed ratio) as it is by the situations included in the questionnaire. This observation is supported by the fact that the row and column correlations for the split-half of the Form 3 data and the Form 4 data are comparably high, while the row and column correlations for neither of the two split-halves of the Form 2 data reached the levels of stability obtained for the other three comparisons. As noted above, the lack of stability of the dimensional structure of the Form 2 data can be attributed to the fact that more of the situations included in that set of situations dealt with boy/girlfriend relationships and status relationships (and/or unfamiliar situations). It may be the case that such situations enhanced the emergence of individual differences. Several comments are in order concerning the comparison of the male and female configurations. A scree line test of the GALILEO solutions indicates four dimensional solutions (see Section 3.3 for details con— cerning the scree line tests for the aggregated solutions) and, as noted earlier, the first four dimensions for both the male and female config- urations are very similar (except for Form 2 data). However, a number 151 of situations were found to correlate poorly across the male-female com- parisons. These situations deal with boy/girlfriend relationships (not to date another person, ask you out, to study more, confide in you more, go to a party where there is mostly your friends), with parents (stay out late, mother to go back to college, brother/sister to go to a four-year University) and with professionals (professor to increase a grade, land- lord to repair plumbing, police officer not to give a ticket). While it may be save to posit a sex difference in the perceptions of boy/girlfriend relationship situations, it ought to be noted that a sex difference may indeed mask a different true relationship. That is, some other variable may correlate with sex and is the correct explanation of the different perceptions. Given the situations dealing with parents and professionals for which the apparent sex differences were obtained, it may be the case that a differentiation of individuals on a measure such as dependence-independence (or autonomy) may account for the apparent sex differences. Specifically, if more of the sampled females are dependent upon others than the sampled males, than it is likely that this depen- dence may be the explanatory variable which accounts for these differ- ences. If this is true, then the "sex differences" obtained may not be replicated in other research using different samples, while a measure such as dependence or autonomy may prove useful. Nonetheless, it would appear that for most of the large dimensions substantive stability in the dimensional structure has been obtained. This is true for both the random split-halves as well as for the male- female comparisons. The exception to this conclusion is the poor dimension to dimension, or column correlations, for the second and third dimensions of the Form 2 data. However, a second random split-half 152 comparison improved congruence and the situations found to be instable in the random split-half and the male-female comparison were different. These observations indirectly support the claim that diverse points of view are operative. The question of the diversity of such points of view will be discussed in Section 3.7. 3.2.2 Comparison of Form 3 and Form 4 Solutions. The single most feature which distinguishes Form 3 and Form 4 data is the difference in the scale provided the respondents. In order to compare the solutions obtained by utilizing an assumed interval scale and an assumed ratio scale, the means matrices for the two sets of data were factored and the coordinates for the Form 3 solution were rotated into congruence with that of the Form 3 solution. The dimension to dimension correlations (column correlations) for the first eight dimensions are as follows: .855, .945, .707, .832, .749, .736, .907 and .560. Thus, it would appear that the type of scale provided the respondents had little impact on the resultant solutions; at least for the first four dimen- sions. However, as will be noted below, the correspondence between dimensional arrays of situations and the twenty-three attributes are dif- ferent for the two solutions; although the interpretation of three of the four dimensions is similar. The reason for this difference may be due to the fact that one situation was located differently across the two configurations: situa- tion eleven ("persuade a professor to increase your grade") correlated only .315. All other row correlations ranged from .734 to .885. Advo- cates of the assumed ratio (unbounded) scale may argue that the assumed interval scale may encumber ceiling effects or not allow sufficient 153 gradations of the scale to allow precise measurement of dissimilarity. Advocates of the assumed interval scale may argue that the above situa- tion is simply measured unreliably by the assumed ratio scale. However, since the purpose of the present study is to interpret a dimensional structure of situations, it should be sufficient to note that the two four-dimensional solutions correspond fairly well. The results of the various analyses concerning stability presented in this section of the paper indicate that a large number of dimensions are stable, for either the random split—half comparisons or for the male- female comparisons. In the comparisons made in these analyses, all of the dimensions obtained in the GALILEO solutions were used. However, it is doubtful whether each of the eight dimensions would be representative of equally important distinctions in the perceptions of situations. In the following section, I shall utilize procedures typically employed in multidimensional scaling to parsimoniously limit the number of dimensions that will be employed in the subsequent analyses. 3.3 The number of dimensions problem resolved One of the basic purposes of multidimensional scaling is to obtain a dimensional configuration of objects in as small a number of dimensions as possible. For each of the three multidimensional scaling procedures employed, a particular criterion is used to decide what number of dimen- sions should be retained for interpretation. Traditionally, a value of stress is calculated in order to measure the "goodness of fit" for each dimensional solution (Kruskal, 1964; Spence and Ogilvie, 1973, Woelfel and Danes, 1977). This criterion is used in conjunction with the §X§I_ (TORSCA) program. 154 When employing the INDSCAL model, a plot of percent of variance accounted for by each dimensional solution is made, and the researcher chooses whether or not there is any advantage of selecting a configura- tion of higher dimensionality. That is, the researcher asks the ques- tion, "how much gain is there in the percent of variance in moving from a one dimensional solution to a dimensional solution of higher dimension- ality?" Usually, no criterion is applied in selecting a minimum number of dimensions when using the GALILEO program, since all dimensions are used in subsequent analysis. For the purposes of the present analysis, a scree line test will be employed (plotting percent of variance accounted for by each dimension). First, I shall present information concerning stress values. Data for each questionnaire were analyzed using the §Z§I_multidimensional scaling program (with TORSCA option) for dimensional solutions one through five. Table 10 presents the stress values for each dimensional representation. Usually, stress values of twenty percent and higher are considered "poor"; stress values of ten percent are "fair"; five percent are "good" two and one-half percent are "excellent"; and zero as "perfect". For a one dimensional solution, the stress values are extremely high (see first row of Table 10). For the two dimensional solutions, the stress values improve markedly, but stress values of 9.6 percent and 14.4 per- cent are still poor. The stress values for the three dimensional solu- tions indicate that the fit again improves, approaching what could be considered a good fit. Stress values for the four dimensional solutions for all sets of data are very good. For the five dimensional solutions, 155 TABLE 10 Stress Values for Dimensions One to Five For Each of the Four Questionnaire Types (KYST Analysis) Questionnaire type: Dimensions 1. 2 3 4 1 34.4 96 30.8 96 21.6 96 33.5 96 2 14.4 96 11.5 96 9.6 96 12.0 96 3 6.7 % 5.2 96 3.1 96 6.3 ‘1”: 4 4.6 96 3.3 96 1.5 96 2.7 96 5 2.9 96 1.9 96 0.9 96 1.3 96 156 several stress values approach the "perfect" category (Forms 2, 3 and 4). Figure 10 plots the stress values by dimensional solutions for all four sets of data. Note that there is a substantial decrease in stress from the one dimensional solutions to the two dimensional solutions; some improvement from the two dimensional solutions and the three dimensional solutions; and very little improvement from the three dimensional solu- tions to the fourth. Overall, there is some advantage in moving from the three dimensional solutions to the four dimensional solutions. For all practical purposes, however, there is not enough of an improvement in stress values between the four and five dimensional solutions to accept the five dimensional solutions. Therefore, I shall accept the four dimen— sional solutions for further analysis -there is a general improvement from the three dimensional solutions to the four dimensional solutions, and all of the four dimensional solutions have "good" stress values. Table 11 presents correlations between the input data and scores computed by the INDSCAL program for dimensional solutions two through six. These correlations are squared to provide the assessment of the variance accounted for by each dimensional solution. A one dimensional solution for each of the questionnaire forms was not obtained because it seemed obvious (from the REST analysis) that a one dimensional represen- tation would be poor. Further, there exists a limit internal to the INDSCAL program such that the number of respondents multiplied by the maximum number of dimensions cannot exceed 3. Thus, in order to assess the dimensional solutions of a dimensionality higher than four, a random split-half of the data was used. As discussed below, the solutions account for substantially less variance than expected. For this reason, solutions for six dimensions were obtained. 157 FIGURE 10 Plot of Stress Values for Dimensions One to Five For Each of the Questionnaire Forms (KYST Analysis) stress values 209611- 10%" 30951 + l I l l I T 1 2 3 ;:5L_ .4 Dimensional Solutions X = Form 1 data + = Form 2 data * = Form 3. data (I) ll Form 4 data TABLE 11 Correlations Between Input Data and Predicted Scores and Percent of Variance Accounted fer by Two through Six Dimensional Solutions (INDSCAL Analysis) . . Questionnaire: DimenSional Solution FORM 1 FORM 2 FORM 3‘ FORM 4 2 Corr. .501 .608 .621 .584 Percent 25.09 36.94 38.59 34.08 3 Corr. .562 .673 .686 .648 Percent 31.55 45.27 47.12 42.03 4 Corr. .613 .707 .722 .709 Percent 37.60 49.96 52.18 49.72 5 (I) Corr. .660 .737 .756 .774 Percent 43.54 54.35 57.14 59.88 5 (II) Corr. .684 .736 .740 .714 Percent 46.54 54.19 54.80 51.04 6 (I) Corr. .672 .772 .788 .775 Percent 49.69 59.54 62.10 60.12 6 (II) Corr. .688 .762 .769 .768 Percent 47.30 58.09 59.10 59.07 159 For the Form 1 data, a two dimensional solution accounts for only 25.09 percent of the variance (see column one of Table 11). Each subse- quent dimensional solution yields a moderately slow improvement in the amount of variance accounted for the by solution. Yet, even a six dimen— sional solution accounts for less than half of the variance (49.69 and 47.30 percent of variance, for the respective split-halves). For the Form 2 data, a two dimensional solution accounts for 36.94 percent of the variance (see second column of Table 11). There is a marked improvement in moving from a two dimensional solution to the three dimensional solution, which accounts for 45.27 percent of the variance. Some improvement can be noted in moving from the three dimensional solu- tion to the four dimensional solution (49.96 percent of the variance) and from the four dimensional solution to the five dimensional solution (54 percent). Roughly speaking, the degree of improvement in moving from the three dimensional solution to the four dimensional solution, and each subsequent step, accounts for the same, small degree of improvement - about 4 to 5 percent of the variance. For this set of data, the degree of improvement seems to be fairly trivial after the three dimensional solutions. For the Form 3 data, the two dimensional solution accounts for 38.59 percent of the variance (see the third column of Table 11). Again, there is a marked improvement in moving to a three dimensional solution (47.12 percent of the variance). A small gain can be observed in moving to a four dimensional solution (52.18 percent of the variance) and to a six dimensional solution (roughly, 61 percent of the variance). Again, the primary gain in improvement occurs between the two and three dimensional solutions, though there is some improvement between the three dimensional 160 and four dimensional solutions. For the Form 4 data, a two dimensional solution accounts for 34.08 percent of the variance (see column four of Table 11). For this set of data, there is roughly the same amount of improvement in moving from a two dimensional solution to a three dimensional solution (42.03 percent of the variance) and from a three dimensional solution to a four dimen- sional solution (49.72 percent of variance). Curiously, the degree of improvement in moving from a four dimensional solution to a five dimen- sional solution depends upon the random split-half that is chosen for comparison; 59.88 percent or 51.04 percent of the variance. Taking a rough average of these two five-dimensional solutions would indicate that the gain obtained from moving from the four dimensional solution to the five dimensional solution is less than the gain in moving from the three dimensional solution to the four dimensional solution. Figure 11 presents a plot of the percent of variance accounted for by each of the dimensional solutions for each of the data sets. It does not appear that the percent of variance tapers off as one would expect. While the linear trend in improvement is gradual, the only subtle break in the degree of improvement appears between the four dimensional solu- tions and the five dimensional solutions (except for the Form 1 data). This conclusion, of course, is arbitrary. Others may note that the pri- mary improvement in moving from one dimensional solution to another occurs between the two and three dimensional solutions and that little is gained by accepting solutions of higher dimensionality. I shall not accept a three dimensional solution for three reasons. First, the assessment of the §z§T_solutions suggested the retainment of four dimensional solutions (as well as the GALILEO assessment discussed 161 FIGURE 11 Plot of Percent of Variance Accounted for by (INDSCAL) Dimensional Solutions Two through Six Percent of Variance 100% 4 90% 4 80% 4 70% ~ 60% 50% 1. ./ -, 4096 w J / 30% n ‘ "”’,,.,~’x 20% v 10% . l J l l l L I I l I F 0 l 2 3 4 5 6 Dimensions X = Form 1 + = Form 2 * = Form 3 $ = Form 4 162 below). Secondly, if one accepted the three dimensional solutions, one would be accepting solutions which account for less than one-half of the variance. Apparently, this criteria for selecting the number of usable dimensions is not as functional for these sets of data as it has been for other researchers. For example, Wish, Deutsch and Kaplan (1976) obtained dimensional solutions in six, five, four and three dimensions. These solutions accounted for 79, 78, 75 and 67 percent of the variance, respectively. These researchers concluded that the degree of improvement tapered off at four dimensions. In the present study, however, the degree of improvement does not taper off drastically, nor do the INDSCAL solu- tions account for 75 percent of the variance; at most, the four dimen- sional solution for the Form 3 data accounts for 52.18 percent of the variance. A third reason for accepting the four dimensional solutions over the three dimensional solutions deals with how the original data for each respondent is adequately represented in the INDSCAL solutions. An option internal to the INDSCAL model was used in order to obtain the correlation between the individual respondent's data to that of the computed scores (configuration of situations). For the three dimensional solution of Form 1 data, 44 of the 64 respondents had correlations less than .60. The average correlation between respondents' data and the INDSCAL solu- tion was .551. For the three dimensional solution of Form 2 data, 18 of the 67 respondents had correlations less than .60. The average correla- tion between the respondents' data and the INDSCAL solution was .665. For the Form 3 three dimensional solution, 13 of the 66 respondents had correlations less than .60, and the average correlation was .677. For the Form 4 three dimensional solution, 22 of the 66 respondents had 163 correlations less than .60, and the average correlation was .635. For the Form 1 four dimensional solution, 26 of the respondents had correlations below .60, and the average correlation was .606. For the Form 2 four dimensional solution, nine of the 67 respondents had correla- tions below .60, and the average correlation was .701. For the Form 3 four dimensional solution, 10 of the 66 respondents had correlations below .60, and the average correlation was .715. Finally, for the Form 4 four dimensional solution, 11 of the 66 respondents had correlations less than .60, and the average correlation was .670. Thus, by accepting the four dimensional solutions over the three dimensional solutions, a substantial improvement can be obtained in the degree to which the indi- vidual's responses are represented in the INDSCAL solution. For three of the sets of data, the number of respondents whose correlations were less than .60 decreased by about one-half when moving to the four dimensional solution (data sets for Forms 1, 2 and 4. Admittedly, the correlation level of .60 in the above comparison is arbitrary. However, a similar pattern could be revealed for almost any correlation cut-off level. The above comparison is important because I shall attempt to cluster individuals on the basis of the dimensional salience weights generated by the INDSCAL model. Obviously, if the INDSCAL solution accepted for analysis subsumes only 36 to 40 percent of the variance for a large number of the respondents, the generality of the obtained clusters would be severely weakened. Thus, the four dimen- sional INDSCAL solutions will be utilized in the subsequent analysis. Table 12 presents the percent of variance accounted for by each of the eight (GALILEO) dimensions for all four sets of data. The plot of these values is presented in Figure 12. For three of the sets of data 164 TABLE 12 Percent of Variance Accounted for By Each of Eight Dimensions (GALILEO Analysis) Questionnaire: Dime"Si°n FORM 1 FORM 2 FORM 3 FORM 4 1 25.224 38.549 38.920 38.828 2 21.360 19.268 20.438 23.141 3 14.646 15.053 16.483 15.180 4 10.576 8.470 6.896 8.343 5 8.675 6.831 6.267 5.553 6 6.705 5.118 4.151 4.788 7 5.272 4.485 3.307 2.527 8 4.012 2.017 1.482 1.640 50% 1 40% 4 3096 . 20% .. 1096 4 165 FIGURE 12 Plot of Percent of Variance Accounted for by Bach Dimension (GALILEO Analysis) “‘=n=-._== . I: 1 l 1 1 1 1‘ L 7 l I l l T 0 l 2 3 4 5 6 7 8 X=Forml +=Form2 *=Form3' Form 4 166 (Forms 2, 3 and 4) the line representing a linear, downward trend in the amount of variance accounted for by each dimension changes slope between the fourth and fifth dimensions. For Form 1 data, however, this change is very subtle. Thus, a conclusion similar to that made when the §X§T_ analysis was presented is apparently supported: a four dimensional solu- tion will be used for interpretation. Before discussing the psychological interpretation of the solutions, it would be of interest to make some comparisons among the three solu- tions provided by the three multidimensional scaling programs. First, it is puzzling that the four dimensional solutions generated by the INDSCAL model account for only one-half of the variance in each of the four solutions (see Table 11). For the GALILEO solutions, considerably higher levels of variance were accounted for by four dimensions (71.906, 81.340, 82.737 and 85.492 percent of variance, respectively). This difference is due to the divergent ways in which the percent of variances are computed. For the INDSCAL model, a correlation between the input data (averaged across individuals) and the computed scores is squared. For the GALILEO model, a means matrix is calculated and fac- tored (obtaining the coordinate loadings, eigenvalues and trace of the matrix). (The same procedure is employed in using the KYST (TORSCA) model, except that only the coordinate loadings and stress values are provided.) Thus, the points of view (individual differences) and unreliability influence the calculation of means in the §Z§T_and GALILEO models. In the three-way multidimensional scaling model (INDSCAL), the operation of divergent points of view (as well as unreliability) strongly influence the quality (or generality, if you will) of the overall dimen- sional structure. For example, if there is one very strong point of view 167 (a point of view which included one-half of the respondents) and two other points of view, the INDSCAL model would generate a solution that would be more or less general for the main point of view, but not as representative of the other two points of view. A major premise of the INDSCAL model is that the solution is gener- ally representative of the solution which would be generated for each individual. However, the INDSCAL model will give a dimensional salience weight of zero for some individuals on some dimensions, and while an in- dividual may have a weight on each dimension, there may be other distinc- tions the individual perceives which are not representative in the over- all solution. (See Woelfel and Danes, 1977, for other comments concern- ing similarities and dissimilarities among the models.) Thus, if there are divergent points of view operative in these sets of data, the per- cent of variance accounted for by the INDSCAL model would be attenuated. _ The fact that so little of the variance is accounted for by dimensional solutions two through six adds indirect support that there are various points of view which are operative in these sets of data. A second note must also be made. While I do not intend to rotate various dimensional solutions to congruence in order to assess differ- ences across the three multidimensional scaling procedures, it would be of interest to briefly mention commonalities among the solutions. Since the KYST (TORSCA) procedure and the GALILEO procedure are two-way factor- ing models using assumed interval and assumed metric data (for respective data sets), one would believe that the four dimensional 5121 solutions would resemble the first four dimensions generated by the GALILEO pro- cedure. This is evident in the correlations between unrotated dimension— al solutions: for the Form 1 data, the correlations are —.99, .99, .99 168 and .98; for the Form 2 data, the correlations are .99, -.95, .97 and -.75; for the Form 3 data, -.93, .95, .97 and .97; and for the Form 4 data, -1.00, .99, -.93 and .93. A rotation of the §X§T_solutions to that of the first four dimensions of the GALILEO solutions would do little to the structure of the obtained solutions, aside from reversing the signs of the loadings of situations on the appropriate dimensions. Correlations between the coordinate loadings of the §Z§I_solutions and the INDSCAL solutions (unrotated) are as follows: for the Form 1 data, -.94, .88, -.81 and .83; for Form 2 data, .57, -.35, .53 and .21; for Form 3 data, -.77, .20, .40 and .54; and for Form 4 data, .68, .35, .03 and .81. The main point to this observation is obvious: if one were to rely solely on either the INDSCAL model or the KYST/GALILEO models, one would obtain a fairly different structure of situations. 3.4 Dimensional Structure: Results of the four dimensional (KYST) solutions Table 13 presents the coordinate loadings of the 12 situations included in Form 1 in the four dimensional representation generated by the KYST (TORSCA) solution. Table 14 presents the standardized regres- sion coefficients obtained by regressing the coordinate loadings of the situations for all four dimensions onto the mean ratings for each attri- bute. In relation to the first dimension, only a few attributes seem to be related to the manner in which situations are arrayed on this dimen- sion: "emotional-unemotional" (8 = -.679); "dominates-treat each other equally" (8 = -.535); "s/he has a major effect-does not have a major effect on my life" (8 = -.676); and "I have a major effect-do not have a major effect on his/her life" (8 = -.463). However, each of these attri- butes apparently lies across the first three dimensions. Thus, the first 169 TABLE 13 Coordinates of 12 Situations (Form 1) in Four Dimensions from KYST Analysis (Stress = 4.65%) Dimension Situation 1 2 _3_ 4- 1. Persuade your roommate(s) to leave phone messages for you while you are out. -l.094 -.l24 -.416 -.560 2. Persuade a professor to let you enroll in an already crowded class. .486 -.546 .629 -.069 3. Persuade a group of people to stop teasing a friend of yours. -.517 .268 .583 .630 4. Persuade your boy/girlfriend to meet your relatives. .107 .709 -.256 -.416 5. Persuade a friend not to swear so much. -.569 .599 .227 .319 6. Persuade your parents to let you stay out late. .613 -.274 -.128 -.102 7. Persuade your boy/girlfriend not to date another person. .482 .788 .481 -.446 8. Persuade a group of people to order a pizza because you feel like eating one. -.587 -.704 -.295 .375 9. Persuade a friend to take a class with you. -.509 .039 -.217 -.l22 10. Persuade your father to take more time off from work and find some- thing relaxing to do. .768 .578 -.621 .537 ll. Persuade a bouncer to let you into a bar even though you have forgotten your I.D. , .156 -.726 .536 -.181 12. Persuade your parents to give you money. .663 -.607 -.522 .036 170 TABLE 14 Standardized Regression Weights for Four (KYST) Dimensions Used to Predict Mean Ratings on Unidimensional Attributes (Form 1; n=64) Dimensions: Attributes 1 2 3 4 R2 friendly- a a unfriendly .168 -.023 .890 .039 .822 difficult-easy a b to resolve conflicts -.O74 .167 -.865 -.l48 .803 short term- b a a long term strategy .315 .831 -.l98 .238 .885 comfortable- b a a uncomfortable .316 -.049 .878 .034 .874 resistant-not resistant c a b to persuasion -.394 -.O83 -.713 -.296 .758 easy to persuade- c a b difficult to persuade .389 .149 .674 .388 .779 I benefit- a a I do not benefit -.089 .629 -.030 .674 .858 I have been-I have never been in this situation .117 .466 .460 -.110 .455 involves a favor- b C involves a right -.l61 .701 -.029 -.161 .544 do something for his/her benefit-not do something a a fbr his/her benefit .169 -.772 .222 -.454 .880 I have relatively low a b b status-high status -.268 .607 -.594 —.025 .794 a fermal-informal -.475b .238 -.750a .059 .848 one totally dominates— b b b b treat each other equally -.535 .566 -.422 -.099 .794 a p < .01 b p < .05 °p<.10 171 TABLE 14 (cont'd.) . Dimensions: Attributes 1 2 3 4 R2 superficial-intimate relationship -.110 .347 -.247 -.408 .360 no commitment to future b b b interaction-commitment .239 .614 -.502 -.158 .710 frank and open- b b a a reserved and cautious .448 -.440 .675 -.077 .856 impersonal-personal b b b relationship .219 .606 -.586 -.104 .769 emotional-unemotional a b a situation -.679 -.592a -.293 .011 .897 I know-I do not know b b a b how to behave .408 .497 .614 .154 .815 he/she will think less of a b me-will not think less -.O37 -.l21 -.820 -.167 .717 he/she has a major effect- does not have a major a b b a effect on my life -.676 -.435 .441 .058 .844 I have a major effect- do not have a major b b b b effect on his/her life -.463 -.519 .546 .080 .787 I must be assertive- a a I need not be assertive -.295 .086 -.710 -.469 .818 p < .01 p < .05 p < .10 172 dimension is not distinctly characterized by any of the attributes. As such, the array of situations does not relate to intimacy or formality. However, this dimensional array of situations roughly corre- sponds to what Insel and Moos (1974) have referred to as a relationship type of dimension. All situations with negative loadings (friend to take a class with you, stop teasing a friend, friend not to swear so much, order a pizza, have roommates leave phone messages) deal with social, casual situations entailing cohesiveness and affiliation. These are informal situations but the formal-informal set of attributes do not cor- respond to the dimensional array of situations simply because the situa- tions which load positive on the dimension are not particularly formal in nature. That is, persuading a father to take time off from work, parents to provide financial support, permission to stay out late, a professor to let you enroll in an already crowded class, etc., are only somewhat more formal ( as indicated by the Beta weight of -.475). Unfortunately, the way in which situations load on this dimension apparently does not coincide with a bipolar notion of formal—informal. Further, the situations which load positive can hardly be referred to as non-affiliation or non-cohesive situations. These situations can best be described as "subordinate relationship" situations in which the persuader is in a subordinate relationship with parents, professors, a bouncer and a boy/girlfriend who they wish to persuade not to date another person. Therefore, this dimension can be labeled a "subordinate relationship vs. social-affiliation relationships" dimension. In terms of a psychological interpretation of the second dimension, respondents perceived situations with positive loadings as those situa- tions where a long-term strategy is preferable ("short term-long term 173 strategy", 8 = .831). Further, situations which have positive loadings on this dimension are perceived as having more to do with the respon- dents' rights, as opposed to favors (B = .701) and entailing more to do with future interactions (8 = .614), higher relative status (8 = .607), treating each other as equals (8 = .566) and more personal relations (8 = .606). Finally, the two benefit items correspond to the second dimension: "I benefit-do not benefit" (8 = .629) an "something for his/her benefit-not for his/her benefit" (8 = -.722). It would appear that this dimension can be labeled as a "short term strategy-I benefit- involves a favor vs. long term strategy-do something for his/her benefit- involves a right" dimension. However, this is not a very parsimonious way of labeling a dimen— sion. Note that situations with positive loadings on this dimension all entail important future consequences for the relationship (not to date another person, meet relatives, friend not to swear so much, father to relax, a group of people to stop teasing a friend). Further, situations with negative loadings have little bearing on the directions of the rela- tionship (bouncer, pizza, parents to provide financial support, etc.). Thus, this dimension may also relate to an "important consequences for the relationship vs. no important consequences for the relationship" distinction. Situations which do not entail important consequences for the relationship usually involve simple requests, a favor, no future interaction and a benefit for the persuader. Of these, some clarifica- tion is in order concerning the attributes "no commitment to future interaction-commitment" and "I benefit-I do not benefit". The commitment-no commitment to future interaction attribute corresponds to this dimension only moderately well (8 = .614). The reason this 174 correspondence is not greater is simply because a number of situations exist in which a commitment to future interaction does not entail any change in the relationship. This is particularly true of the parent situations. The "I benefit-I do not benefit" attribute also corresponds only moderately well with this dimension (8 = .629). I suspect that the attributes used to assess "benefits" do so only at a global level. "Benefits", it should be noted, can be long term benefits (such as in the improvement of relationships —-situations with positive loadings on this dimension) or short term gains (as in the situations with negative loadings). Further, the situations that load positive also entail mu- tual benefits -both people benefit. Obviously, future research in this area should be well advised to differentiate the general notion of "benefits". All of the attributes which formed the "easy to persuade-difficult to persuade" set of attributes correspond with the third dimensional array of situations: "easy to persuade-difficult to persuade" (B = .674); "I know how to behave-I do not know how to behave" (B = .614); "comfortable-uncomfortable" (B = .878); and "resistant-not resistant to persuasion" (B = -.7l3). More importantly, however, the attributes which ought to form a friendly-unfriendly factor correspond highly to this dimensional array of situations: "friendly-unfriendly" (6 = .890); "difficult-easy to resolve conflicts" (B = -.865); "frank and open vs. reserved and cautious" (B = .675); and "s/he will think less of me-will not think less of me" (8 = -.820). Further, the attribute "assertive-I need not be assertive" corresponds to this dimensional array of situa- tions (8 = -.710), as well as the "formal-informal" attribute (B = .750); 175 however, the remaining status items do not correspond well to this di- mension. It would appear that the third dimension is an "easy to persuade-friendly vs. difficult to persuade-unfriendly" dimension in which the attributes which tap friendliness and resistance merge together in the sense that they are predicted to by the same dimensional array of situations. Apparently, the fourth dimensional array of situations corresponds to only one attribute: "I benefit—I do not benefit" (8 = .674). This suggests that the benefit attribute relates to two dimensions and further suggests how the notion of a "benefit" can be differentiated. The second dimension related to long term benefits in the sense of improved rela- tionships. With regard to the fourth dimension, the term benefit can be used to refer to both long term benefits (boy/girlfriend not to date another, meet relatives) and short term benefits (roommates to leave phone messages, bouncer to let you into bar, etc.). It should be noted that there are nine situations in the Form 1 set of situations which can be subsumed under the heading of a general category of personal (long- and short-term benefits), while only three situations entail a persuasion attempt for another person's benefit (persuade a group of people to stop reasing a friend, a father to take time off from work and a friend not to swear so much). These situations load high and positive on the fourth dimension. Thus, this dimension appears to correspond to a "general benefits" distinction. Table 15 presents the coordinate loadings of the twelve situations included in Form 2 in the four dimensional §Z§I_solution. Table 16 pre- sents the results of the regression analysis. The first dimension can be interpreted as the friendly-unfriendly dimension (see column one of 176 TABLE 15 Coordinates of 12 Situations (Form 2) in Four Dimensions from KYST Analysis (Stress = Situation 1. 2. 10. 11. 12. Persuade your boss to let you leave work early. Persuade a group of people to go to the bar with you. . Persuade a professor that an answer on a test is incorrect. Persuade a friend to go shopping with you. . Persuade your boy/girlfriend to go to a party where there is mostly your friends. Persuade your boy/girlfriend not to date another person. Persuade a friend to give you a ride. Persuade a sister/brother to go to a four-year University. Persuade a boy/girlfriend to study more. Persuade a person to ask you out. Persuade a group of people to participate in an experiment fOr your group project. Persuade parents that marijuana is not bad. 3.30) Dimension 1 2 i 5. .307 -1.042 .258 .145 -.963 -.224 -.052 -.396 .876 —.333 -.065 .324 -.996 -.041 -.l40 -.381 -.549 .605 .107 .085 .361 .755 .632 .027 -.766 -.l79 -.063 .575 .560 .294 -.884 -.272 .254 .490 -.299 .494 ’-.329 -.016 .635 -.113 -.095 -.275 -.367 -.O23 1.341 -.O33 .235 -.465 177 TABLE 16 Standardized Regression Weights for Four (KYST) Dimensions Used to Predict Mean Ratings on Unidimensional Atffibutes (Form 2; n=67) Dimensions: Attributes . 1 .. 2 3 4 R2 friendly-unfriendly -.905a .052 .333b .035 .933a difficult-easy a a to resolve conflicts .833 .071 .275 .257 .841 short term- b 0 long term strategy .569 .228 .181 .442 .604 comertable- a a a uncomfortable .808 .113 .515 .141 .951 resistant-not a a resistant to persuasion .923 .031 .193 .141 909 easy to persuade- a a difficult to persuade .957 .062 .140 .042 .941 I dg not benefit .530 .379 .486 .225 .711 I have been-I have never a b been in this situation .787 .246 .088 .280 .766 involves a favor— a b involves a right .816 .155 .111 .131 .720 do something fOr his/her benefit-not do something for his/her benefit .171 .487C .279 .525 .620‘2 I have relatively low b a status—high status .447 .702al .301° .201 .824 . b formal-informal .657a .515b .229 247 . 811 one totally dominates- a treat each other equally .682a .588a -.118 .222 .874 ap<.01 b p < .05 °p<.1o TABLE 16 (cont’d.) 178 . Dimensions: Attributes l. '2 g 3 4 R2 superficial-intimate b relationship .176 .845a .076 246 . 811 no commitment to future b interaction-commitment .213 .678 .104 .190 .552 frank and open- a a C a reserved and cautious .576 .572 .370 .144 .818 impersonal- a 6 personal relationship .094 .787 .106 .297 .728 emotional-unemotional a a a a situation .524 .685 .449 .065 .950 I know-I do not know a C b a how to behave .737 .290 454 .145 .853 he/she will think less of a a a a m-gill not think less .719 .037 .664 .018 .959 he/she has a major effect- does not have a major C b C effect on my life .488 .643 .126 .110 .679 I have a major effect- do not have a major a b effect on his/her life .184 .810 .008 .185 .725 I must be assertive- a C I need not be assertive .807 .087 .125 .152 .698 a p < .01 b p < .05 cp<.10 179 Table 16): "friendly-unfriendly" (B = .905); "difficult-easy to persuade" (B = -.833); "comfortable-uncomfortable" (B = .808); "resistant-not resistant" (8 = -.923); "easy to resolve—difficult to resolve conflicts" (B = .957); "he/she will think less of me-will not think less of me" (8 = -.719); and "I must be assertive-I need not be assertive" (B = -.807). Further, this dimension also corresponds to familiarity: "I have been-I have never been in this situation" (8 = .787) and "I know how to behave-I do not know how to behave" (B = .737). Finally, the more positive the loadings in this situation, the more the respondents perceive the situation as involving their rights, as opposed to favors (8 = .816). A number of the attributes from the superficial-intimate factor cor- respond with the second dimensional.array of situations: "superficial- intimate" (B = .845); "impersonal-personal" (8 = .787); and "no commitment-commitment to future interactions" (8 = .678). Further, the two items which assess influencability also correspond with this dimen- sion: "I have-I do not have a major effect on his/her life" (8 = -.810), and "s/he has a major effect-does not have a major effect on my life" (8 = -.643). Two of the status related items also appear to correspond somewhat to this dimension ("I have relatively low-high status", 8 = .701, and "dominates-treat each other as equals", 8 = .588); however, the attributes which formed the "formal—informal" set of attributes generally correspond to both the first two dimensions. Finally, the "frank and open-reserved and cautious" attribute and the "emotional-unemotional" attribute correspond to this dimension (8 = -.572 and -.685, respective- ly). Nonetheless, this dimension is clearly a "superficial-intimate" dimension. 180 Only two attributes correspond to the third dimension: "comfortable- uncomfortable" (B = .515) and "s/he will think less of me-will not think less of me" (8 = -.664). I believe that this dimension corresponds to an "intense feelings vs. non-intense feelings" distinction. Situations that load positive on this dimension deal with more intense feelings: per- suade a person to ask you out, persuade a boy/girlfriend not to date another person, a boss to let you leave work early, parents that mari- juana is not bad. Situations which load negative on this dimension are comfortable, unemotional and entail considerably less intense feelings (i.e., persuade a sister/brother to go to a four-year University, a group to participate in a group project, a boy/girlfriend to study more, a friend to go shOpping). None of the attributes correspond to the fourth dimension. Further, the manner in which the situations are arrayed on the dimension does not offer any interpretation. Therefore, this dimension will not be labeled. Table 17 presents the coordinate loadings of the twelve situations included in Form 3 in the four dimensional §:§T_solution. Table 18 pre- sents the results of the regression analysis. A number of attributes from the "superficial-intimate" factor and the "no major effect-major effect" factor correspond to the first dimensional array of situations (see column one of Table 18); "superficial-intimate" (B = -.914); "impersonal-personal" (B = -.928); "no commitment-commitment to future interaction" (8 = -.838); "s/he has a major effect on my life-does not have a major effect" (8 = .858); and "I have a major effect-I do not have a major effect" (8 = .919). Further, this dimensional array corresponds with the type of strategy to be employed ("short term strategy-long term strategy", 8 = —.90l), perceived level of emotional feelings ("emotional- 181 TABLE 17 Coordinates of 12 Situations (Form 3) in Four Dimensions from KYST Analysis (Stress = 1.48%) Dimension 8 ituat ion 1 2 3 4 l. Persuade a boy/girlfriend that there is nothing wrong in occa- sionally dating another person. -l.115 -.679 .093 .001 2. Persuade your roommate to keep quiet when you are trying to sleep. .415 .594 -.723 -.426 3. Persuade your boy/girlfriend to confide in you more. -.855 -.l99 -.402 -.059 4. Persuade a professor that a test item was ambiguous. .769 -.303 .415 -.133 5. Persuade a professor that his grading scale is too hard. .635 -.254 .340 -.188 6. Persuade a person you know in class to help you with your home- work. .213 .179 —.736 .065 7. Persuade your landlord to fix the plumbing in your apartment. .645 .200 -.409 .612 8. Persuade an old boy/girlfriend that the relationship is worth saving. -.927 -.431 -.226 .032 9. Persuade your father into getting a checkup. -.481 .845 .470 -.O7O 10. Persuade a police officer not to give you a ticket. .652 -.487 .315 .172 ll. Persuade a professor to increase your grade. .586 -.300 .196 -.160 12. Persuade your mother to go back to college and get a degree. -.536 .835 .665 .154 182 TABLE 18 Standardized Regression Weights for Four (KYST) Dimensions Used to Predict Mean Ratings on Unidimensional Attributes (Form 3; n=66) Dimensions: a p < .01 b p < .05 c p < .10 Attribute? . 1... . 2 3 4 R2 friendly-unfriendly .659a -.581a .146 -.112 807b difficult-easy a b b a to resolve conflicts -.647 .516 -.390 -.024 839 short term- a a long term strategy -.901 .138 .137 .133 .868 uncomfbrtable .466 -.741 .187 -.080 .808 resistant-not b a a resistant to persuasion -.365 .480 -.742 .026 .916 easy to persuade- a a a a difficult to persuade .330 -.479 .781 064 .952 I do not benefit -.707 .442a .525a .091 .980 .I have been-I have never a a been in this situation .244 .043 .574 .665 .833 involves a favor- involves a right .234 -.147 -.035 ".019 .076 do something fer his/her benefit-not do something a a C a for his/her benefit .806 -.483 -.l94 -.131 .939 I have relatively low a b a a status-high status -.683 .387 -.480 -.033 .848 formal-infernal -.708a .541a -.403a -.076 .962a one totally dominates- a b a treat each other equally -.706 290 -.592a -.043 .936 TABLE 18 (cont'd.) 183 ------ . Dimensions: Attributes l . 2 '3 4 R2 superficial- int imate C C b b relationship -.941 .293 -.103 -.O94 .991a no commitment to future C C C interaction-commitment - . 838 . 469 - .104 . 079 . 940 frank and open- a a b a reserved and cautious .659 .576 .354 .030 .892 impersbnal- C C b a personalrelationship -.928 .326 - 100 .121a .992 emotional-unemot ional a b a situation .843 .344 —.201 .133 .887 I know-I do not know c a a b - how to behave .256 .597 .600 .317 .883 he/she will think less of C C a me-will not think less -.359 .779 -.800 .120 750 he/she has a major effect- does not have a major a a a a effect on my life .858 .312 -.334 .150 .966 . I have a major effect- do not have a major C C b C effect on his/her life .919 .347 -.118 -.027 .979 I must be assertive- a C a a I need not be assertive -.645 .308 -.571 - 116 .850 l . . , ap<.Ol bp<.05 cp<.10 184 unemotional", B = .843), and who benefits from the persuasion ("I benefit-I do not benefit", 8 = -.707; and "do something for his/her benefit-not do something for his/her benefit", 8 = .806). This dimen- sion is the "superficial-intimate" dimension, subsuming the two items which assess influencability. Only two attributes appear to correspond to the second dimension: "comfortable—uncomfortable" (B = -.741) and "s/he will think less of me- will not think less of me" (8 = .779). This dimension can be labeled similarly to the third dimension of the Form 2 solution: "intense feelings vs. non-intense feelings". There is good reason to believe that the situations arrayed on the third dimension correspond to the attributes which formed the "easy to persuade-difficult to persuade" factor: "easy to persuade-difficult to persuade" (B = .781), "resistant-not resistant to persuasion" (B = —.742L and "I know how to behave..." (B = .600). A careful scrutiny of how the situations are arrayed on this dimension will support this conclusion. First, when persuading a mother to go back to college, or when persuading a father to get a checkup, one has to overcome some non-trivial obstacles. Other situations which have positive loadings on this dimension include the situations dealing with professors and the police. Resistance in such situations is obvious. In contrast to the "easy to persuade- friendly vs. difficult to persuade-unfriendly" dimensions obtained in the solutions for the Form 1 and Form 2 data, the friendly-unfriendly set of attributes do not correspond to the same dimension of situations as the "easy to persuade-difficult to persuade" attributes in the Form 3 solution. The friendly-unfriendly set of attributes correspond moder- ately to the first two dimensions. 185 The first thing that should be noted about the fourth dimension is that most of the situations have essentially a zero coordinate loading. In reality, this dimension appears to be marked by three situations with positive loadings (repairing plumbing, .612; police officer, .172; and persuading your mother to go back to college, .154), and by four situa- tions with negative loadings (roommate(s) to keep quiet, -.426; a pro- fessor about his grading scale, -.188; a professor about an increase in a grade, -.160, and a professor that a test item was ambiguous, -.l33). The only attribute which faintly corresponds to this dimensional array of situations is the attribute "I have been-I have never been in this situation" (8 = .665). Given the manner in which the situations load on this dimension, it seems that this dimension may be a familiarity dimen- sion, in which the respondents have rarely, if ever, attempted to per- suade their mother to go back to college, to persuade a police officer not to ticket them or ever needed to persuade a landlord to repair the plumbing. On the other hand, the respondents have apparently requested their roommate(s) to be quiet a number of times, and have discussed grading scales and increases in grades with professors at least occasionally. Table 19 presents the coordinate loadings of the twelve situations included in Form 4 in the four dimensional gig: solution. Table 20 presents the results of the regression analysis. Since the situations included in Form 3 are the same as in Form 4, one would expect to obtain an extremely similar interpretation of the Form 4 dimensional solution. However, it appears as though the first dimensional array corresponds to five of the six factors of attributes, and to a number of additional attributes (see column one of Table 20): "friendly-unfriendly" (B = 186 TABLE 19 Coordinates of 12 Situations (Form 4) in Four Dimensions from KYST Analysis (Stress = 2.73%) Dimension Situation 1 2 3 4 l. Persuade a boy/girlfriend that there is nothing wrong in occa- sionally dating another person. .386 -.784 .655 .091 2. Persuade your roommate to keep quiet when you are trying to sleep. .340 1.044 .559 .150 3. Persuade your boy/girlfriend to confide in you more. .762 -.290 .139 -.547 4. Persuade a professor that a test item was ambiguous. -.836 .180 -.157 -.396 S. Persuade a professor that his grading scale is too hard. -.928 -.O22 -.096 -.133 6. Persuade a person you know in class to help you with your homework. .384 .733 .021 -.462 7. Persuade your landlord to fix the plumbing in your apartment. -.512 .339 .126 .627 8. Persuade an old boy/girlfriend that the relationship is worth saving. .730 -.S25 .114 .025 9. Persuade your father into getting a checkup. .711 -.O36 -.448 .475 10. Persuade a police officer not to give you a ticket. -.776 -.337 .266 .182 11. Persuade a professor to increase your grade. -.903 -.343 -.314 -.109 12. Persuade your mother to go back to college and get a degree. .642 .042 -.864 .097 187 TABLE 20 Standardized Regression Weights for Four (KYST) Dimensions Used to Predict Mean Ratings on Unidimensional Attributes (Form 4; n=66) . Dimensions: Attribute? .l... .2 3 .4 .122 friendly-unfriendly -.813a -.065 .345c .251 .847a difficult-easy a a to resolve conflicts .927 .274 -.049 -.l66c .985a short term- b b b lung term strategy .619 -.511 -.238 .218 .749 comfortable- a b b uncomfortable -.843 -.342 .283 .144 .929a resistant-not a a C - resistant to persuasion .789 .488 .017 -.247 .92“ easy to persuade- a a a difficult to persuade -.728 -.572 -.147 .169 .908 I benefit- b b b I do not benefit .542 -.307 -.530 .241 .727 I have been-I have never _ b c been in this situation -.l47 -.314 -.363 .638 .659 involves a favor- involves a right -.LH+0 .076 .425 .338 .494 do something for his/her benefit-not do something a c 7b a for his/her benefit --800 -283 .35 -.203 .890 I have relatively low a a status-high status .933 .063 .093 .038 .884 formal-informal .940a .302a - . 028 . 027 .9775l one totally dominates- a c .909 .156 .242 -.01s .910a treat each other equally a p < .01 b p < .05 c p < .10 TABLE 20 (cont'd.) 188 , Dimensions: Attr ”butes 1. .2 3 4 22 superficial—intimate a b a relationship .954 .224 .006 -.O4O .962 no commitment to future a interaction-comitment . 932 . 150 . 056 . 060 . 899 frank and open- a C a memd and cautious - 881 .2714 .022 ".079 .858 impersonal- b personal relationship .945a .235 .018 -.044 .950 emotional-unemotional situation - 483a .828a .130 .030 .938 I knowbI do not know b how to behave -.519 .838a .147 .379b .838 he/she will think less of a b me-will not think less .746 .358 .474a .045 .910 he/she has a major effect- does not have a major a a C effect on my life -.728 .502 .268 .090 .8823 I have a major effect- do not have a major a b effect on his/her life -.879 .315 .117 -.169 .914 I must be assertive- a b b I need not be assertive .89]. .289 .01-l7 - 265 951 ..... ..... bp<.05 cp<.10 189 -.813); "difficult-easy to resolve conflicts" (8 = .927); "frank and open-reserved and cautious" (B = -.881); "comfortable-uncomfortable" (8 = -.843); "resistant-not resistant to persuasion" (B = .789); "easy- difficult to persuade" (B = -.728); "superficial-intimate" (B = .954); "impersonal-personal" (B = .945); "no commitment to future interaction— commitment" (8 = .932); "formal-informal" (B = .940); "I have relatively low status-high status" (8 = .933); "dominates-treat each other as equals" (8 = .909); "s/he will think less of me-will not think less of me" (8 = .746); "I benefit—I do not benefit" (8 = .542); "do something for his/her benefit-not do something for his/her benefit" (8 = -.800); "s/he has a major effect-does not have a major effect on my life" (8 = -.728); "I have a major effect-I do not have a major effect on his/ her life" (8 = -.879); and "I must be assertive-I need not be assertive" (B = .891). When I discussed the interpretation of the four dimensional solution of Form 3 data, I labeled the first dimensional array of situations as the "superficial vs. intimate" dimension (subsuming the two influenca- bility items). It was also noted that the attributes in the formal- pinformal factor of attributes (along with those attributes which assess the perceived level of friendliness) corresponded to both of the first two dimensional arrays of situations. Apparently, the use of the assumed ratio scale (or a change in the mean ratings on attributes) was suffi- cient to have merged the various factors such that attributes from five different factors relate to a single dimensional array of situations. In terms of providing a psychological interpretation of the second dimensional array of situations, only the attribute "emotional- unemotional" (B = .828) corresponds to this dimension. Situations with 190 negative loadings are situations perceived as more emotional than the situations which have positive loadings. In the solution for Form 3, the "comfortable-uncomfortable" attribute corresponded with the second dimension and the attribute "emotional-unemotional" corresponded to both the first two dimensions, primarily the first. In the Form 4 solution, the "comfortable-uncomfortable" attribute corresponds to the first three dimensions, but primarily the first. Apparently, a difference in scale usage (assumed interval vs. asumed ratio) resulted in substantial changes in the dimensional structure; the second dimensional array of situations in the Form 4 data is an "emotional-unemotional" dimension. Since the zero order correlation between the two second dimensions for the Form 3 and Form 4 solutions is .954, it seems legitimate to conclude that this dimension is an "intense feelings" dimension, a construct assessed by either of these two attributes. Apparently, there is no attribute which corresponds to the third dimensional array of situations. Despite the fact that the zero order correlation between the third dimensions for the Form 3 and Form 4 solu- tions is .707 and despite the fact that the third dimension of the Form 3 solution was labeled "easy to persuade vs. difficult to persuade", the third dimension of the Form 4 solution does not appear to correspond to this distinction. This dimension will not be labeled. Finally, the fourth dimension is apparently the familiarity dimen- sion: "I have been-I have never been in this situation" (8 = .638). It is similar to the fourth dimension obtained for the Form 3 data. Situations with negative loadings are situations in which the respond- ents have been more often than the situations with positive loadings. 191 The following can summarize the obtained dimensions, and attributes relevant to the dimensions: I. II. III. IV. I. II. For Form 1: "subordinate relationship vs. social affiliation" emotional-unemotional (B = -.679), s/he has a major effect on my life-does not have a major effect (8 = -.876). "improves relationships-personally important goal vs. does not improve relationship—not personally important goal" I benefit—I do not benefit (8 = .629), do something for his/her benefit-not do something for his/her benefit (8 = -.772), involves a favor-involves a right (8 = .701), short term-long term strategy (8 = .831). "easy to persuade-friendly vs. difficult to persuade-unfriendly" friendly-unfriendly (B = .890), easy to persuade-difficult to persuade (B = .674), easy to resolve conflicts-difficult to resolve conflicts (B = -.865), comfortable-uncomfortable (B = .878), s/he will think less of me-will not think less of me (6 = -.820), informal—formal (B = .750). General Benefits I benefit-I do not benefit (8 = .674). For Form 2: "easy to persuade-friendly vs. difficult to persuade-unfriendly" friendly-unfriendly (B = -.905), easy to resolve conflicts-difficult to resolve conflicts (8 = -.833), comfortable-uncomfortable (B = .808), not resistant-resistant (B = -.923), s/he will think less of me-will not think less of me (8 = -.7l9), easy to persuade-difficult to persuade (B = .957), favors-rights (B = .816). Superficial vs. Intimate superficial-intimate (B = .845), impersonal-personal (B = .787), s/he has-does not have a major effect on my life (8 = -.643), III. IV. II. III. 192 - I have-do not have a major effect on his/her life (8 = -.8lO), ° I have relatively high-low status (8 = .702). Intense Feelings ° comfortable-uncomfortable (B = .515), ' s/he will-will not think less of me (8 = -. 64). Unlabeled For Form 3: Superficial vs. Intimate - superficial-intimate (8 = -.94l), ° no commitment to future interaction-commitment (B = -.838), ' impersonal-personal (8 = -.928), ° s/he has-does not have a major effect (8 = .858), ° I do-do not have a major effect on his/her life (8 = .919), ° short term strategy-long term strategy (8 =.901), ' do something-do not do something for his/her benefit (8 = .806), ° I benefit-I do not benefit (8 = ~.707), ° emotional-unemotional (8 = .843). Intense Feelings ' comfortable-uncomfortable (B = -.74l), ° s/he will-will not think less of me (8 = .779). "easy to persuade vs. difficult to persuade" ° easy to persuade-difficult to persuade (B = .781), ' not resistant-resistant to persuasion (B = .742), ° I know how to behave-I do not know how to behave (B = .600). Familiarity - I have-have never been in this situation (8 = .665). For Form 4: Friendly-intimate-easy to persuade vs. unfriendly-superficial- difficult to persuade. ° friendly-unfriendly (B = -.813), ° difficult to resolve conflicts-easy to resolve conflicts (B = .927), ° comfortable-uncomfortable (B = -.843), ° easy to persuade-difficult to persuade (B = -.728), - not resistant-resistant (B = .789), ° high status-low status (8 = .933), ° informal-formal (B = .940), 193 ° treat each other equally-one totally dominates (B = .909), : intimate-superficial (B = .954), ° commitment-no commitment to future interaction (8 = .932), ' impersonal-personal (B = .945), ° s/he has-does not have a major effect on my life (8 = -.728), ° I have-do not have a major effect on his/her life (8 = -.879), ° I need not be assertive-I must be assertive (B = .891), ° s/he will-will not think less of me (8 = .746), ° do something-not do something for his/her benefit (8 = -.800). II. Intense Feelings ° emotional-unemotional (B = .808), ° I know how to behave-I do not know how (8 = -.636). III. Unlabeled IV. Familiarity 0 I have been-I have never been in this situation (8 = .638). While a number of differences can be noted across the various solu- tions, the occurrence of strong commonalities is very encouraging. First, it is clear that an "easy to persuade vs. difficult to persuade" dimen- sion is replicated and generalizable. Further, a "superficial vs. intimate" dimension is also replicated. (Note, in additional analysis using the GALILEO coordinates, there was some indication that the fifth dimension of the Form 1 solution corresponds moderately well with the superficial-intimate distinction. Therefore, it would not be surprising to find that future research would obtain solutions of higher dimension- ality than four. However, the replication of the "superficial vs. intimate" dimension across the other three sets of data and across two sets of situations is sufficient to warrant the conclusion of replication.) Unfortunately, attributes which can assess the levels of intensity of feelings and emotions were lacking in this study. Because of this failure, few attributes corresponded to the "intense feelings" 194 dimensions. Further, while I have suggested that the first dimension of the Form 1 solution is a "subordinate relationships vs. social affilia- tion", there is some indication that this dimension may be best labeled an "intense feelings" dimension. I say this because the several attri- butes which did correspond with the first dimension of the Form 1 solu- tion also corresponded moderately well with the "intense feelings" dimensions for the Form 2, 3, and 4 solutions. However, this is not sufficient reason to relabel the dimensions. Future research is needed in order to assess levels of emotions and intensity of feelings. The most important consequence of the exclusion of the attributes which could assess intensity of feelings is concerned with the generality of the factor analytic solution. Although six factors were obtained, none of the factors corresponded with an intensity of feelings distinc- tion while the multidimensional scaling solutions indicate replication of such a factor. Another important recommendation concerning future research deals with the notion of who benefits. The term "benefits" is too global a term to capture a number of subtle differences among types of benefits. Specifically, some benefits are long-term, some short-term, some only _ involve the persuader, some involve the persuadee, some involve a third person and some involve a benefit for both individuals (mutual benefit) or an entire group of peOple. Similarly, a distinction must be made between altruistic and selfish motives. Developing a more differentiated notion of benefits would help to provide a bettern understanding of the second and fourth dimensions of the Form 1 solution. 195 3.5 The hierarchical structure of situations: Results of Johnson's (1967) hierarchical clustering of situations In order to make the above discussion concerning the interpretation of the dimensional structure of situations more concrete, some support for the proposed system of interpretations of dimensional arrays of sit- uations can be found in the nature of the clusters of situations. Figure LBpresents the tree diagram for the cluster analysis of the Form 1 data. The entry levels (see columns at left of Figure 13) indicate the distance between situations at the point of merger. Table 21 presents the situa- tions included in each of the clusters (in descending order of strength of entry). Cluster one consists of two of the three situations which deal with parents -the third parent item (father to find something relaxing to do) is included in cluster four; cluster two consists of situations in which the respondent is interacting with an individual who is involved in a specific role (professor and "bouncer"); cluster three consists of the two boy/girlfriend situations; cluster four consists of those situations in which the respondent is persuading someone else to do something for their (or a third person's) benefit; cluster five consists of situations dealing with friends and roommates. Clusters one and two include situations in which the persuader attempts to gain some benefit from individuals who are superordinates. These situations are four situations which loaded positive on the first dimension of the Form 1 solution and can be contrasted to the situations in cluster four (loading negative on the first dimension). These situa- tions are social, casual encounters. This distinction supports the labeling of the first dimensional array of situations as a "subordinate relationships vs. social-affiliation relationships" dimension. Tree Diagram of Results of Johnson's Clustering of Form 1 Data entry levels: 3.266 3.969 4.125 4.156 4.891 5.688 6.063 6.641 7.516 7.813 8.406 3.266 3.969 4.125 4.156 4.891 5.688 6.063 6.641 7.516 7.813 8.406 196 FIGURE 13 HO (DO (00 : XXX : XXX XXXXX XXXXX OH O O O O 0 1 O 1 3 5 4 7 2 1 6 2 . : : XXX . : : : : XXX XXX : XXX : : XXX XXX : XXX : : XXX XXX : XXX XXX XXX XXX : XXX XXX XXX XXX : XXX XXX XXXXXXX XXXXX : XXXXX XXXXXXX XXXXXXX XXXXXXXXXXXXXXX XXXXX XXXXXXXXXXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXX 4———e>3—-—%’5—€'2-€r1 (1967) Hierarchical situation cluster 197 TABLE 21 Situations Included in Clusters Generated from Johnson's (1967) Hierarchical Clustering Program (Form 1 Data) Cluster 1: 6. 12. Cluster 2: 2O 11. Cluster 3: Cluster 4: 8. g. 1. Cluster 5: u 7. Situations (in order of entry) Persuade your parents to let you stay out late. Persuade your parents to give you money. Situations (in order of entry) Persuade a professor to let you enroll in an already crowded class. Persuade a bouncer to let you into a bar even though you have forgotten your I.D. Situations (in order of entry) Persuade a group of people to stop teasing a friend of yours. Persuade a friend not to swear so much. Persuade your father to take more time off from work and find something relaxing to do. Situations (in order of entry) Persuade a group of people to order a pizza because you feel like eating one. Persuade a friend to take a class with you. Persuade your roommate(s) to leave phone messages for you while you are out. Situations (in order of entry) Persuade your boy/girlfriend to meet your relatives. Persuade your boy/girlfriend not to date another person. 198 Situations comprising cluster five load positive on the second dimension, followed by situations in cluster three. These situations entail actions which have important consequences for the development of the relationship. Situations inclusive of clusters two, one and four (which load negative on the second dimension) do not entail any actions which entail important consequences concerning the future of the rela- tionship. It would appear that situations in cluster two are situations which are perceived as unfriendly and where the persuasion is difficult, while situations in clusters one and four are perceived as more friendly and entailing less resistance. This is congruent with the manner in which situations are arrayed on the third dimension and supports the interpre- tation of that dimension. Further, situations in cluster three load positive on the fourth dimension and entail a persuasion attempt which provides a benefit for a person other than the persuader, while situations in clusters five and two (as well as situation one) load negative on the fourth dimension and entail a persuasion attempt for the persuader's benefit. Figure 14 presents the results of the hierarchical clustering of the Form 2 data. There are four clusters. Table 22 presents the situations included in each cluster. Cluster one includes situations that deal with friends and includes one of the situations that deal with a boy/girlfriend relationship (go to a party). All of the situations included in this cluster entail joint activities (friendly) and short-term persuasion. Cluster two consists of situations that deal with tasks -selecting a University, studying and obtaining participants for a group project. The first two situations 199 FIGURE 14 Clustering of Form 1 Data entry levels: 1.896 4.119 4.194 4.866 4.881 5.179 5.881 7.045 7.493 8.373 9.299 1.896 4.119 4.194 4.866 4.881 5.179 5.881 7.045 7.493 8.373 9.299 0 O O 2 4 5 XXX : XXX : XXXXX : XXXXX XXXXXXX XXXXXXX XXXXXXX XXXXXXX XXXXXXX 0 1 O O 1 O O 1 6 O 8 9 1 1 3 2 : XXX : : : XXX : : XXX XXX : XXX XXX : : : : XXX XXX : : XXX XXX XXXXX : XXX XXX XXXXX XXXXX XXXXXXXXX XXXXX XXXXXXXXXXXXXXXXX XXXXX XXXXXXXXXXXXXXXXXXXXXXX 1-———9'3—$'2-——>l+——9 Tree Diagram of Results of Johnson's (1967) Hierarchical, situation cluster 200 TABLE 22 Situations Included in Clusters Generated from Johnson's (1967) Hierarchical Clustering Program (Form 2 Data) Cluster 1: Situations (in order of entry) 2. Persuade a group of people to go to the bar with you. 4. Persuade a friend to go shopping with you. 5 Persuade a boy/girlfriend to go to a party where there is mostly your friends. 7. Persuade a friend to give you a ride. Cluster 2: Situations (in order of entry) 8. Persuade a sister/brother to go to a four-year university. 9. Persuade a boy/girlfriend to study more. 11. Persuade a group of people to participate in an experi- ment for your group project. Cluster 3: Situations (in order of entry) 6. Persuade your boy/girlfriend not to date another person. 10. Persuade a person to ask you out. Cluster 4: Situations (in order of entry) 3. Persuade a professor that an answer on a test is incorrect. 12. Persuade parents that marijuana is not bad. 1. Persuade your boss to let you leave work early. 201 deal with another person's benefitting from the persuasion, and the situation dealing with a group project merges later. Thus, this cluster may have something to do with benefits, as well as with tasks. Cluster three consists of two situations which deal with initial stages of a boy/girlfriend relationship. Cluster four included situa- tions which merged fairly late; situations that deal with professors, parents and a boss. It appears as though the situations dealing with parents sampled in Form 2 do not involve the parent's benefit, and situations which deal with both parent relationships and professionals could involve more emotional interactions; possibly more difficult to persuade situations, than the situations sampled in Form 1. In relation to the first dimensional array of situations, the friendly-unfriendly dimension, situations from cluster one load highly negative (friendly), while situations from cluster four have highly positive loadings (unfriendly, resistance to persuasion). The second dimensional array was labeled the superficial vs. intimate dimension and situations from clusters two and three have positive loadings on this dimension (intimate) and situations in clusters four and one have nega- tive loadings (superficial). The last two dimensions of the Form 2 solu- tion did not predict to any of the attributes. However, it was argued that the third dimension may be related to perceived level of intensity of interaction. In general, situations from clusters three and four have positive loadings on this dimension (intense, emotional) and situations in cluster two load negative (non—intense, unemotional). Figure 15 presents the results of the cluster analysis of Form 3 data. There are four global clusters of situations, with the situations in the fourth cluster merging at fairly high entry levels. Table 23 202 FIGURE 15 Tree Diagram of Results of Johnson's (1967) Hierarchical Clustering of Form 3 Data entry levels: 2.364 2.833 2.879 3.546 4.500 4.894 4.955 5.864 6.849 7.636 8.894 2.364 2.833 2.879 3.546 4.500 4.894 4.955 5.864 6.849 7.636 8.894 _ MO 00 \10 OH : XXX : XXX ' 430 01c: 14:4 : XXXXX : : XXXXX : : XXXXXXX : : XXX XXXXXXX : XXX XXXXXXX XXXXX XXXXXXX XXXXXXXXXXXXX XXXXXXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXX 4——-—91—-—-—4>2-——?3-9 cluster H0 (.00 (DO LOO NH : XXX : : XXX XXX XXX : XXX XXXXX XXXXX XXXXX XXX XXX XXX XXX XXX XXX XXXXXXXXX situation 203 TABLE 23 Situations Included in Clusters Generated from Johnson's (1967) Hierarchical Clustering Program (Form 3 Data) Cluster 1: 4. 5. 11. 1o. Cluster 2: Cluster 3: g. 12. Cluster 4: Situations (in order of entry) Persuade a professor that a test item was ambiguous. Persuade a professor that his grading scale is too hard. Persuade a professor to increase your grade. Persuade a police officer not to give you a ticket. Situations (in order of entry) Persuade your boy/girlfriend to confide in you more. Persuade an old boy/girlfriend that the relationship is worth saving. Persuade a boy/girlfriend that there is nothing wrong in occasionally dating another person. Situations (in order of entry) Persuade your father into getting a checkup. Persuade your mother to go back to college and get a degree. Situations (in order of entry) Persuade a person you know in class to help you with your homework. Persuade your landlord to fix the plumbing in your apart- ment. Persuade your roommate to keep quiet when you are trying to sleep. 204 presents the situations included in each cluster. The first dimensional array of situations was labeled a superficial vs. intimate dimension. Situations which load positive, in clusters one and four, are situations involving superficial relationships, while situations with negative coordinate loadings, in clusters two and three, involve more intimate relationships. The second dimensional array of situations obtained for the Form 3 solution was labeled intense feelings. Situations from clus- ters three and four have positive loadings on this dimension (non- intense interactions). Situations in clusters one and two are perceived as intense interactions. The third dimensional array of situations was labeled "easy to per- suade vs. difficult to persuade".. In general, situations in clusters three and one entail persuasion attempts that are perceived as difficult. Situations in clusters two and four are perceived as entailing little resistance. Situations which correspond to the familiarity (fourth) dimension cannot be identified in any combination of the clusters. Figure 16 presents the results of the cluster analysis of the Form 4 data. There are four clusters. Table 24 presents the situations in each of the four clusters. Since the situations are the same in Forms 3 and 4, and since three of the four dimensions were interpreted as the same across the two four-dimensional configurations, it should be suffi- cient here to note the commonalities and differences in the two cluster analysis solutions. Aside from a few situations which merge at rela- tively different entry levels, the obtained four clusters are the same except for situation seven (persuade your landlord to fix the plumbing in your apartment). In the cluster analysis results of the Form 4 data, situations seven merged into cluster one, a set of situations which are 205 FIGURE 16 Clustering of Form 4 Data Tree Diagram of Results of Johnson's (1967) Hierarchical O 1 O O 1 O O O O O O 1 . . 7 0 4 5 1 1 3 8 2 8 9 2 Sltuatl°n entry levels: : : 24.288 24.288 XXX : 36.712 36.712 XXXXX : : : : 41.561 41.561 XXXXX : XXX : : : : 48.409 48.409 : XXXXX : XXX : : XXX 56.106 56.106 : : XXXXX : XXX XXX XXX 59.667 59.667 : XXXXXXX : XXX XXX XXX 72.591 72.591 : XXXXXXX XXXXX XXX XXX 80.985 80.985 XXXXXXXXX XXXXX XXX XXX 95.424 95.424 XXXXXXXXX XXXXX XXXXXXX 113.258 113.258 123.394 123.394 XXXXXXXXX XXXXXXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXX 1———> 2——-) 4-—)3—-) cluster 206 TABLE 24 Situations Included in Clusters Generated from Johnson's (1967) Hierarchical Clustering Program (Form 4 Data) Cluster 1: Cluster 3: 90 12. Cluster 4: 2. 6. Situations (in order of entry) Persuade a professor that his grading scale is too hard. Persuade a professor to increase your grade. Persuade a professor that a test item was ambiguous. Persuade a police officer not to give you a ticket. Persuade your landlord to fix the plumbing in your apart- ment. Situations (in order of entry) Persuade your boy/girlfriend to confide in you more. Persuade an old boy/girlfriend that the relationship is worth saving. Persuade a boy/girlfriend that there is nothing wrong in occasionally dating another person. Situations (in order of entry) Persuade your father into getting a checkup. Persuade your mother to go back to college and get a degree. Situations (in order of entry) Persuade your roommate to keep quiet when you are trying to sleep. Persuade a person you know in class to help you with your homework. 207 somewhat unfriendly and entail a persuasion attempt that may involve some resistance in a superficial relationship. This situation merged fairly late in the clustering process. In the Form 3 cluster analysis results, this situation merged into cluster four, a cluster of situations which are neither very friendly or unfriendly, nor superficial or intimate. Situations in this cluster may be characterized as involving simple requests. Given that the situations merged late into the "weakest" cluster of the Form 3 and Form 4 results, it would appear that this difference in the solution is not a major dif- ference. Obviously, some set of individuals may perceive interactions with landlords as unfriendly, while others perceive landlords as friendly and somewhat cooperative. Note that the situation that exhibited the most change in the dimen- sional structure comparison (situation eleven, "persuade a professor to increase a grade") only experiences a change in degree of entry levels into the first clusters of both the Form 3 and Form 4 solutions. This suggests that the degree of similarity between situation 11 and situa- tions 4, 5 and 10 are fairly strong across the two data sets, and that the lack of stability of the situation may be due to instability with other situations. However, situation seven was stably located in the dimensional structure comparison, and its lack of stability in the clus- ter analysis results is probably due to the fact that it only weakly cor- responds to each of the respective clusters it merges into (see above). In hierarchical cluster analysis, it is possible to discuss how clusters continue to merge together until all objects have merged into one cluster. However, by observation, one can note that situations arranged from left to right in Figures l3, 14, 15 and 16 are arranged 208 more or less in terms of types of relationships. For instance, in Figure 13, situations in clusters one and two merge (situations which involve the respondent in subordinate relationships), situations in clusters three and five merge (involving important relationships), followed by situations in clusters one, two, three and five merging and ultimately situations in cluster four merging last (social/affiliation situations not involving resistance). The results of the exploratory study (see Figure 1) also illustrate this observation. More global clus- ters identified relationship types, while various types of actions (behaviors, feelings goals) could be identified as subclusters. Nonethe- less, the interpretation of the cluster analysis results is, as expected, quite compatible with the results of the interpretation of the dimension- al structures. 3.6 Dimensional Structure: Results of INDSCAL analysis ‘vY It was noted above that the solutions provided by the INDSCAL model were often substantially different than the ones obtained by using either the KYST (TORSCA) or GALILEO models. I will only briefly discuss the interpretation of the INDSCAL solutions. Since the main reason for employing the INDSCAL model lies in the advantage of obtaining dimension- al salience weights for each individual on all dimensions, it will be convenient to discuss points of view that may be operative in each of the four sets of data at the same time the interpretation of dimensions is presented. Therefore, in this section of the paper, I shall present the interpretations of the INDSCAL solutions and the results of hierarchi- cally clustering the respondents based on their salience weights. Fur- ther, I shall identify subsets of individuals based on the results of the 209 cluster analysis to be used in subsequent analysis (see Section 3.7) in order to assess in greater detail the differences in points of view. Table 25 presents the coordinate loadings of the twelve situations included in Form 1 in the INDSCAL solution. Table 26 presents the stan- dardized regression weights obtained by regressing the coordinate load- ings of the situations onto the mean ratings on unidimensional attri- butes. The first dimension appears to correspond to the "easy to persuade vs. difficult to persuade" set of attributes, although there is substan- tial overlap between the first and third dimensions. There is also an indication that this dimension corresponds to levels of perceived emo- tional involvement (8 = .789). The second dimension appears to be an "I benefit-I do not benefit" (8 = .694) dimension (also, "s/he benefits-s/he does not benefit", 8 = -.7ll). While there is considerable overlap in how the attributes correspond to the first and third dimensions, it appears that the third dimension corresponds to the friendly-unfriendly set of attributes, the status set of attributes and the personal- impersonal set of attributes. None of the attributes correspond to the fourth dimension. For convenience, I will label the four INDSCAL dimensions as: (1) "easy to persuade-comfortable vs. difficult to persuade-uncomfort- able"; (2) "I benefit vs. I do not benefit"; (3) "friendly-personal vs. unfriendly-impersonal"; and, (4) Unlabeled. Table 27 presents the coordinate loadings of the twelve situations in Form 2 in the four dimensional INDSCAL solution. Table 28 presents the standardized regression weights obtained by regressing the coordinate loadings of the situations onto the mean ratings of the unidimensional attributes. 210 TABLE 25 INDSCAL Coordinates for Form 1 (Four Dimensions) Situation 1. 10. ll. 12. Persuade your roommate(s) to leave phone messages for you while you are out. Persuade a professor to let you enroll in an already crowded class. . Persuade a group of people to stop teasing a friend of yours. Persuade your boy/girlfriend to meet your relatives. Persuade a friend not to swear so much. Persuade your parents to let you stay out late. Persuade your boy/girlfriend not to date another person. Persuade a group of people to order a pizza because you feel like eating one. Persuade a friend to take a class with you. Persuade your father to take more time off from work and find some- thing relaxing to do. Persuade a bouncer to let you into a bar even though you have forgotten your I.D. Persuade your parents to give you money. Dimension 1 .608 -.263 -.O35 -.010 .162 -.238 -.232 .343 .335 -.362 -.O67 -.243 -.O45 -.256 .430 .235 .433 -.300 .345 -.209 -.013 -.292 -.402 up .110 -.416 -0330 .313 -.062 .238 .071 -.l37 .039 .441 .248 [4: -0080 -.23l .532 -.305 .049 -0022 -.435 .479 -0057 .181 -0279 .168 211 TABLE 26 Standardized Regression Weights for Four (INDSCAL) Dimensions Used to Predict Mean Ratings on Unidimensional Attributes (Form 1; n=64) Dimensions: Attributes 1 2 3 4 R2 friendly-unfriendly -.489b .302 -.750a -.157 .787b difficult-easy b a to resolve conflicts .392 -.180 .872 .095 .847a short term- b a a long term strategy -.406 .800 .541 .046 .883a comfortable- a a uncomfortable -.598 .232 -.738 -.204 .874a resistant-not a b b resistant to persuasion .663 -.275 .613 .105 .765 easy to persuade- a b difficult to persuade -.690 .334 -.564 -.031 .750 I benefit- a b I do not benefit -.266 .694 .112 .479 .747 I have been-I have never c c been in this situation -.229 .502 -.289 —.481 .578 involves a favor- C involves a right .112 .577c .385 -.212 .548 do something for his/her benefit-not do something a b for his/her benefit -.004 -.711 -.338 -.238 .714 I have relatively low b b a a status-high status .341 .387 .765 .172 .901 formal-informal .809a .069 .882a .303 .903a one totally dominates- a b a a treat each other equally .586 .463 .526 .045 .861 a b C p < .01 p < .05 p < .10 TABLE 26 (cont'd.) 212 Dimensions: Attributes 1 2 3 4 R2 superficial- intimate relationship .256 .195 .515 -.172 .379 no commitment to future a a interaction-commitment -.O90 .268 .845 -.034 .843 frank and open- a c a a reserved and cautious -.532 -.266 -.692 -.259 .908 impersonal- a a personal relationship -.056 .251 .889 .022 .897 emotional-unemotional a situation .789a -.481 -.122 .197 .913 I know-I do not know a b a how to behave -.651 .587 -.360 -.254 .845 he/she will think less of b c me-will not think less .393 -.454 .637 .018 .631 he/she has a major effect- does not have a major a a a effect on my life .511 -.046 -.759 .020 .936 I have a major effect- do not have a major b a a effect on his/her life .298 .140 -.859 -.O25 .910 I must be assertive- a b I need not be assertive .848a -.181 .703 -.145 .793 a p < b p < C p< .01 .05 .10 10. 11. 12. 213 TABLE 27 INDSCAL Coordinates for Form 2 (Four Dimensions) Situation 1. Persuade your boss to let you leave work early. 2. Persuade a group of people to go to the bar with you. 3. Persuade a professor that an answer on a test is incorrect. 4. Persuade a friend to go shopping with you. Persuade your boy/girlfriend to go to a party where there is mostly your friends. Persuade your boy/girlfriend not to date another person. . Persuade a friend to give you a ride. . Persuade a sister/brother to go to a fbur-year University. Persuade a boy/girlfriend to study more. Persuade a person to ask you out. Persuade a group of people to participate in an experiment for your group project. Persuade parents that marijuana is not bad. Dimension 1 —.451 -.340 -00‘46 ”02"1’0 -0018 .371 -.251 .475 .359 -0012 -.086 .240 [M .192 -.127 .434 —.152 -0288 -.055 '0168 .057 -.240 .019” -0162 .704 [w .0361". .296 -.322 .318 .310 .272 -.287 .028 |-F-' O 321 ‘0337 .222 -.400 -.059 .491 -.300 “0333 -.008 .342 -0038 .099 214 TABLE 28 Standardized Regression Weights for Four (INDSCAL) Dimensions Used to Predict Mean Ratings on Unidimensional Attributes (Form 2, n=67) Dimensions: Attributes 1 2 3 4 R2 friendly-unfriendly .288b .535a -.054 .503a .905a difficult-easy c c a a to resolve conflicts -.110 -.316 .296 .633 .877 short term- long term strategy .508 .361 .326 .361 .523 comfortable- a a a uncomfortable .083 .383 -.091 .752 .973 resistant-not c b b a resistant to persuasion -.257 -.450 .246 -.473 .881 easy to persuade- b a c a a difficult to persuade .358 .485 -.227 .406 .923 I benefit- a C b I do not benefit .734 .238 -.217 -.427 .773 I have been-I have never b been in this situation .637b .301 -.055 .279 .766 involves a favor- o a b involves a right .406 .694 .087 .180 .796 do something for his/her benefit-not do something for his/her benefit -.544 .275 -.188 .241 .428 I have relatively low a a b a status-high status .593 -.562 .230 -.400 .906 formal-informal .333b --.336b .459a -.588a .915a one totally dominates- b a a a a treat each other equally .365 -.521 .457 -.381 .925 a p < .01 p < .05 C p < .10 TABLE 28 (cont'd) 215 Dimensions: Attributes l 2 3 4 R2 superficial-intimate a b relationship .721 -.207 .583 -.l72 .820b no commitment to future b interaction-commitment .682 .033 .359 -.001 .500 frank and open- b b C a reserved and cautious -.404 .414 -.284 .595 .867a impersonal- a b personal relationship .708 -.095 .591 -.187 .739b emotional-unemotional a a a situation -.731 -.070 -.379 -.519 .948a I know-I do not know a c a how to behave .518 .277 .160 .598 .881a he/she will think less of b a me-will not think less -.172 -.438 -.188 -.707 .897a he/she has a major effect- does not have a major b effect on my life -.717 -.334 -.307 -.019 .610C I have a major effect- do not have a major a c b effect on his/her life -.822 -.O24 -.402 .156 .704 I must be assertive- c b I need not be assertive -.219 -.332 .473 -.270 .728 a p < .01 b C p < .05 p < .10 216 It appears that the first dimension corresponds to intimacy (superficial-intimate, 8 = .721; impersonal-personal, 8 = .706; no commitment-commitment to future interaction, 8 = .682; s/he has a major effect on my life-does not, 8 = -.7l7; and, I have a major effect-I do not, B = -.822). Thus, this dimensional array of situations can be labeled the "intimate vs. superficial" dimension. Only one attribute corresponds to the second dimension, and it cor- responds only at a moderate level: involves a favor-involves a right, 8 = .694. Thus, I shall refer to this dimensional array of situations as the "favors vs. rights" dimension. There are no attributes which even moderately correspond to the third dimension. Several attributes relate to the fourth dimension: comfortable-uncomfortable, B = .752; difficult-easy to resolve conflict, 8 = -.633; formal-informal, B = -.588; s/he will think less of me-will not think less of me, 8 = -.707; and, I know-I do not know how to behave, 8 = .596. I shall refer to this dimensional array of situations as the "comfortable vs. uncomfortable" dimension. The status/formality attributes generally correspond with more than two of the dimensions. Although the process of labeling these dimensions is extremely tenuous, the following labels will be provided: I. superficial vs. intimate; II. favors vs. rights (with overtones of friendly vs. unfriendly); III. unlabeled; IV. comfortable vs. uncomfortable (with overtones of friendly-easy to persuade vs. unfriendly-difficult to persuade). Table 29 presents the coordinate loadings of the twelve situations in Form 3 in the four-dimensional INDSCAL solution. Table 30 presents the standardized regression weights for the four INDSCAL dimensions used to predict to mean ratings on the unidimensional attributes. 217 TABLE 29 INDSCAL Coordinates for Form 3 (Four Dimensions) Situation 1. 10. 11. 12. Persuade a boy/girlfriend that there is nothing wrong in occa- sionally dating another person. Persuade your roommate to keep quiet when you are trying to sleep. Persuade your boy/girlfriend to confide in you more. Persuade a professor that a test item was ambiguous. Persuade a professor that his grading scale is too hard. Persuade a person you know in class to help you with your homework. Persuade your landlord to fix the plumbing in your apartment. Persuade an old boy/girlfriend that the relationship is worth saving. Persuade your father into getting a checkup. Persuade a police officer not to give you a ticket. Persuade a professor to increase your grade. Persuade your mother to go back to college and get a degree. Dimension 1 .250 -.215 .116 - 0130 -.O93 -0258 -.379 .126 0:458 —.256 -0182 .563 -.388 -0210 .262 “.392 .178 .384 .230 .182 .360 -01472 .026 .304 .297 -.276 -.273 .196 .172 .261 -0280 [42’ .365 - 013]. .229 .397 .122 .155 -.236 -.O22 218 TABLE 30 Standardized Regression Weights for Four (INDSCAL) Dimensions Used to Predict Mean Ratings on Unidimensional Attributes (Form 3, n=66) . Dimensions: Attrlbutes 1 2 3 4 R2 . . b b b friendly-unfriendly -.494 .260 .539 -.l4l .774 difficult-easy a a a to resolve conflicts .352 -.589 -.566 -.051 .867 short term- a a long term strategy .658 -.255 -.001 .293 .841 comfortable-uncomfortable -.453b .241 .6783 .056 .786b resistant-not a a a resistant to persuasion -.112 -.653 -.720 .029 .912 easy to persuade- a a difficult to persuade .144 .699 .728 .059 .960 I benefit- a a I do not benefit .939 .169 -.l3l .147 .971 I have been-I have never a been in this situation .046 .747 .032 .602 .879 involves a favor- . involves a right -.233 .039 .133 -.O23 .081 do something for his/her benefit-not do something a b b a for his/her benefit -.748 .066 .310 -.282 .934 I have relatively low a b a status-high status .220 -.647 -.493 .104 .852 formal-informal .383a -.saga -.610a .033 .985a one totally dominates- a b a treat each other equally .138 -.792 -.464 .066 .960 a p < .01 p < .05 C p < .10 TABLE 30 (cont'd.) . Dimensions: Attrlbutes 1 2 3 4 R2 superficial-intimate a a a a relationship .605 -.515 -.247 .168 .997 no commitment to future b a c a interaction-commitment .558 -.320 -.421 .245 .927 frank and open- a a a reserved and cautious -.408 .558 .572 .052 .893 impersonal- a a b a personal relationship .619 -.506 -.266 .141 .987 emotional-unemotional c b b a situation -.473 .296 -.402 -.378 .878 I know-I do not know a a a how to behave -.O73 .659 .688 .366 .885 he/she will think less of a c me-will not think less .502 .017 -.625 -.024 .685 he/she has a major effect- does not have a major a a effect on my life -.886 .155 .052 -.092 .973 I have a major effect- do not have a major a a a a a effect on his/her life -.734 .255 .216 -.254 .980 I must be assertive- a a a I need not be assertive .138 -.762 -.486 .017 .883 a p < .01 b p < .05 c p < .10 220 The first dimensional array of situations corresponds to the benefit items: I benefit-I do not benefit (8 = .939); do something for his/her benefit—not do something for his/her benefit (8 = -.748). This dimension also corresponds to a number of the superficial-intimate and influenca- bility attributes: s/he has a major effect on my life-does not have a major effect (8 = -.886); I have a major effect-I do not have a major effect on his/her life (8 = -.734); impersonal-personal, B = .619; superficial-intimate, 8 = .605; no commitment-commitment to future interaction, 8 = .558; and short term strategy-long term strategy, 8 = .658. Several of the superficial-intimate attributes also correspond to the second dimension (see Table 31). It is extremely difficult to label any of the remaining dimensional arrays of situations. Many of the friendly-unfriendly attributes correspond moderately with either the first and third dimensions, or the second and third dimensions. The same is true for a number of the "easy to persuade-difficult to persuade" set of attributes. Many of the status/formal set correspond to both the second and third dimensions. The attribute "I know how to behave-I do not know how to behave" also corresponds to two dimensions (two and four) and the attribute "involves a favor-involves a right" does not correspond to any of the four dimensions. However, for the purpose of labeling the dimensions, the following dimensions will be applied to the four INDSCAL dimensions (see Table 30): I. I benefit-intimate vs. I do not benefit-superficial; II. one dominates vs. treat each other as equals; III. comfortable-easy to persuade vs. uncomfortable-difficult to persuade; IV. unlabeled. Table 31 presents the coordinate loadings of the twelve situations in Form 4 in the four INDSCAL dimensions. Table 32 presents the 221 TABLE 31 INDSCAL Coordinates for Form 4 (Four Dimensions) Situation 1. Persuade a boy/girlfriend that there is nothing wrong in occa- sionally dating another person. 2. Persuade your roommate to keep quiet when you are trying to sleep. 3. Persuade your boy/girlfriend to confide in you more. 4. Persuade a professor that a test item was ambiguous 5. Persuade a professor that his grading scale is too hard. 6. Persuade a person you know in class to help you with your homework. 7. Persuade your landlord to fix the plumbing in your apartment. 8. Persuade an old boy/girlfriend- that the relationship is worth saving. 9. Persuade your father into getting a checkup. 10. Persuade a police officer not to give you a ticket. ll. Persuade a professor to increase your grade. 12. Persuade your mother to go back to college and get a degree. Dimension 1 .336 O 3u2 .201 .150 .174 . 144 .457 .280 . 3u5 .238 .086 .431 .2. .3. E. ‘0511 0207 -0048 .0111 ‘.’469 . -0058 -0u07 -0109 -0039 0298 0229 -0396 .268 .332 -.300 .012 -.452 -.480 0071 -00146 0360 -.441 .029 .183 .192 -.272 .433 .065 .336 .275 .285 .366 -.174 .278 -.192 .242 222 TABLE 32 Standardized Regression Weights for Four (INDSCAL) Dimensions Used to Predict Mean Ratings on Unidimensional Attributes (Form 4, n=66) Dimensions: Attributes 1 2 3 4 R2 . . b b b frlendly-unfrlendly -.545 -.013 .625 .065 .717 difficult-easy b c a a to resolve conflicts .399 -.211 -.792 -.l28 .948 short term- b c long term strategy .672 -.l60 .023 .259 .713 comfortable-uncomfortable -.43Ob -.O45 .836a .082 .894a resistant-not a a resistant to persuasion .149 -.212 -.863 -.215 .890 easy to persuade- b a difficult to persuade .084 .278 .903a .148 .942 I benefit- a I do not benefit .847a .335c —.l38 .252 .840 I have been-I have never c b been in this situation -.011 .288 .314 .819a .785 involves a favor- involves a right -.574 -.241 .310 .169 .376 do something for his/her benefit-not do something a b a for his/her benefit -.728 -.014 .307 -.293C .878 I have relatively low b a status-high status .392 -.395 -.8123 .019 .899 - a a a formal-informal .385 -.168 -.828 -.017 .987 one totally dominates- a treat each other equally .238c -.517a -.648a -.O42 .932 a p < .01 b p < .05 c p < .10 223 TABLE 32 (cont'd.) . Dimensions: Attributes 1 2 3 4 R2 superficial-intimate relationship .855a -.384a -.397a .038 .974 no commitment to future a b a interaction-commitment .599 -.297 -.453 .103 .906 frank and open- c c a reserved and cautious -.279 .261 .764 -.034 .869 impersonal- a a a personal relationship .676 -.348 -.383 .024 .967 emotional-unemotional a a a situation -.577 .523 -.392 -.l87 .954 I know-I do not know a a how to behave -.O41 .104 .827 .525 .898 he/she will think less of b a me—will not think less .401 .268 -.784 .174 .851 he/she has a major effect- does not have a major a effect on my life -.859 .072 .060 -.194 .927 I have a major effect- do not have a major a C b b effect on his/her life -.677 .232 .291 -.263 .938 I must be assertive- b c a I need not be assertive .352 -.260 -.759 -.l73 .888 224 standardized regression weights for the four INDSCAL dimensions used to predict to mean ratings on the unidimensional attributes. The first dimension is the "I benefit vs. I do not benefit" (I benefit-I do not benefit, 8 = .847; do something for his/her benefit-not do something for his/her benefit, 8 = -.728). I shall leave the second dimensional array of situations unlabeled (no attributes correspond to this dimension and it does not appear to relate to the intense feelings distinction). The third dimension corresponds to the "easy to persuade- difficult to persuade" (easy to persuade-difficult to persuade, 8 = .903; resistant-not resistant, B = -.863; comfortable-uncomfortable, B = .836). Several other attributes also correspond to this dimension: formal- informal, 8 -.828; and, I know how to behave-I do not know how to behave, 8 = .827. The fourth dimension corresponds with familiarity (I have been-I have never been in this situation, 8 = .819). The friendly- unfriendly set of attributes and status set of attributes correspond to dimensions one and three or to dimensions two and three, respectively. Thus, the four dimensional INDSCAL solution for the Form 4 data can be labeled as follows: I. I benefit vs. I do not benefit; 11. unlabeled; III. easy to persuade-comfortable vs. difficult to persuade- uncomfortable; and IV. familiarity. Obviously, the process of labeling the INDSCAL dimensions has not been a rigorous one. Technically, the INDSCAL solutions should be rota- ted to congruence with the KYST solutions. However, this rotation was not performed for the purpose of the present study because the model pur- ports to provide dimensional solutions which need no rotation for inter- pretation (Carroll, 1972; Carroll and Chang, 1970). The fact that many 225 of the factors of attributes corresponded highly to respective orthogonal dimensions in the interpretation of the §X§T_solutions while the ortho- gonal factors obtained by the factor analysis usually corresponded only moderately with several of the INDSCAL solutions suggests that the INDSCAL solutions must be rotated in order to enhance interpretability. Undoubtedly, by employing only the INDSCAL model to the type of data collected for the purpose of this study, one would eventually label dimensions which would have little correspondence to the results of the factor analysis or to the REST or GALILEO solutions. One reason for this potential problem of a lack of dimensional "generality" is the operation of diverse points of views (individual differences). As Woelfel and Danes (1977) have noted, individuals who perceive three objects in a different order (even in a different §§§k_ order, I might add) could not have their exact point of view represented in an aggregated solution; i.e., no degree of shrinking or stretching of axes will represent both individual's initial matrix. Thus an assessment of points of view becomes a central methodological question. Not only is it of critical importance to demonstrate the ways in which individuals vary in their perceptions of situations, but now a critical methodologi- cal question arises concerning one of the basic assumptions of the INDSCAL model: Is there a common aggregated multidimensional configura- tion which can subsume the initial configuration of each respondent? Stated another way, the question becomes, do the same dimensions of perception exist for all respondents, such that individual differences merely take the form of stretching or shrinking of some set of dimen- sional axes? 226 In order to address the question of individual differences, corre- lations were computed between respondents in each of the four groups on the basis of dimensional salience weights. These correlations were utilized as input into Johnson's (1967) hierarchical clustering program. The results of this analysis revealed that there are diverse points of view: eight global clusters were obtained for the Form 1 data, nine for the Form 2 data, eight for the Form 3 data and ten for the Form 4 data. By "global cluster" I mean that the cluster did not merge with another cluster until the correlation was (approximately) equal to zero. The clusters varied in size (from five to twelve individuals) and in the level of strength of clustering (in "weak" clusters, individuals merged at a low correlation; in "strong" clusters, individuals merged at corre- lations of about .65 or better -often a correlation of about .8). It should also be noted that a number of dimensional weights were zero, and some were negative, across the four data sets. This observa- tion, along with the facts that (a) the INDSCAL solutions accounted for so little of the variance and (b) a large number of clusters of indivi- duals were obtained, strongly suggests that each of the aggregated INDSCAL solutions do not represent a common configuration for all indi- viduals, and that individual differences are not likely to be represented by the expansion or contraction of some set of dimensional axes. To provide some exposition of individual differences, I shall select two clusters of individuals from the Form 3 data and illustrate how diverse several of the points of view are. Cluster 4 (Form 3) is a fairly strong cluster (the last individual merged at r = .673). This cluster consists of ten individuals —-three males and seven females. Cluster six (Form 3) is another strong cluster of individuals (last 227 individual merging at r = .764), and consists of nine individuals (three males and six females). 3.7 Investigation of Individual Differences In order to assess the difference between the clusters of indivi— duals, each set of paired comparison data was analyzed using the §Z§I/ TORSCA program. Dimensional solutions for dimensions one through five were obtained and the stress values for these solutions indicated four dimensional solutions (stress values of 1.93% and 2.78%). Table 33 presents the coordinates of twelve situations of Form 3 in the four dimensional §§§T_solution for cluster four individuals. Table 34 presents the standardized regression weights obtained from regressing the coordinate loadings of the situations onto the median ratings on the unidimensional attributes. The median ratings were used because the sample sizes are so small that the mean ratings may not be very robust. Table 35 presents the coordinates of twelve situations of Form 3 in the four dimensional EXSI solution for cluster six individuals. Table 36 presents the results of the regression analysis. When rotated, the row and column correlations for these two configurations never exceeded .7. These configurations are extremely different. In the following discus- sion, the unrotated §Z§T_solutions will be employed. For cluster four individuals, the first dimensional array of situa- tions corresponds to the superficial-intimate set of attributes (superficial-intimate, B = -.883; no commitment to future interaction- commitment, B = -.690; impersonal-personal, B = -.876; also, frank and open-reserved and cautious, B = .743), the influence set of attributes (I have-I do not have a major effect on his/her life, 8 = .905; s/he 228 TABLE 33 Coordinates of 12 Situations for the Fourth Cluster of Respondents (Form 3) in Four Dimensional KYST Solution Dimensions: 1 2 3 4 1. Persuade a boy/girlfriend that there is nothing wrong in occa- sionally dating another person. -.904 .819 .038 .441 2. Persuade your roommate to keep quiet when you are trying to sleep. .599 -.769 .207 .406 3. Persuade your boy/girlfriend to confide in you more. -.420 .163 .719 .430 4. Persuade a professor that a test item was ambiguous. .511 .394 .437 .181 5. Persuade a professor that his grading scale is too hard. .479 .538 .284 .094 6. Persuade a person you know in class to help you with your homework. .223 -.289 .479 .692 7. Persuade your landlord to fix the plumbing in your apartment. .624 -.810 .019 .407 8. Persuade an old boy/girlfriend that the relationship is worth saving. -.464 .269 .747 .161 9. Persuade your father into getting a checkup. -.829 -.633 .441 .128 10. Persuade a police officer not to give you a ticket. .668 .355 .207 .106 ll. Persuade a professor to increase your grade. .502 .474 .233 .031 12. Persuade your mother to go back to college and get a degree. -.989 -.511 .592 .066 229 TABLE 34 Standardized Regression Weights for Four (KYST) Dimensions Used to Unidimensional Attributes (Form 3, Cluster 4) Predict Median Ratings on . Dimensions: Attributes 1 2 3 4 R2 . . a a c a friendly-unfriendly .712 .504 .310 .107 .869 difficult-easy to resolve conflicts -.462 -.453 -.382 .064 .568 short term- a a long term strategy -.858 -.l23 -.222 .220 .848 comfortable-uncomfortable .573b .607b .229 .026 750b resistant-not b b resistant to persuasion -.l70 -.531 -.698a .096 .808 easy to persuade- a b difficult to persuade .084 .625 .650 -.113 .8333 I benefit- a I do not benefit -.774 -.l74 .088 .167 .885c I have been-I have never been in this situation .069 -.221 .430 .072 244 involves a favor- involves a right .360 .093 .158 .032 .164 do something for his/her benefit-not do something a a for his/her benefit .948 .112 -.008 .238 .969 I have relatively low c status-high status -.527 -.411 -.337 .086 .569 formal-informal -.730a -.390b -.433a -.Oll .872a one totally dominates- c b c treat each other equally -.438 -.430C -.544 .061 .676 a p < .01 b p < .05 C p < .10 230 TABLE 34 (cont'd.) . Dimensions: Attrlbutes 1 2 3 4 R2 superficial-intimate a a a relationship -.883 -.l45 -.369 -.064 .942 no commitment to future a b b a interaction-commitment -.690 -.429 -.414 .077 .838 frank and open- a b b a reserved and cautious .743 .372 .429 -.098 .884 impersonal- a b a a personal relationship -.876 -.l9l -.388 -.077 .961 emotional-unemotional a b a situation .750 -.261 .229 .401 .845 I know-I do not know how to behave .251 .357 .035 —.154 .215 he/she will think less of me-will not think less -.272 -.517 -.l86 -.034 .376 he/she has a major effect- does not have a major a c a effect on my life .928 .088 -.056 .212 .917 I have a major effect- do not have a major a a c a effect on his/her life .905 .290 .118 .203 .959 I must be assertive- a b I need not be assertive -.302 -.159 -.753 -.154 .707 p< p< p< .01 .05 .10 231 TABLE 35 Coordinates of 12 Situations for the Sixth Cluster of Respondents (Form 3) in Four Dimensional KYST Solution Dimensions: 1 2 3 4 1. Persuade a boy/girlfriend that there is nothing wrong in occa- sionally dating another person. -l.088 -.367 -.350 —.357 2. Persuade your roommate to keep quiet when you are trying to sleep. -.O39 -.240 1.085 .041 3. Persuade your boy/girlfriend to confide in you more. -l.l28 -.265 .035 .169 4. Persuade a professor that a test item was ambiguous. .782 -.345 .093 -.l26 5. Persuade a professor that his grading scale is too hard. .711 -.085 .235 —.l23 6. Persuade a person you know in class to help you with your homework. -.059 -.279 -.189 .822 7. Persuade your landlord to fix the plumbing in your apartment. .631 -.252 -.378 .295 8. Persuade an old boy/girlfriend that the relationship is worth saving. -l.110 -.055 -.041 -.239 9. Persuade your father into getting a checkup. -.O29 .953 .217 -.O28 10. Persuade a police officer not to give you a ticket. .719 -.128 -.274 -.421 ll. Persuade a professor to increase your grade. .715 -.O78 -.243 -.l4O l2. Persuade your mother to go back to college and get a degree. -.106 1.142 -.191 .107 232 TABLE 36 Standardized Regression Weights for Four (KYST) Dimensions Used to Predict Median Ratings on Unidimensional Attributes (Form 3, Cluster 6) . Dimensions: Attributes l 2 3 4 R2 friendly-unfriendly .818a -.558a .022 .333 .805a difficult-easy a b to resolve conflicts -.692 .338 .185 .466 .881 short term— a b b long term strategy -.780 .407 .192 .044 .813 comfortable- a a uncomfortable .533 .529 .009 .601 .928a resistant-not resistant to persuasion -.848a .141 .261 .369 .944a easy to persuade- a difficult to persuade .821 .044 .354 .373 .942a I benefit- a b b I do not benefit -.O47 .934 .021 248 .938 I have been-I have never c been in this situation .527 .280 .349 .156 .503 involves a favor— involves a right .223 .239 .389 .198 .297 do something fOr his/her benefit-not do something a b b for his/her benefit .680 .488 .016 .153 .724 I have relatively low c status-high status -.452 .116 .081 .534 .510 formal-informal -.795a .372a .208b .372a .950a one totally dominates- a a treat each other equally -.886 .034 .136 .354 .930 a p < .01 b p < .05 c p < .10 TABLE 36 (cont'd.) 233 Dimensions: Attributes l 2 3 4 R2 superficial-intimate a a relationship -.895 .403 .099 -.022 .973 no commitment to future a c interaction-commitment -.734 .389 .045 .310 .789 frank and open- a reserved and cautious .810 .172 -.115 .225 .753 impersonal- a a personal relationship -.853 .442 .186 .097 .968 emotional-unemotional a b situation .888 .030 .052 .331 .902 I know-I do not know b how to behave .661 .096 -.057 .247 .510 he/she will think less of C c me..will not think less -.32'+ .1488 .108 .482 .538 he/she has a major effect- does not have a major a a effect on my life .660 .627 .018 .314 .928 I have a major effect- do not have a major a a effect on his/her life ~310 ~529 .084 .122 959 I must be assertive- a I need not be assertive -.813 .166 -.194 .074 .732 p < .05 23” has a major effect-does not have a major effect on my life, 8 = .928), the benefit set of attributes (I benefit-I do not benefit, 8 = -.774; do something-not do something for their benefit, 8 = .948) and the strategy preference attribute (short term-long term strategy, 8 = -.858). For cluster six individuals, the first dimensional array of situa- tions also corresponded to the superficial-intimate set of attributes (superficial-intimate, B = -.895; impersonal-personal, B = -.853; no commitment to future interaction-commitment, B = -.734; also frank and open-reserved and cautious, B = .810), several of the easy to persuade- difficult to persuade set of attributes (easy to persuade-difficult to persuade, B = .821; resistant—not resistant to persuasion, B = -.8u8; I know how to behave-I do not know how to behave, B = .661; also, emotional~unemotional, B = .888), strategy preference (short term-long term strategy, 8 = -.780), assertiveness (B = .810), and the two influence attributes (I have a major effect-I do not have a major effect on his/her life, 8 = .810; s/he does not have a major effect-does have a major effect on my life, 8 = .660). Also, two status attributes corres- pond to this dimension: formal-informal, B = -.795, and one totally dominates-treat each other as equals, 8 = -.886. While some similarities between the first dimensions for both clus- ters of individuals can be noted, there are several major differences between these two dimensions. For both dimensions, the situations deal- ing with police, landlords and professors load positive, but are ordered differently. The main differences, however, appear to be related to the relative levels of perceived intimacy of the relationships. For indivi— duals in cluster four, the situations dealing with one's father (situa- tion nine) and one's mother (situation twelve) have much higher loadings 235 on this dimension than do the situations dealing with boy/girlfriend relationships. The opposite is true for individuals in cluster six; situations dealing with boy/girlfriends are perceived as more intimate and personal than the situations dealing with parents, particularly the situations dealing with one's father (see Table 35). A second major difference between the two configurations deals with perceived levels of assertiveness. For the cluster four individuals, the third dimension corresponds to the attribute "I must be assertive-I need not be assertive" (B = -.753). But the first dimension of the cluster six individuals corresponds to the assertiveness attribute. Again, this difference appears to be related to differences in the perceptions of the father and mother situations. For cluster four individuals, situations which deal with parents, professors and police are perceived as entailing assertiveness (situations that load positive on the third dimension —-see Table 33). For cluster six individuals, situations dealing with police, professors and landlords entail some degree of assertiveness (situations which load positive on the first dimension -see Table 35), but situa» tions which deal with parents, particularly one's mother, are situations that do not appear as though they entail a need to be assertive. Apparently, cluster four individuals perceive situations dealing with parents as more intimate and personal and feel that they can (or should) be assertive in making these types of requests. Individuals in cluster six perceive the two situations dealing with parents as less intimate and personal than situations dealing with boy/girlfriends, and perceive that they cannot (or do not need to) be assertive when attempt- ing to gain compliance from parents. It may be the case that for indi- viduals whose parents are more distant, past persuasion tactics entailing 236 assertiveness have been counter-productive. These individuals, them, would learn to avoid being assertive in trying to get their parents to do something for their own good. Further, individuals who are engaged in a more personal and intimate relationship with their parents feel that they can be assertive (or need to be assertive) in order to gain the com— pliance of their parents in a task that entails a benefit for their parents. A third difference between these two points of View deals with the question of who benefits from the persuasion. The "I benefit-I do not benefit" attribute corresponds to the first dimension of the cluster four solution (B = -.77u), also the do "something for his/her benefit-not do something for his/her benefit" attribute (B = -.9u8). Situations with positive loadings on this dimension are situatins in which the respondent benefits from compliance; situations with negative loadings are situa- tions in which the other individual benefits. The situations involving police, professors, landlords and situations such as roommate(s) to keep quiet and a person to help you with your homework are situations in which the respondent benefits from the persuasion. Further, it is not surpris- ing that the situations which deal with one's father and mother entail a persuasion attempt for the other person's benefit. Individuals in clus- ter four also perceive the situations dealing with boy/girlfriends as an action from which the boy/girlfriend will benefit. For the cluster six individuals, the situations with negative load- ings on the second dimension are situations in which the respondents is perceived as obtaining benefits. Individuals in cluster six and cluster four generally agree as to who benefits from compliance for all situa- tions except the situations dealing with boy/girlfriend relationships. 237 Specifically, a major difference exists between the two points of view in that individuals in cluster six perceive situation eight (persuade an old boy/girlfriend that the relationship is worth saving) and situation one (persuade a boy/girlfriend that there is nothing wrong in occasion- ally dating another person) as having more to do with their person bene- fits than cluster four individuals. The second dimensional array of situations for the cluster four individuals appears as though it is an "intense feelings" dimension. I conclude this because the situation "persuade your boy/girlfriend that there is nothing wrong in occasionally dating another person" is a fairly intense interaction, as well as the situations dealing with pro- fessors and police. The other two boy/girlfriend situations are also sometwhat intense (they have small positive loadings on this dimension). The situations dealing with parents, roommates and the landlord are not intense interactions. Further, it may be the case that the fourth dimensional array of situations for cluster six individuals is also an "intense feelings" dimension. If this is true, then the situation "persuade a person you know in class to help you with your homework" entails the least intense interaction in the perceptions of this cluster of individuals, followed by the landlord situation, "persuade a boy/girlfriend to confide in you more", the two parent situations and the roommate situation. Situations involving intense feelings include situations dealing with professors, "persuade an old boy/girlfriend that there is nothing wrong in occasion- ally dating another person" and, particularly, "persuade a police officer not to give you a ticket". 238 Despite the fact that some dimensions are not interpretable, I shall label the dimensions as follows, for cluster four individuals: I. Superficial-(I benefit-short term strategy) vs. intimate-(I do not benefit-long term strategy); II. Intense Feelings; III. Assertive vs. no need to be assertive (with overtones of "easy to persuade-difficult to persuade"); and IV. Unlabeled. And, for cluster six individuals: I. Superficial-(difficult to persuade-short term strategy-assertive) vs. intimate-(easy to persuade-long term strategy-no need to be assertive); II. I benefit vs. I do not benefit; III. Unlabeled; and IV. Intense feelings. With as many different clusters of individuals as obtained in this study, it is possible to provide an endless list of comparisons among individuals. Obviously, there is a critical need to provide a parsi- monious procedure for accounting for individual differences in the per- ceptions of situations. As noted ealrier, one variable that may prove useful is the individual's level of autonomy or dependence. The autono- mous individual ought to shy away from situations in which s/he would be expressing a dependence upon others and ought to be assertive in situa- tions which require self-reliance. The comparison of several clusters of individuals noted above sug- gests that the quality of one's relationships with parents will infouence the relative location of parent situations on the "superficial-intimate" dimension. Other differences obtained above support the notion of a useful differentiation of respondents on an autonomy type of measure; if one can assume that an autonomy measure relates to avoidance of subordi- nate relationships, the desire to date others (which they perceive as a personal benefit) and higher levels of assertiveness in dealing with 239 others who may have higher relative status. Aside from differences in perceptions due to relative levels of intimacy of parental situations, differences between individuals in situation familiarity and individual differences due to autonomy (possibly dominance), it is difficult to sug- gest person variables that would be centrally related to individual dif- ferences in the perception of situations. As Block (1968) noted, there is a critical need to integrate environmental factors with personality factors and there is considerable additional research needed in order to provide an understanding of how individuals perceive the behavioral requirements of situations, and what accounts for individual differences in such perceptions. SUMMARY As noted, several dimensions of persuasion situations were obtained. Of these, three dimensions were found to be replicated: "superficial vs. intimate", "easy to persuade vs“ difficult to persuade" and "intense feelings". Future research ought to differentiate the concept of "benefits" in a more articulate manner and employ attributes that better assess intensity of feelings. I strongly recommend that one should resist the temptation of claim- ing any sort of superiority for either the overall factor analytic approach or the multidimensional scaling approach. The overall factor analytic solution indicated the degree to which attributes form into fac- tors across all three sets of situations. The multidimensional scaling solutions indicate the degree of correspondence among the attributes within each of the four sets of individuals and three sets of situations. Further, only by employing the multidimensional scaling analysis would 2'40 "intensity of feelings" factor have been uncovered. Thus, when both techniques are used jointly, any factors which may be missed by the fac- tor analytic approach can be identified by the multidimensional scaling approach, while at the same time one can obtain an indication of the attribute factors across the various sets of situations. Obviously, future research in this area could certainly benefit from the use of both techniques jointly. It should be noted that the process of randomly sampling situations from a population of situations does not exactly follow Stewart's (197R) recommendation for comparing factor analytic and multidimensional scaling solutions in order to assess the "generality" of factors. As noted, the overall factor analysis solution is based on the degree to which attri- butes are correlated across all three sets of situations and four sets of individuals, and it is of no surprise that correlations among attri- butes would be somewhat different within each of the sets of situations. Nonetheless, the fact that the KYST (TORSCA) and GALILEO solutions are so highly similar, plus the fact that the several orthogonal factors of attributes obtained via factor analysis clearly correspond with ortho- gonal dimensional arrays of situations, supports the claim that the dimensions are not artifacts. Further, the interpretation of the gig: dimensions and the interpretation of the clusters of situations were very compatible. In contrast, it was concluded that the dimensional solutions ob- tained by the INDSCAL model are not "general" because (a) innumerable clusters of individuals were obtained, (b) a number of individuals had zero (or even negative) dimensional weights, (c) the INDSCAL solutions accounted for less than half of the variance and the solutions of higher 241 dimensionality did not indicate any major gain in the amount of variance accounted for, and (d) the INDSCAL solutions were different than the solutions provided by GALILEO or EZSI: While the INDSCAL purports to provide dimensions that need no rotation to enhance interpretability, the results of this study suggest that the INDSCAL solutions do not provide "general" factors without rotation. Admittedly, the number of diverse points of view obtained in this study was quite surprising. The INDSCAL model was found by Wish, Deutsch and Kaplan (1976) to be able to account for individual differences by an expansion and contraction of the various dimensional axes. Further, Magnusson and Ekehammar (1975) found only three homogeneous subgroups of individuals on the basis of anxiety responses. Thus, it was initially assumed at the start of this study that either the INDSCAL model could account for the individual differences or that some small number of clus- ters of individuals could be identified. Clearly, a more parsimonious way of assessing individual differences is needed. Differentiation of individuals on the basis of the quality of interpersonal relationships or on the basis of autonomy or dominance may prove useful. The need to find a more parsimonious way of differentiating indivi- duals on the basis of perceptions of situations becomes critical if one considers that the procedure employed in this study to differentiate individuals (the cluster analysis of individuals on the basis of dimen- sional salience weights) is not only post hoc but prone to error in classifying individuals. That is, with diverse points of view operating, the INDSCAL solutions may provide dimensions that are artifactual and the dimensional salience weights are not likely to be reliable. Thus, there may be fewer homogeneous clusters of individuals than eight (as obtained 242 here) and the individuals comprising clusters four and six may be less representative of homogeneous points of view than other sets of indivi- duals. Other points of view, however, could only hazardously be presented here -nearly all individuals were of the same age, nearly all were Caucasian and academic majors are too diverse to encourage a clus- tering of individuals on the basis of interests. Further, while the male- female comparisons indicated that the locations of some situations could be attributed to sex, the obtained differences in male-female configura- tions do not represent major differences in perspective (not as major as that obtained in the comparison of the two clusters of individuals pre- sented above). Also, it was suggested that the "sex differences" ob- tained may not exist if some other variable were to be partialed out (i.e., autonomy or dependence-independence). Fortunately, some limit can be given concerning the diversity of individual preferences. By summing across situations, a factor analysis of persons was performed for each of the four data sets. For each solu- tion, only one factor had an eigenvalue greater than one; though the per- cent of variance accounted for by the four one-factor solutions was only between forty-two percent and forty-eight percent of the variance. While it may be possible to accept factors with eigenvalues less than unity in order to obtain factors of persons, this observation suggests that the individual differences in multidimensional scaling solutions probably reflect (a) different patterns in the way situations are arrayed on dimensions and/or (b) a difference along some dimension(s) not repre- sented in the attributes employed in this study. Some support for (a) can be noted in the comparison made above for cluster four and six indi- viduals; where the superficial relationship situations were ordered 243 somewhat differently along the "superficial vs. intimate" dimension; while the intimate relationship situations were ordered much differently. It would appear that the respondents do not vary so much in terms of the attributes provided them as they do in terms of relative location of situations in the configurations. 244 NOTES The GALILEO program was used in order to assess reliability because the program automatically provides stability information. Further, although many dimensions are used in this procedure, providing re- liability for all real dimensions in this text enables other research- ers using the metric multidimensional scaling program to compare their estimates of reliability to those presented here. Finally, it is quite safe to assume that reliability of the KYST solutions will be similar to those provided in the text. CHAPTER 4 DISCUSSION Admittedly, the study reported here is primarily exploratory in na- ture, though it is not without relevant methodological importance and not without theoretical significance. Further, because the process of con- structing and assigning meanings to situations and events is (a) a pro- cess that unquestionably occurs with Considerable frequency, (b) repre- sents a process not well understood, and (c) determines limits on the pre- diction of behavior, the suggestions concerning future research are quite innumerable. Thus, in this final section of the paper I shall discuss (1) the methodological importance and implications of the results, (2) theo- retical significance, (3) summary, and (4) future research. 4.1 Methodological Importance A number of conclusions concerning the utility of methodologies em- ployed in this study for future research have already been noted in the previous chapter. These will be summarized here. First, as noted above, the utility of both the multidimensional scaling approach and the factor analytic approach is strongly encouraged. Second, when diverse points of View are operative, the utility of the INDSCAL model is severely lacking because of possible artifactual dimensions. This occurrence makes the dimensional salience weights suspect. Ironically, the sole advantage of the INDSCAL model is the ability to analyze individual differences, yet I concluded that a more parsimonious procedure is necessary when individ- uals order stimuli differently along the dimensional axes (also, see Woel- fel and Danes, 1977). 245 246 While individual differences in the perceptions of situations are considerable, the aggregate solutions provided by §Z§I_(and GALILEO) were found to be general, non-artifactual, and compatible with the factor analysis solutions and the interpretations of the cluster analysis re- sults. The operation of diverse points of views, however, suggests that the aggregate solutions may not be generalizable across respondents. If it is true, however, that the respondents vary primarily in terms of or- dering situations differently along dimensions, then this lack of gener- alizability across respondents is only a question of degree of intimacy, friendliness, etc., and not necessarily involving major differences such as the employment of totally different dimensions of perceptions. Some comparisons concerning similarities between subsets of the aggregate solutions for the Form 3 data and the solutions obtained for two clusters of Form 3 individuals were discussed above. For both clus- ters of individuals, the "superficial vs. intimate" dimensions and the "intense feelings" dimension are replicated and, as in the solution for cluster four individuals, the third dimension corresponds with both the assertiveness attribute and the "easy to persuade vs. difficult to per- suade" factor of attributes. Thus, three of the feur dimensions are rep- licated in one subcluster and two in another subcluster of individuals. However, the cluster four individuals perceive the need to be assertive somewhat differently than the aggregate, and the cluster six individuals apparently perceive the benefits attribute much differently than the aggregate (it is the only attribute that corresponds to the second dimen- sion for the cluster 6 individuals). Nonetheless, these observations suggest, at the least, that those dimensions found replicated across the four data sets appear to have some generalizability across persons. 247 Perhaps the most important aspect of the approach undertaken in this study is the utilization of random sampling of stimuli from a (pretested) population. Employing this tactic made it possible to replicate factors, uncover generalizable factors and claim, with a good deal of certainty, that a dimension (intense feelings) is not an artifact (which may be the case if only one set of situations were employed), even when it was not obtained in the factor analysis solution or in the Form 1 solution. Thus, a major distinction in the perception of situation missed by employing factor analysis was uncovered. Finally, a comment is in order concerning the utility of the assumed ratio and assumed interval scales. In the dimensional structure compari- son, one situation was found to be located differently in the eight GALILEO dimensions, yet the first four dimensions corresponded fairly well and three of the four (EZSI) dimensions were interpreted as the same dimensions. In the cluster analysis (of situations), only one situation clustered into a different cluster of situations, and it was either the last situation to merge into a strong (first) cluster (Form 4) or in the last cluster to merge (Form 3). Note that both of these differences in solutions are equally likely to occur because of a true difference in per- spective across the two samples of respondents, as opposed to any super- iority of the assumed ratio scale (i.e., precision), or any liability of the assumed ratio scale (i.e., increased unreliability). The third observed difference between the Form 3 and Form 4 data is that the clustering of individuals on the basis of dimensional salience weights resulted in eight clusters of individuals for Form 3 data and ten clusters of individuals for the Form 4 data. None of these three differ- ences between dimensions, clusters of situations or clusters of 248 individuals can be construed as representative of major differences that result from the type of scale employed. Both types of scales are equally likely to provide reliable dimensions of situations. 4.2 Theoretical Significance I concluded Chapter 1 by noting that there are valid reasons for why individuals should not be "consistent" across situations, even situations that an experimenter may perceive as very similar in nature, and that the categorization of the individual as "inconsistent" is warranted only under those instances in which the individual perceives situations as functionally equivalent and yet behaves differently. In essence, while the interactionists argue that behavior is a function of the person and the situation, my contention is simply that behavior is a function of the person and the person's construction of the situation. First, and most obviously, the results of this study suggest that it is quite impossible to assume that any set of situations is functionally equivalent for a set of respondents. Situations within a general cate- gory of intimate relationships-commitment to future interaction vary con- siderably in terms of levels of friendliness, resistance to persuasion, who benefits (and types of benefits), levels of assertiveness, etc. Fur- ther, individuals vary in terms of relative levels of these variables. It is quite apparent that future research needs to integrate personality factors and environmental factors. I must concede the argument that the results of this study do not shed any light on what personality or disposition types of measures can be used to parsimoniously account for individual differences in the per~ ceptions of these situations. During the initial stages of the study, it was implicitly assumed that differences among the respondents could be 249 identified on the basis of dimensional salience weights (i.e., the "friendly" individuals would have higher weights on the "friendly vs. unfriendly" dimension, etc.) or that a small number of clusters of indi- viduals would be uncovered. Obviously, alternative strategies must be found. First, it is likely that disposition measures would relate to major differences in the content of the individual's social environment and different frequencies of occurrence of some situations as well as differ- ent behavior responses in many situations. Therefore, one can first differentiate individuals on a disposition measure (friendly, anxiety), pretest situations and uncover the differences in both the content of situations inclusive of social environments as well as the difference in dimensional structures. This approach is similar to Forgas (1976), but differentiates individuals on the basis of dispositions and not in terms of occupations. I have already noted that an autonomy (or dependence-independence) measure may prove quite useful in differentiating individuals in a study using the approach employed in this paper. Alternatively, it seems that the claim that some individuals who are "situationally bound" (Mischel, 1968) or high in "role variability" (Block, 1961) are likely to have di- verse social environments and also differentiate between situations more than consistent individuals. These variables are only suggested here, though the work on integrating personality factors with relevant aspects of the individual's environment is only beginning. A considerable amount of attention has been given to the question of whether trait measures have predictive utility. Endler and his colleagues (see Chapter 1) argue that the interactions of persons and 250 situations spell doom for the omnibus inventories. While such a conclu- sion seems unwarranted, especially given the review in Chapter 1, it should be noted that trait measures are possibly much more useful than assumed (and tested) by Endler. Given that trait measures can predict the occurrence of behavior in some instance at a correlational level of .33 is quite encouraging in the face of possible individual differences in the perceptions of situations (and differential behavior-outcome expec- tancies). Further, if the effects of idiosyncratic differences in situa- tion perception are attenuated by averaging across observations, then the trait measure is likely to be very useful (c.f., McGowan and Gormly, 1976). Further, across the same type of situation over time, instructors' ratings of verbal performance and ratings of predisposition toward verbal behavior are quite high (Mortenson and Arnston, 1974). Finally, if dis- positions are related to differences in the content, scope and dimension- al structure (as noted in the above paragraph) of social environments, then a full exposition of the importance of the disposition and the glo- bal tendency to behave in certain ways in relation to the individual's daily functioning can be actualized. In terms of the ultimate question of the prediction of behavior, then, it seems intuitively obvious that trait measures alone cannot pre- dict behavior in some instance. This point is readily conceded (also, see McGowan and Gormly, 1976). However, the analysis of person, mode of response and situational requirements (let alone the individual's percep- tion of situational requirements) have not been fully provided in any re- search program in such a way as to lead to improvements of prediction. Further, the prediction of behavior is only one type of prediction that involves the trait measure. Trait measures may be useful in predicting 251 whether or not the individual engages in an activity, the perceptions of the activity and, finally, the mode of response exhibited in the activity. Of course, the reciprocal nature of engaging in some activities with the consequences on personality (such as self concept; see Miller and Steinberg, 1975) poses substantive problems for arguing any casual prior- ity. That is, the friendly individual has engaged in more friendly in- teractions and will undoubtedly engage in the same in the future. To the extent that future potentially hostile events are avoidable, the friendly individual's tendency to avoid the event should be predictable The "sit- uational constraints" noted by Goldberg (1972a) in this regard to the per- ceived level of profit involved as a consequence of such engagement. Thus, the apparent limited utility of trait measures to predict behavior is in relation to only one of the various ways in which trait measures - have utility. 4.3 Summary and implications of the dimensions of situations A number of dimensions obtained in the EZSI analyses can be unques- tionably judged as replicated and generalizable dimensions: intense feel- ings, superficial vs. intimate relationship, and friendly-easy to per- suade vs. unfriendly-difficult to persuade. The fact that the above three dimensions were replicated across several sets of randomly sampled sets of situations means that these three characteristics of situations are generally regarded as critically important criteria in the perceptions of situations. Of course, less important criteria may also be used to conveniently differentiate these situations, but it may prove useful first to assess how situations precalibrated on these dimensions vari- ably influence message strategies and modes of responses. Further, for other sets of randomly sampled situations, other criteria may be perceived 252 as critical. This occurred in the Form 1 solution, where the subordi- nate vs. social affiliation characteristic of the relationship and the perceived importance of the goal were apparently judged as important cri— teria. It should go without saying that the results of this study should not be generalized to a wide range of persons. Indeed, generalizability to a wide range of situations is the key focus of the present paper, and to accomplish such a task some limitation of the generalizability to per- sons was inevitable. With this in mind, and in conjunction with the above observation that for different sets of randomly sampled situations differ— ent characteristics vary in importance (or, in applicability to a set of situations) it must be strongly and unequivocally stated that any re- search seeking to analyze modes of responses across situations must pro— vide an assessment of the perceptions of situations for the particular set of sampled situations for the particular set of sampled respondents. Hopefully, the present study will be informative of how this can most productively be accomplished. 4.4 Future Research The directions of future research are quite numerous, and only a few general directions will be noted here. In terms of the specific do- main of persuasion situations, replications of this study ought to inves- tigate more fully aspects of situations which have bearing on the inten- sity of feelings and to differentiate the type of benefits operative in the various types of persuasion attempts. Across the four sets of data, dimensions of relationship and feelings/behaviors were replicated. The absence of goal or work related types of dimensions, however, is not sur- prising given that so many of the situations obtained in the pretests 253 dealt with social encounters. However, the "personally important goal vs. not personally important goal" dimension obtained for the Form 1 sol- ution indicates that considerations of the consequence of the interaction may indeed result in a separate dimension, though this dimension was ap- parently not replicated across the sets of situations. Nonetheless, several attributes which deal with the importance of the goals, etc., could lend support to the labeling of this dimension and possibly prove useful if solutions of higher dimensionality were obtained. In comparing the replicated dimensions in this study with other solutions obtained by the perception approach, the "intimate vs. super- ficial" dimension obtained here replicated the "intimacy" dimension ob- tained by Forgas (1976), the "social interaction-close friends" dimen- sion obtained by Magnusson (1971), and the "formal and open vs. reserved and cautious" dimension obtained by Wish (1975). Further, the "easy to persuade vs. difficult to persuade" dimension obtained here, and often corresponding to a "friendly vs. unfriendly" distinction, roughly cor— responds to the "friendly vs. unfriendly" dimension obtained by Wish (1975) and Wish, Deutsch, and Kaplan (1976). Similarly, the "intense feelings" dimension also replicated a distinction obtained by Wish (1975) and Wish et al. Thus, some commonalities appear to exist between the domain of persuasion situations and the more general category of social behavior situations. While several distinctions are specific to the domain of persuasion (i.e., favors vs. rights and benefits), at least these three general dimensions of situations can serve as a starting basis for future research concerned with other types of situations. The relationship between situations as they are precalibrated on the various dimensions and behaviors (i.e., compliance gaining strategies) 254 is another main avenue for further research. However, it should be noted that the difference in moving from one situation located at one location in the dimensional structure to another situation may lead to no change in strategy selection or lead to a large number of diverse strategies. For instance, it may be the case that only a few types of strategies are employed when engaged in superficial relationships while diverse types of strategies are obtained when ehgaged in intimate relationships with (a) strategies involving boy/girlfriend and parent relationships over- lapping, and (b) strategy selection dealing with intimate relationships being strongly influenced by other features of the situations (resist- ance, benefits, to mention two). Thus, situations which represent ext- remes on some dimensions may result in different strategy selection, while different strategies may be selected within some categories of sit- uations. Further, the most useful extension of this research would be to investigate how disposition measures correspond to the individual's soc- ial environment and to strategy selection. A content analysis of strate- gies employed for both the dependent and autonomous individuals (to use autonomy as a measure) according to respective features of social envir- onments would indicate how the types of individuals both structure their social environments as well as indicate the relevant strategies used to influence others in their immediate social worlds. 4.4 Summary Three sets of situations randomly sampled from a pretested set of situations comprising a population of persuasion situations were analyzed by three multidimensional scaling programs, and four dimensional solu- tions were obtained. Attributes used to interpret the dimensional . ‘- 255 solutions were factor analyzed. While six factors were obtained in the factor analysis solution, the degree of correspondence between orthogon- al classes of attributes for the various four dimensional solutions with orthogonal factors provided by the factor analysis, the high degree of correspondence between the dimensional solutions for two of the multi- dimensional scaling analyses and the compatibility of interpretation of the dimensional solutions with cluster analysis solutions suggests that the various four dimensional solutions obtained are not artifacts. Further, the replication of the "intense feelings" dimension suggests that a major distinction was not obtained in the factor analysis solu- tion. Finally, the INDSCAL model was not found to be a convenient way of uncovering individual differences. Replications of this work should (a) investigate the notion of in- tensity of feelings more fully, and (b) differentiate the global notion of benefits. Extensions of this work include (a) investigating a wider domain of situations-~in which case several of the replicated dimensions in this study will prove useful-~and, (b) investigate the correspondence of behavior or strategy selection along precalibrated dimensions. Fur- ther, a more parsiminious means of accounting for individual differences is needed. Thus, it was suggested that the global dispositions, on which omnibus inventories purport to reliably differentiate individuals may be useful in not only predicting the average of preformance across situa- tions but also may relate to differences in the content of social envir- onments as well as in the structures used to categorize social environ- ments . APPENDI CBS APPENDIX A INITIAL LIST OF SITUATIONS 10. ll. 12. l3. 14. 15. 16. 17. 18. 19. 20. 21. Persuade your Persuade your Persuade your Persuade your Persuade your 256 APPENDIX A INITIAL LIST OF SITUATIONS parents to let you stay out late. parents to let you get your own phone and car at 16. parents to let you bring a car to campus for awhile. parents to send you to Florida for the break. brother/sister to attend a state university rather than a community college. Persuade your it was parked Persuade your Persuade your Persuade your Persuade your Persuade your Persuade your Persuade your love to go. Persuade your father into believing that someone hit his car when when actually you sideswiped another car while drunk. brother/sister to drive you home for the weekend. brother to mow the lawn. parents not to punish you. parents to give you money. parents to visit you for a weekend. sister to stop seeing a divorced man you don't like. parents to take a vacation someplace that you would family not to go someplace for vacationing that you do not want to go. Persuade your brother/sister to go to a high school dance because you want him/her to have fun. Persuade your Persuade your sure you will Persuade your parents to let you go to a college in a warmer climate. parents to build a darkroom for you, when they are never use it. parents to buy you clothes. Persuade one of your parents to stop drinking. Persuade one of your parents to stop smoking. Persuade your parents to sign a lease for you. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. an. 45. Persuade your Persuade your Persuade your Persuade your Persuade your has something Persuade your Persuade your Persuade your 257 boy/girlfriend to tell you "frat" (sorority) secrets. boy/girlfriend to quit smoking. boyfriend to shave or cut his hair. boyfriend to grow a beard. boy/girlfriend to go somewhere with you when he/she important to do. boy/girlfriend to go to the store for you. fiancee to meet your ex-boy/girlfriend. fiancee to go on a trip with you. Persuade an old boy/girlfriend that the relationship is worth saving. Persuade your Persuade your girl. Persuade your Persuade your boy/girlfriend to live with you. boy/girlfriend of 2 years to stop seeing anOther boy/ boy/girlfriend that your relationship should end. boy/girlfriend to stay in school. Persuade a friend to let you use their car so that you can go on a date. Persuade a friend to go shopping with you. Persuade a friend to switch presents with you. Persuade a friend to approve of your buying an item that they are buying (the same dress or shirt). Persuade a friend to give you a ride. Persuade a friend to buy you something. Persuade a friend to take a class with you. Persuade a friend to drive more carefully. Persuade a friend to come over and visit. Persuade a friend not to go to a party because you don't want to drive the twenty miles to get there. Persuade your friends not to study because you feel like partying. 46. 47. 48. L‘9. 50. 51. S2. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 258 Persuade a friend not to commit suicide. Persuade a friend not to criticize and make fun of another person. Persuade a friend to lend you a large sum of money. Persuade your friends to use some kind of birth control. Persuade a date to make sexual advances. Persuade a date not to make sexual advances. Persuade your friend to have a fight with someone else. Persuade a friend to go swimming with you. Persuade your roommates to go somewhere so that you can be alone. Persuade your roommate to go to Florida during the break with you. Persuade your roommate not to have his/her boy/girlfriend and all of her/his friends over every weekend. Persuade your roommate to move to a nicer apartment. Persuade your roommate, who is constantly depressed and having problems, to seek professional help. Persuade your roommate, who owes you 80 dollars, to pay you. Persuade your roommates to do their share of the housework. Persuade your roommate, who you car pool with, to be on time more often. Persuade your roommate to keep quiet when you're trying to sleep. Persuade your neighbors to keep their stereo turned down. Persuade your boss to let you leave work early. Persuade a cop not to give you a ticket. Persuade a store manager to sell you something for less than the regular price. Persuade a professor to give you an extension on a project. Persuade a boss to let you take a leave of absence to attend school so that you would have employment next summer. Persuade a boss to give you a raise. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 91. 259 Persuade a professor that an answer on a test is incorrect. Persuade your landlord to fix the plumbing in your apartment. Persuade a professor to let you enroll in an already crowded class. Persuade a doctor to let you go home from the hospital. Persuade the manager of a clothing store to hire you. Persuade a judge not to suspend your drivers license. Persuade a group of people to participate in an experiment for your group project. Persuade a person to ask you out. Persuade a person to sell you a car for less than they are asking, when you know you can sell for more than they are asking. Persuade a group of people to go to the bar with you. Persuade a group of people to drive someplace, instead of flying, so that you can afford to go with them. Persuade a group of people to order a pizza because you feel like eating one. Persuade a person you know in your class to help you with your homework. Persuade a person you don't know well in your class to study with you for the final. Persuade a girl/boy not to go to the movies, but to go drinking and to bed. Persuade an "I.D. checker" to let you into the I.M. when you forgot your I.D. Persuade someone who has consumed too much alcohol to get someone else to drive them home. Persuade someone to marry you. Persuade someone to sign a petition. Persuade someone not to rob you. Persuade someone to be a reference for you. Persuade a group of people to stop teasing you. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. Persuade Persuade Persuade Persuade Persuade stead of Persuade pression Persuade over the Persuade Persuade 260 a group of people to stop teasing a friend of yours. a customer to buy a certain item. your husband to take you to Hawaii. your child to get "A"s in school. your parents to lave your brother in jail for awhile in- bailing him out immediately. your fiance to cut his hair in order to make a good im- on your parents. your roommate to go to New England (rather than Florida) Spring break. a friend to believe something that is not true. a friend to wear something because you think they look good in it. Persuade dress or Persuade Persuade modeling Persuade Persuade theirs. Persuade a friend to let you buy an item that you both want (same shirt). a friend not to wear something because you want to wear it. a photographer to compile a portfolio of you in return for at your convenience. a principle not to expel you from school. someone that your solution to a problem is better than a professor to excuse you from class. APPENDIX B INSTRUCTIONS FOR SORTING TASK PRETEST 261 APPENDIX B INSTRUCTIONS FOR SORTING TASK PRETEST Before you there is a deck of cards. On each card you will find a description of a situation. There are forty situations described on forty cards. Your task is to sort them into categories. First, read through the cards in order to obtain an idea as to the kind of situations you will be sorting today. Place those situations which are similar to each other in the same pile (or category). Place those situations which are NOT SIMILAR in different piles. After you have sorted all the situations into various piles, read through the situations you have sorted into each pile to check if the situations in each pile are all similar to each other and different from the situations in other piles. FEEL FREE TO MOVE SITUATIONS FROM ONE FILE TO ANOTHER UNTIL ALL SITUATIONS IN EACH PILE ARE SIMILAR TO EACH OTHER. You are free to determine what constitutes "similarity" between the situations on your own. When you feel confident that the situations in each pile are similar to each other and different from those in other piles, you are finished. Tell the researcher that you are finished. I wish to ask you a few ques- tions about how you sorted the situations. If you have any questions, do not hesitate to ask them. Please begin. APPENDIX C LIST OF NOMINATED SITUATIONS 10. ll. 12. 13. 14. 15. 16. 17. 18. 19. 262 APPENDIX C LIST OF NOMINATED SITUATIONS Persuade your parents to let you stay out late. (does not apply, 32.5%) Persuade your parents into giving you money. (does not apply, 10%) Persuade your parents to visit you for a weekend. (does not apply, 25%) Persuade your parents to take a vacation someplace that you would like to go. (does not apply, 15%) Persuade your parents into buying clothes for you. (does not apply, 20%) Persuade your parents not to punish you. (does not apply, 25%) Persuade your boy/girlfriend to go somewhere with you when he/she has something important to do. (does not apply, l5%) Persuade your boy/girlfriend to go to the store with you. (does not apply, 15%) Persuade an old boy/girlfriend that the relationship is worth saving. (does not apply, 25%) Persuade your boy/girlfriend to stay in school. (does not apply, 32.5%) Persuade your boy/girlfriend to quit smoking. (does not apply, 40%) Persuade a friend to go shopping with you. (does not apply, 2.5%) Persuade a friend to give you a ride. (does not apply, 10%) Persuade a friend to take a class with you. (does not apply, 15%) Persuade a friend to drive more carefully. (does not apply, 7.5%) Persuade a friend to come over and visit. (does not apply, 10%) Persuade your friends not to study because you feel like partying. (does not apply, 37.5%) Persuade a friend not to criticize and make fun of another person. (does not apply, 5%) Persuade your friends to use some kind of birth control. (does not apply, 25%) 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 263 Persuade a date to make sexual advances. (does not apply, 30%) Persuade a date not to make sexual advances. (does not apply, 30%) Persuade a friend to go swimming with you. (does not apply, 10%) Persuade your roommate(s) to go somewhere so that you can be alone. (does not apply, 28.95%) Persuade your roommate, who is constantly depressed and having prob- lems, to seek professional help. (does not apply, 28.95%) Persuade your roommate to move to a nicer apartment. (does not apply, 44.74%) Persuade your roommate(s) to do their share of the housework. (does not apply, 13.51%) Persuade your roommate to keep quiet when you are trying to sleep. (does not apply, 21.62%) Persuade your neighbors to keep their stereo turned down. (does not apply, 25%) Persuade a group of people to go to the bar with you. (does not apply, 15%) Persuade a group of people to participate in an experiment for your group project. (does not apply, 15%) Persuade a person to ask you out. (does not apply, 30%) Persuade a group of people to order a pizza because you feel like eating one. (does not apply, 2.5%) Persuade a person you know in your class to help you with your home- work. (does not apply, 15%) Persuade a person you don't know well in your class to study with you for the final. (does not apply, 27.5%) Persuade an "I.D. checker" to let you in the I.M. when you have for- gotten your I.D. (does not apply, 20%) Persuade someone who has consumed too much alcohol to get someone else to drive them home. (does not apply, 15%) Persuade someone to sign a petition. (does not apply, 17.5%) Persuade a professor to be a reference for you. (highest nominated for this item on pretest) 264 39. Persuade an employer to be a reference for you. (second highest nominated for this item on pretest) 40. Persuade a group of people to stop teasing you. (does not apply, 30%) 41. Persuade a group of people to stop teasing a friend of yours. (does not apply, 5%) 42. Persuade a customer to buy a certain item. (does not apply, 12.5%) 43. Persuade your boss to let you leave work early. (does not apply, 5%) 44. Persuade a cop not to give you a ticket. (does not apply, 12.5%) 45. Persuade a store manager to sell you something for less than the regular price. (does not apply, 15%) 46. Persuade a professor to give you an extension on a project. (does not apply, 2.5%) 47. Persuade a boss to let you take a leave of absence to attend school so that you would have employment next summer. (does not apply, 22.5%) 48. Persuade your boss to give you a raise. (does not apply, 17.5%) 49. Persuade a professor to let you enroll in an already crowded class. (does not apply, 15%) 50. Persuade a doctor to let you go home from the hospital. (does not apply, 20%) 51. Persuade a landlord to repair the plumbing in your apartment. (does not apply, 35%) 52. Persuade a professor that an answer on a test is incorrect. (does not apply, 10%) NOMINATED SITUATIONS ADDED AFTER THE NOMINATING SITUATIONS PRETEST A. Parents and Family (in order of frequency of nomination) l. Persuade your parents to approve of your moving off campus. 2. Persuade your mother into finding some way to spend her time (volun- teer work, a job, etc.). 3. Persuade your brother/sister to go to a four-year University. B. 265 Persuade your father into getting a checkup. Persuade your parents to take a real vacation by themselves (a sec- ond honeymoon). Persuade your mother into going back to college and getting a degree. Persuade your parents to financially support your education. Persuade your father into taking more time off from work and find something relaxing to do. Persuade your parents that marijuana is not bad. Boyfriends and Girlfriends (in order of frequency of nomination) 1. Persuade your boy/girlfriend that the relationship should end. Persuade your boy/girlfriend to confide in you more. Persuade your boy/girlfriend to have sex. Persuade your boy/girlfriend that there is nothing wrong in occa- sionally dating another person. Persuade your boy/girlfriend to go to a party where there is mostly your friends. Persuade your boy/girlfriend into meeting your relatives. Persuade your boy/girlfriend to study more. Persuade your boy/girlfriend to lend you money. Persuade your boy/girlfriend not to date another person. Persuade your boy/girlfriend that you only want to date. Friends (in order of frequency of nomination) l. 2. 3. u. Persuade a friend to study more. Persuade a friend to pay money back. Persuade a friend to diet. Persuade a friend not to swear so much. Persuade a friend to stop smoking. 266 Roommates (in order of frequency of nomination) Persuade your roommate to leave phone messages while you are out. Persuade your roommate to leave for the night. Persuade your ex-roommate to forward your mail. People (in order of frequency of nomination) Persuade someone in class to lend you notes. Persuade a bouncer to let you into a bar even though you have for- gotten your I.D. Professionals and Others (in order of frequency of nomination) Persuade a professor to increase your grade. Persuade a professor that a test item was ambiguous. Persuade a professor that his grading scale is too hard. Persuade a landlord to pay for paint, or some repairs, for your apartment (or house). APPENDIX D INSTRUCTIONS FOR ASSUMED INTERVAL SCALE 2657 PERCEPTIONS OP SITUATIONS. PART I. INSTRUCTIONS. One of the major ways of making comparisons between situations is to simply state whether two situations are similar to each other or difforent from each other. It is possible, then, to cluster together a set of situations which are similar to each other, and diffhrent from other sets of situations. How do you report situations as being similar or dissimilar? On the next several pages, you will find one situation presented on the left of the page and one on the right. Between the two you will find an eleven category scale, such as: Persuade a friend to loan / you money. Persuade a boy/ girlfriend to quit sacking. / Dissimilar If'you perceive these two situations as exactly alike, check the cate- gory marked 1. If you perceive the two Eituations as somewhat similar check one of the categories marked 2, 3, 4 or 5. If you perceive Ehe two situations as neither similar nor dissimilar, check the category marked 6. If you perceive them as somewhat dissimilar, check one of the categories marked 7, 8, 9 or 10. If you perceive t5; two situations to be totally diffbrent, check the category marked ll. Please be sure to answer every comparison. In order to use the scale effbctively, it would be helpful for you to know the situations I as providing you. Carefully read the follow- ing situations; they are the ones you will be comparing today. 1. Persuade a boy/girlfriend that there is nothing wrong in occasion- ally dating another person. 2. Persuade your root-ate to keep quiet when you are trying to sleep. 3. Persuade your boy/girlfriend to confide in you more. 4. Persuade a professor that a test item was ambiguous. S. Persuade a profhssor that his grading scale is too hard. 6. Persuade a person you know in class to help you with your homework. 7. Persuade your landlord to fix the plumbing in your apartment. 26E3 8. Persuade an old boy/girlfriend that the relationship is worth saving. 9. Persuade your father into getting a check-up. 10. Persuade a police officer not to give you a ticket. ll. Persuade a professor to increase your grade. 12. Persuade your mother to go back to college and get a degree. Remember, you can take as much time as you want. If you have questions, do not hesitate to ask them. Again, thank you. Michael Cody I.D. Col. l-e SEX Col. 7 FOR! Col. O-lo Form 300 APPENDIX E INSTRUCTIONS FOR ASSUMED RATIO SCALE 2639 PLEASE READ THESE INSTRUCTIONS CAREFULLY INSTRUCTIONS In this questionnaire, I would like you to tell me how different (or, in other words, "how far apart") certain situations are from each other. All of the situations in this questionnaire deal with persuading (or influencing) someone else to do something. Differences between these concepts can be mea- sured in Social Uhits. "Social Units" refer to the psychological distance or dissimilarity_be- tween situations. You may perceive the underlying characteristics of some situations to be extremely different, in which case they are "far apart." You may perceive the underlying characteristics of some situations to be exactly alike or very similar, in which case they are "very close" to each other. To aid in understanding what I mean by "distance," please use the following rule: The distance between PERSUADE A GROUP or PERSUADE YOUR BOY/GIRLFRIEND and PEOPLE TO GO THE BAR TO MEET YOUR RELATIVES --m NITH YOU is 100 units. I would like you to tell me how many Social Units apart situations are from «a: other. Mr, the more different two situatiF-ns are from each other, the bigger the number of SocIEI Units. The less different they are, the smaller the number of Social Units. The following can serve as an example. First, you are given the rule and then you are asked to report the distances between pairs of situations. For example: I? THE DISTANCE BETWEEN PERSUADE YOUR BOY/GIRLFRIEND TO MEET YOUR RELATIVES AND PERSUADE A GROUP OF PEOPLE TO GO TO THE BAR WITH YOU IS 100 UNITS, HOV FAR APART ARE: PERSUADE A PROFESSOR PERSUADE YOUR BOY/GIRLFRIEND THAT AN ANSWER ON A TO MEET YOUR RELATIVES and TEST IS INCORRECT units. PERSUADE YOUR BOY/GIRLFRIEND PERSUADE A FRIEND T0 TO MEET YOUR RELATIVES and GO SHOPPING NITH YOU units. PERSUADE A PROFESSOR THAT AN PERSUADE A FRIEND TO ANSWER ON A TEST IS INCORRECT and GO SHOPPING WITH YOU units. If you perceive "Persuade your boy/girlfriend to meet your relatives" as being very close to the meaning and to your perceptions of the situation "Persuade a professor that an answer on a testis incorrect," then you would write, in the space provided, a small number of Social units. If you perceive the two situations as 3352 distant from each other (that they involve different feelings, expectations, behaviors, etc.). then write a number that is very 270 INSTRUCTIONS , cont '6 Similarly, if you perceive the two situations, "Persuade your boy/girlfriend to meet your relatives" and "Persuade a friend to go shopping with you," as ve similar situations, then report a small number in the space provided. If the two situations are extremely different from each other, report a large number. FEEL FREE TO USE ANY NUMBER TO REPORT AS ACCURATEUY AS POSSIBLE THE DIS- TANCE THAT YOU SEE BETNEEN THE TWO SITUATIONS. You may use a number over 100 if the distance between any two situations is ter than the distance between the situations of "Persuade your boy/ girlggIend to meet your relatives" and "Persuade a group of people to go to the bar with you." If you perceive two situations as extremely similar report a number that is very small. If you think that there’is no difference in the meaning of two situations, then you may write zero (0) to represent no distance between them. On the following pages, you will find lists of pairs of situations similar to those shown above. Please write a number in the blank space after each pair of situations. Ignore the column of numbers next to the blanks; they are for clerical use only. Please try not to skip any item. Try to report some distance between each pair of situations. Keep in mind that there is no one correct answer; all that I ask is that you give an honest and careful response to each pair. It would be helpful to you to know what situations you will find in the questionnaire. The situations included in this questionnaire include; 1. Persuade your parents to approve of your moving off campus. 2. Persuade your parents to financially support your education. 3. Persuade a boy/girlfriend that there is nothing wrong in occasionally dating another person. u. Persuade your roommate to keep quiet when you are trying to sleep. 5. Persuade your boy/girlfriend to confide in you more. 6. Persuade a professor that a test item was ambiguous. 7. Persuade a professor that his grading scale is too hard. 8. Persuade a person you know in class to help you with your homework. 9. Persuade your landlord to fix the plumbing in your apartment. lO. Persuade an old boy/girlfriend that the relationship is worth saving. ll. Persuade your father into getting a check-up. . Persuade a police officer not to give you a ticket. 13. Persuade a professor to increase your grade. 1%. Persuade your mother to go back to college and get a degree. If you have any questions, please do not hesitate to ask me or one of my assistants. Thank you. Michael Cody Form uoo 271. PERCEPTIONS OF SITUATIONS. PART II. (cont'd) Note that the person checked the category marked a on the second scale, indicating he perceived that it is somewhat difficult to resolve conflicts with his/her roommate. Obviously, this is how this'person perceives his/ her roommate, and your responses may be very different. Note that when you persuade a person to do something, you can use a short term strategy or a long term strategy. For example, if you wanted to perw suade a group of people to go to the movies tonight, you would probably use a "short term" strategy. The strategy you employ would take place at one point in time-—you do not "campaign" over time to get your way. How- ever, if you wanted to persuade your parents to take a vacation somewhere you wanted to go, you.m§y_use a long term strategy in order to influence their decision. That is, you bring up the topic several times over several weeks (or months). In the example given above, the person checked the category marked 2, indicating that s/he is inclined to use a short term strategy. Another distinction between situations is whether or not you have status over the other individual(s). In the example given above, the person checked the category marked 6, indicating that neither s/he nor his/her roommate have higher status. Finally, you may be persuading a person to do you a favor, as in persuad- ing a roommate to give you a ride or to pick up some food from the grill. Some situations deal‘with your rights, or infringements of your rights--. as in persuading a friend to pay you back money they borrowed or'in per- suading a record company to stop sending you bills for records you never ordered or received. In the example given above, the person checked the category marked 11, indicating that s/he perceived the situation as one involving his/her rights. There are a large number of other scales provided in this questionnaire. Please read all of these scales carefully. Note that there is no correct answer to any of the responses on these scales--all I am asking for is your most honest and frank evaluation of each situation on all scales. Please, do not skip any items--try to give some answer for each scale. If you.haye any questions now, or during the completion of this question- naire, do not hesitate to ask them. Thank you. Michael Cody APPENDIX E ORDER BY WHICH ATTRIBUTES WERE PRESENTED 272 monstrous or SITUATIONS. PART II. FOR! 51 (In order to match your second questionnaire with your first questionnaire, please write your I.D. number here: . Remember, all answers are kept confidential.) INSTRUCTIONS Another way of making comparisons between situations is indirectly, through the use of rating scales. In this questionnaire, you will be asked to rate each situatia: on a meter of rating scales. For example, a situation will be given, followed by a manner of scales: PLEASE READ ALL SCALES CAREFULLY. ON THE SCALES PROVIDED BELON, PLEASE RATE YWR FEELING TOHARD THE FOLLOWING SITUATION: PERSUADE YOUR RMNNATE NOT TO EAT YOUR FOOD. Very friendly Very unfriendly situation] I I_I__I I __I I__I_I)_(_I Isituation . TTauTT Tasmfi Difficult to re- Easy to resolve solve conflicts conflicts with withthispersaal I I IXI I I- I I I I Ithisperson I"??? S-T'TF'TID'II Iwouldusea Iwouldusea short term long term strategy// X/I/____/II/_//st!etegy T's—TTTLO'E I have relative- I have relative- lyhighstatus/ I / I I IX/ I I I I __Ilylowstatus 0'1 vo'\ 0'): ‘l ’23“??? T39 onmissituation This situation involves one of involves ef vor/__ I__I_ oIx __Imy riate .1’6'J/93—T'5" In this (hypothetical) exanple, the person marked the space indicating that the situation was very close to being "very unfriendly." If sIhe perceived the situation as "very friendly," sIhe would have checked the category marked 1 (or 2). If sIhe had perceived the situation as less friendly, sIhe would have checked either category 3, It, or 5. If sIhe had perceived the situation as somewhat unfriendly, sIhe would have checked either category 7, 8, or 9. If sIhe perceived the situation as neither friendly nor unfriendly, sIhe could have checked the category mm Category 6 (six) can be used if the situatia: is neutral on the rating scale, or if the rating scale does not apply to the situation. 2753 PERCEPTIONS OE SITUATIONS. PART II. (cont'd) Note that the person checked the category marked u on the second scale, indicating he perceived that it is somewhat difficult to resolve conflicts with his/her roommate. Obviously, this is how this person perceives his/ her roommate, and your responses may be very different. Note that when you persuade a person to do something, you can use a short term strategy or a long term strategy. For example, if you wanted to per- suade a group of people to go to the movies tonight, you would probably use a "short term" strategy. The strategy you employ would take place at one point in time--you do not "campaign" over time to get your way. How- ever, if you wanted to persuade your parents to take a vacation somewhere you wanted to go, you use a long term strategy in order to influence their decision. That , you bring up the topic several times over several weeks (or months). In the example given above, the person checked the category marked 2, indicating that s/he is inclined to use a short term strategy. Another»distinction between situations is whether or not you hate status over the other individual(s). In the example given above, the person checked the category marked 6, indicating that neither sIhe nor his/her roommate have higher status. Finally, you may be persuading a person to do you a favor, as in persuad- ing a root-ate to give you a ride or to pick up some food from the grill. Some situations deal with your rights, or infringements of your rights--. as in persuading a friend to pay you ack money they borrowed or in per- suading a record company to stop sending you bills for records you never ordered or received. In the example given above, the person checked the category marked 11, indicating that sIhe perceived the situation as one involving his/her rights. There are a large number of other scales provided in this questionnaire. Please read all of these scales carefully. Note that there is no correct answer to any of the responses on these scales--all I am asking for is your most honest and frank evaluation of each situation on all scales. Please, do not skip any items-~try to give some answer for each scale. If you.have any questions now, or during the completion of this question- naire, do not hesitate to ask them. Thank you. Michael Cody 271+ PM]: READ ALL SCALES CAWY. ON mm SCALES PROVIDED BEL“, PLEASE TATE. Yfl-R FEELINGS TO‘JARD THE EOILONTIG SITUATION: PERSUADE A POY/GIRLERIDW THAT THERE IS NOTHING mono IN OCCASIONALLY‘DATINC ANOTHER PEfiON. Very friendlyl / I I I I I I I I "cry unfriendly TTTTTsTTTEE Difficult to re- Say to resolve solve cmflicts conflicts with with this person! I I I I I I I I I I Ithis person TTTTTTTTTEE I would use a I would we a short term 1mg term t t I I I I I I '"mr'rrr “rrrlrr’m Verycomfortablel I I I I I I I I I I I’Jerytmcomfortable TTTuTTTTTUE ize/she will be lie/she will not be very resistor to I very resistant to my pereuesion/ I I I I I I I I I lay persuasion TTTTTTTTTWE III I strategy Itwouldbevery It wouldbevery easy to persuade difficult to his/bal I I I I I I I I I Ipersuade him/her TTTTTTTTTEK I benefit from I do not benefit thispersuasion/ I I I I I I I I I I Ifruthispersuuion TTTTTTTTSTH I have been in this type of I have never been situatiu: many in this type of times before/ I I I I I I I I I I Isituation before TTTTTFTTTHH This situation This situation involves one of involvesafsvor/IIIII III/Imyri ts. TTTTTHTTTWE AL' I as persuading I am not persuading her/him to do her/him to do something for herI something for his own benefit! I I I__ I I I I I I Iher/his on benefit TTTT TTTTsEH I have relatively I have relative ly lumm/I III III/mwmm FTHTVTTTTTTH Very formal Very informal situation/ I I I I I I I I I I Isituation lrausevarfiu PLEASE MAKE SURE THAT YOU HAVE ANSHERED EVERY ITEM. THANK YOU. of M is for ' clerical 13.1u 15,16 17.18 '19.20 21.22 23.25 25.26 ‘ 27.23 29.30 my 3% 35.36 275 PLEASE READ ALL SCALE CARENLLY. ON THE SCALES PROVIDED DEW, PLEASE RATE 701R mums WARD THE WWII“; SITUATION: PERSUADE A BOYI GIRLFRIEND THAT THERE IS NOTHING HRDNG IN OCCASIO‘ALL‘! DATING ANOTHER “13011:t One totally don- ' heat each other lnates the other/ I I I I I I I I I I In equal TTTTTTTTWH Very superficial Very intiaete relationahip/ I I I I I I I___ I I I Irelationship TTTTTTTT TEE Ho coo-ituat to Cos-item to future interaction/__I I I__I I I I I I___/ Ifuture interact ion 1TTnTTTTTmE Reserved and Frank and open/ I I__I__I I I I I I I Icautious TTanTTTTTHH Very iapersonal Very personal relationahip/ I I I I I I I Irelationship .TTT/THTTTTTTH Very enotimal Very usaot ional eituation/ I I I I I I I I I I I situation TTTTTTTTTWE I know how to I do not know behave in this how to behave in situation/ I I I I I I I I I I Ithia situation TTTTTTTTTWE There is a good There is a good chance that he/she chance that heIehe will think less of will not think less no because of ety of me because of II/I/I/III/I t -mmt1-rrrr rrrrrmrr'ymw She/he does not She/he has a aajor have a nejor effect effectonnylife/ I I I I I I I I I I Ionaylife TTTTTTTTTEH Ihaveanajor ' Idonothawea effect on his/her njor effect on life! I I I__ I I I I I Ihither life TTTT TTTHTTEE I nut he ' I need not he nesrtive in order assertive in order to persuade-this to persuade this pereon/_ IT I3 __I___I I__I___I___I I I Iperson uTs7aTTofi PLEASE MAKE SURE THAT YOU HAVE AISNEHED EVERY ITEM. THANK Ydl. ”I.M.“ one of um clerical #3.» %M %W 51.52 flfl 55.55 57.53 276 Finally, I would like to as): three additional questions. l. Today, you compared a set of situations and have rated those situations on a nunber of scales. Considering the scales that I have given you, are there any salient or important distinctions between these situa— tions that I have not asked questions about? 2. What do you think of this project? Was this interesting, or dull? Was this laborious or easy to do? In general, do you have any helpful criticisms to make about this project? 3. Are there some situations (in this questionnaire) in whid: you would not try to persuade (or influence) the other person? If yes, which ones? Don't forget to pick up a handout explaining the goals of this project!!! REFERENCES Abelson, R. P. A technique and a model for multidimensional attitude scaling. Public Opinion Quarterly, 1955, lg, HOS-H18. Abelson, R. P. Situational variables in personality research. In S. Messick and J. Ross (Eds.), Measurement in Personality and Cognition. New York: Wiley, 1962. Aiken, L. S., and Brown, D. R. Visual form perception: congruence among spatial configurations. Perception and Psychophysics, 1969, E3 155- 160. Alker, H. A. Is personality situationally specific or intrapsychically consistent? Journal of Personality, 1972, fig; 1-16. Allport, G. Personality: a psychological interpretation. New York: Holt, 1937. Allport, G. W. Traits revisited. American Psychologist, 1966, 213 1-10. Allport, G., and Odbert, H. Trait-names: a psycholexical study. Psychological Monographs, 1936, 1, Whole No. 2111. Anderson, A. B. Brief report: additional data on distortion due to aggregation in non-metric multidimensional scaling. Multivariate Behavioral Research, 1973, 23 519-522. Anderson, A. B. Brief report: the effect of aggregation on non—metric multidimensional scaling solutions. Multivariate Behavioral Research, 1970, _5_, 369-373. Argyle, M., and Little, B. R. Do personality traits apply to social be- havior? Journal for the Theory of Social Behavior, 1972, 23 1-35. Barker, R. G. Ecological Psychology. Stanford: Stanford University Press, 1968. Barker, R. (Ed.) The stream of behavior. New York: Appleton-Century- Crofts, 1963. Barker, R., and Gump, P. Big school, little school. Stanford: Stanford University Press, 196a. Bem, D. J. Constructing cross-situational consistencies in behavior: some thoughts on Alker's critique of Mischel. Journal of Personality, 1972, 393 17-26. Bem, D., and Allen, A. On predicting some of the people some of the time: the search for cross-situational consistencies in behavior. Psychological Review, 197a, Bl! 506-520. 277 278 Bishop, D. W., and Witt, P. A. Sources of behavioral variance during leisure time. Journal of Personality and Social Psychology, 1970, 16, 352-360. _ Block, J. Some reasons for the apparent inconsistency of personality. Psychological Bulletin, 1968, 293 210-212. Block, J. Bgo identity, role variability and adjustment. Journal of Consulting Psychology, 1961, 2E3 392-397. Boster, P. J. Test of a Model of Compliance-Gaining Message Strategy Selection. Unpublished dissertation, Department of Communication, Michigan State University, East Lansing, Michigan, 1977. Bowers, K. S. Situationism in psychology: an analysis and a critique. Psychological Review, 1973, 80, 307-336. Also in N. Endler and D. Magnusson (Bds.), Interactional psychology and personality. New York: John Wiley and Sons, 1976, pp. 125-165.;‘ Burgoon, J., and Burgoon, M. Unwillingness to communicate, an anomie- alienation, and communication apprehension as predictors of small group communication. Journal of Psychology, 1974, §§2 31-38. Burton, R. V. Generality of honesty reconsidered. Psychological Review, 1963, 12, 481-999. Campus, N. Transituational consistency as a dimension of personality. Journal of Personality and Social Psychology, 197R, 22’ 593-600. Carroll, J. D. Individual differences and multidimensional scaling. In R. N. Shepard, A. K. Romney, and S. B. Nerlove (Eds.), Multidimension- al Scaling: Theory and Applications in the Behavioral Sciences, Vol. 1 Theosy. New York: Seminar Press, 1972, 105-155. Carroll, J. D., and Chang, J. J. Analysis of individual differences in multidimensional scaling via an N-way generalization of "Eckart-Young" decomposition. Psychometrika, 1970, pg, 283-319. Cartwright, D. Trait and other sources of variance in the S-R inventory of anxiousness. Journal of Personality and Social Psychology) 1975, sg, nos-tun. Christie, R., and Geis, P. (Bds.) Studies in machiavellianism. New York: Academic Press, 1970 (a). Christie, R., and Geis, F. Implications and Speculations. In R. Christie and P. Geis (Eds.), Studies in machiavellianism. New York: Academic Press, 1970 (b), pp. 339-358. Cliff, N. Liking judgments and multidimensional scaling. Educational ”and'PsycholOgical Measurement, 1969, 22) 73-85. 279 Cliff, N. Consistencies among judgments of adjective combinations. In A. K. Romney, R. W. Shepard and S. B. Nerlove (Eds.), Multidimensional scaling: theory and applications in the behavioral sciences, Vol. II. New York: Seminar Press, 1972, pp. 163-180. Cliff, N., and Young, F. W. On the relation between unidimensional judg- ments and multidimensional scaling. Organizational Behavior and Human Performance, 1968, s) 269-285. Craik, K. H. Environmental psychology. In K. H. Craik, B. Kleinmuntz, R. L. Rosnow, R. Rosenthal, J. A. Cheyne, and R. H. Walters (Bds.), New Directions in Psychology, 4. New York: Holt, Rinehart and Winston, 1970, 1-121. 'I" Cronbach, L. J., Gleser, G. C., Nanda, H., and Rajaratnam, N. The depen- dability of behavioral measurements: theory ofgeneralizability for scores and profiles. New York: Wiley, 1972. Ekehammar, B. Interactionism in personality from a historical perspec- tive. Psychological Bulletin, 1979, g1, 1026-1048. Ekehammar, B., Magnusson, D., and Ricklander, L. An interactionist approach to the study of anxiety. Scandinavian Journal of Psychology, 197a, 1s, u-1u. Ekehammar, B., and Magnusson, D. A method to study stressful situations. Journal of Personality and Social Psychology, 1973, 21, 176-179. Ekehammar, B., Schalling, D., and Magnusson, D. Dimensions of stressful situations: a comparison between a response analytic and a stimulus analytic approach. Multivariate Behavioral Research, 1975, 199 155- 16a. Bkman, G. Dimensions of emotion. Acta Psychologica, 1955, 123 279-288. Endler, N. S. The case for person—situation interactions. Canadian Psychological Review, 1975, is) 12-21. Endler, N. S. The person versus the situation--a pseudo issue? A re- sponse to Alker. Journal of Personality, 1973, 31, 287-303. Endler, N. 8., and Hunt, J. McV. Generalizability of contributions from sources of variance in the S-R inventory of anxiousness. Journal of Personality, 1969, 229 1-2u. Endler, N. 8., and Hunt, J. McV. S-R inventories of hostility and com- parisons of the proportions of variance from persons, responses, and situations for hostility and anxiousness. Journal of Personality and Social Psychology) 1968, 23 309-315. Endler, N., and Hunt, J. McV. Sources of behavioral variance as measured by the S-R inventory of anxiousness. Psychological Bulletin, 1966, gs, 336-3u6. 280 Endler, N. 8., Hunt, J. McV., and Rosenstein, A. J. An S-R inventory of anxiousness. Psychological Monographs, 1962, 2&3 No. 17 (Whole No. 536). Endler, N., and Magnusson, D. (Bds.) Interactional psychology and per- sonality. New York: John Wiley and Sons, 1976—(a). Endler, N., and Magnusson, D. Personality and person by situation inter- actions. In N. Bndler and D. Magnusson (Eds.), Interactional psychol- sgy and personality, New York: John Wiley and Sons, 1976 (b). Exline, R., Thibaut, J., Hickey, C., and Gumpert, P. Visual interaction in relation to Machiavellianism and an unethical act. In R. Christie and F. Geis (Bds.), Studies in Machiavellianism. New York: Academic Press, 1970, pp. 5n-76. Festinger, L. A theory of cognitive dissonance. Stanford: Stanford University Press, 1957. Fiedler, F. A theory of leadership effectiveness. New York: McGraw- Hill, 1967. Fiedler, P. Validation and extension of the contingency model of leader- ship effectiveness: a review of empirical findings. Psychological Review, 1971, ls, 128-1u8. Forgas, J. The perception of social episodes: categorical and dimen- sional representations in two different social milieus. Journal of Personality and Social Psychology, 1976, E39 199-209. Frederiksen, N. Toward a taxonomy of situations. American Psychologist, 1972, 21, 11u—123. Prederiksen, N., Jensen, 0., and Beaton, A. E. Prediction of organiza- tional behavior. New York: Pergamon Press, 1972. Geis, F., Christie, R., and Nelson, C. In search of the Machiavel. In R. Christie and P. Geis (Eds.), §£udies in Machiavellianism. New York: Academic Press, 1970, pp. 78—95. Geis, P., Weinheimer, S., and Berger, D. Playing legislature: cool heads and hot heads. In R. Christie and F. Geis (Bds.), Studies in Machiavellianism. New York: Academic Press, 1970, pp. 190-209. Gibson, W. A. Three multivariate models: factor analysis; latent struc- ture analysis and latent profile analysis. Psychometrika, 1959, 233 229-252. Goffman, E. Frame analysis: an essay on the organization of experience. New York: Harper and Row, 197a. Goldberg, L. A historical survey of personality scales and inventories. In P. McReynolds (Bd.), Advances in psychological assessment (Vol. 2). Palo Alto, California: Science and Behavior Books, 1971. 281 Goldberg, L. Some recent trends in personality assessment. Journal of Personality Assessment, 1972, BE) 597-560 (a). Goldberg, L. Student personality characteristics and optimal college learning conditions: an extensive search for trait-by-treatment in- teraction effects.’ InStructionaI Science, 1972, 13 153-210 (b). Golding, S. Plies in the ointment: methodological problems in the analysis of the percentage of variance due to persons and situations. Psychological Bulletin, 1975, g2) 278-288. Gough, H. The California psyshological inventosy, Palo Alto, California: Consulting Psychology, 1957. Harmon, H. H. Modern factor analysis (2nd ed.). Chicago: University of Chicago Press, 1967. Hartshorne, H., and May, M. Studies in the nature of character. Vol. 1. Studies in deceit. New York: Macmillan, 1928. Hastorf, A. H., Richardson, 8. A., and Dornbush, S. “I. The problem of relevance in the study of person perception. In R. Taguiri and L. Petrullo (Eds.), Person perception and interpersonal behavior. Stanford: StanfOrd University Press, 1958, pp. 54-62. Hewes, D., Haight, L., and Szalay, S. The utility and cross-situational consistency of measures of verbal output: an exploratory study. ICA Top Three paper, presented at the annual convention of the Inter- national Communication Association convention, Portland, Oregon, 1976. Hodges, W. P., and Felling, J. P. Types of stressful situations and their relation to trait anxiety and sex. Journal of Consulting and Clinical Psychology, 1970, 33, 333-337. Insel, P., and Moos, R. Psychological Environments: Expanding the scope of human ecology. American Psychologist, March, 197a, 179-188. Jackson, D. N., Messick, S., and Solley, C. M. A multidimensional scaling approach to the perception of personality. Journal of Psychology, 1957, 333 311-318. Johnson, 8. C. Hierarchical clustering schemes. Psychometrika, 1967, fig, 2u1-25u. Jones, L., and Young, F. Structure of a social environment: longitudin- al individual difference scaling for an intact group. Journal of Personality and Social Psychology, 1972, 322 108-121. Kaminski, B., McDermott, 8., and Boster, F. The use of compliance- gaining strategies as a function of machiavellianism and situation. Paper presented at the annual convention of the Central States Speech Association, Southfield, Michigan, April, 1977. 282 Koltuv, B. Some characteristics of intrajudge trait intercorrelations. Psychological Monogsaphs, 1962, 19 (33, Whole No. 552): 1-33. Krause, M. S. Use of social situations for research purposes. American Psychologist, 1970, 748-753. Kruskal, J. B. Multidimensional scaling by optimizing goodness-of-fit to a nonmetric hypothesis. Psychometrika, 1964, 223 1-27 (a). Kruskal, J. B. Nonmetric multidimensional scaling: a numerical method. Psychometrika, 1964, 22, 115-129 (b). Lazarsfeld, P. P., and Henry, N. W. Latent structure analysis. Boston: Houghton Mifflin, 1968. Levin, J. Three-mode factor analysis. Psyshological Bulletin, 1965, s3, uuz-usz. McGovern, J., and Gormly, J. Validation of personality traits: a multi- criteria approach. Journal of Personality and Social Psychology, 1976, s3, 791-795. McGuire, W. Inducing resistance to persuasion. In L. Berkowitz (Ed.), Advances in experimental social psychology, Vol. I. New York: Academic Press, 1964, pp. 191-229. Magnusson, D. An analysis of situational dimensions. Perceptual and Motor Skills, 1971, 32, 851-867. Also in N. S. Endler and D. Magnusson (Eds.), Interactional-psychology and personality. New York: John Wiley and Sons, 1976, pp. 516-532. Magnusson, D., and Ekman, G. A. A psychophysical approach to the study of personality traits. Multivariate Behavioral Research, 1970, E, 255—273. Magnusson, D., and Ekehammar, B. Perceptions of and reactions to stress- ful situations. Journal of Personality and Social Psychology, 1975, s1, 1147-1154. Magnusson, D., and Ekehammar, B. An analysis of situational dimensions: a replication. Multivariate Behavioral Research, 1973, s, 331-339.' Magnusson, D., Jerzen, M., and Nyman, B. The generality of behavioral data: I. Generalization from observations on one occasion. Multi- variate Behavioral Research, 1968, 2, 295-320. Magnusson, D., and Heffler, B. The generality of behavioral data: III. Generalization potential as a function of the number of observation instances. Multivariate Behavioral Research, 1969, 2, 29-42. Magnusson, D., Heffler, B., and Nyman, B. The generality of behavioral data: II. Replication of an experiment on generalization from obser- vations on one occasion. Multivariate Behavioral Research, 1968, s, 415-422. 283 Marwell, G., and Schmitt, D. Dimensions of compliance-gaining behavior: an empirical analysis. Sociometry, 1967, 229 350-364. Miller, D. R. The study of social relationships: situation, identity and social interaction. In S. Koch (Ed.), Psychology: a study of a science (Vol. 5). New York: McGraw-Hill, 1963. Miller, G. R. Poreward. In G. R. Miller (Ed.), Explorations in inter- personal communication. Beverly Hills, California: Sage Publications, 1976, pp. 9-16. Miller, G. R., Boster, P., Roloff, M., and Seibold, D. Compliance- gaining message strategies: a typology and some findings concerning effects of situational differences. Paper presented at the annual convention of the International Communication Association, Portland, Oregon, April, 1976. Miller, G. R., and Rokeach, M. Individual differences and tolerance for inconsistency. In R. P. Abelson et al. (Eds.), Theories of cognitive consistency: a sourcebook. Chicago: Rand McNally and Co., 1968. Miller, G. R., and Steinberg, M. Between people: a new analysis of interpersonal communication. Chicago: Science Research Associates, Inc., 1975. Mischel, W. Toward a cognitive social learning reconceptualization of personality. Psychological Review, 1973, 82, 252-283. Mischel, W. Personality and assessment. New York: Wiley, 1968. Mischel, W., Ebbesen, B., and Zeiss, A. Selective attention to the self: situational and dispositional determinants. Journal of Personality and Social Psychology, 1973, 31, 129-142. Mischel, W., and Staub, E. Effects on expectancy on working and waiting for larger rewards. Journal of Personality and Social Psychology, 1965, _2_, 625-633. Moos, R. H. Differential effects of ward settings on psychiatric pa- tients. Journal of Nervous and Mental Disease, 1967, 145, 272-283. Moos, R. H. Situational analysis of a therapeutic community milieu. Journal of Abnormal Psychology, 1968, 22, 49-61. Moos, R. H. Sources of variance in responses to questionnaires and in behavior. Journal of Abnormal Psychology, 1969, 22, 405-412. Moos, R. H. Differential effects of psychiatric ward settings on patient change. Journal of Nervous and Mental Disease, 1970, s, 316-321. Moos, R. H. Conceptualizations of human environments. American Psychol- ogist, 1973, 322 652-665. 284 Mortensen, C., and Arnston, P. The effect of predispositions toward verbal behavior on interaction patterns in dyads. Quarterly Journal of Speech, 1974, EB, 421-430. Nelson, B., Grinder, R., and Mutterer, M. Sources of variance in beha- vioral measures of honesty in temptation situations: methodological analyses. Developmental Psychology, 1969, 1, 265-279. Nowlis, V. Research with the mood adjective check list. In S. Tomkins and C. Izard (Eds.), Affect, cognition and personality. New York: Springer, 1965. Osgood, C., and Tannenbaum, P. The principle of congruity in the predic- tion of attitude change. Psychological Review, 1955, 62, 42-55. Pace, R. College and university environment scales. (Technical manual, 2nd ed.). Princeton, New Jersey: Educational Testing Service, 1969. Parks, M. R., Farace, R. V., Rogers, L. E., Albrecht, T., and Abbott, R. Stochastic process analysis of relational communication in marital dyads. Paper presented at the annual convention of the International Communication Association, Portland, Oregon, April, 1976. Pervin, L. A free-response description approach to the analysis of person-situation interaction. Journal of Personality and Social Psychology, 1976, §_4_, 465-474. Pervin, L. A. Performance and satisfaction as a function of individual- environment fit. Psychological Bulletin, 1968, Pg, 56-68. Petterson, R., Centra, J., Hartnett, R., and Liner, R. ‘Institutional functioning inventory: preliminary technical manual. Princeton, New Jersey: ’Educational Testing Service, 1970. Phillips, G. Reticence: pathology of the normal speaker. Speech Mono- graphs, 1968, 22, 39-49. Roloff, M. Communication strategies, relationships and relational changes. In G. R. Miller (Ed.), Psplorations in Interpersonal Commun- ication. Beverly Hills, California: Sage Publications, Inc., 1976, pp. 173-196. Rosenberg, S., and Sedlak, A. Structural representations of implicit personality theory. In L. Berkowitz (Ed.), Advasses in Experimental Social Psychology, New York: Academic Press, 1972, pp. 235-297. Sandell, R. Effects of attitudinal and situational factors on reported choice behavior. Journal of Marketing Research, 1968, 5. Also in N. Endler and D. Magnusson (Eds.), Interactional psychology and per- sonaliEy. New York: John Wiley and Sons, 1976, pp. 391-397. 285 Schalling, D. Tolerance for experimentally induced pain so related to personality. Scandinavian Journal of Psychology, 1971, 132 271-281. Schneider, D. J. Implicit personality theory: a review. Psychological Bulletin, 1973, 13, 294-309. Sells, S. B. An interactionist looks at the environment. American Psychologist, 1963, 122 696-702 (a). Sells, S. B. Dimensions of stimulus situations which account for beha- vior variance. In S. B. Sells (Ed.), Stimulus determinants of beha- vior. New York: Ronald, 1963 (b). Sherif, M., and Hovland, C. Social judgment. New Haven: Yale Univer- sity Press, 1961. Sherman, R. C. Individual differences in perceived trait relationships as a function of dimensional salience. Multivariate Behavioral Re- search, 1972, 2, 109-129. Shouval, R., Zakay, D., and Halfon, Y. Autonomy or the autonomies? Trait consistency and situation specificity. Journal of Multivariate Behavioral Research, 1977, 12, 143-157. Spence, I., and Ogilvie, J. C. A table of expected stress values for random rankings in non-metric multidimensional scaling. Multivariate Behavioral Research, 1973, P, 511-517. Stagner, R. Psychology of personality. New York: McGraw-Hill, 1937. Stern, C. G. People in context: measuring person-environment congruence in education and industry. New York: Wiley, 1970. Stewart, T. R. Generality of multidimensional representations. Multi- variate Behavioral Research, 1974, P, 507-519. Tucker, L. R., and Messick, S. An individual differences model for mul- tidimensional scaling. Psychometrika, 1963, PP, 333-367. Tucker, L. R. Relations between multidimensional scaling and three-mode factor analysis. Psychometrika, 1972, s1, 3-27. Trickett, B., and Moos, R. Generality and specificity of student reac- tions in high school classrooms. Adolescence, 1970, P, 373-390. Vanghan, G. M., and Corballis, M. C. Beyond significance: estimating strength of effects in selected ANOVA designs. Psychological Bulletin, 1969, 22, 204-213. iiallace, J. An abilities conception of personality: some implications for personality measurement. American Psychologist, 1966, 233 132-138. 286 Wallach, M., and Leggett, M. Testing the hypothesis that a person will be consistent: stylistic consistency versus situational specificity in size of children's drawings. Journal of Personality, 1972, 32, 309-330. Ward, L. M. Multidimensional scaling of the moral physical environment. Multivariate Behavioral Research, 1977, 12, 23-42. Wiggins, N., and Hoffman, P. J. Dimensions of profile judgments as a function of instructions, cue-consistency and individual differences. Multivariate Behavioral Research, 1968, s, 3-20. Wish, M. Role and personal expectations about interpersonal communica- tion. Paper presented at the U.S.-Japan Seminar on Multidimensional Scaling, San Diego, California, August, 1975. Wish, M., Deutsch, M., and Kaplan, S. Perceived dimensions of interper- sonal relations. Journal of Personality and Social Psychology, 1976, g, 409-420. Woelfel, J. Metric measurement of cultural processes. Paper presented to the Speech Communication Association, Chicago, December, 1974. Woelfel, J., and Danes, J. New techniques for multidimensional analysis of communication and conception. In P. Monge and J. Cappella (Eds.), Multivariate analysis in communication research. Forthcoming. Woelfel, J., Saltiel, J., McPhee, R., Danes, J., Cody, M., Barnett, G., and Serota, K. Orthogonal rotation to theoretical criteria: compari- son of multidimensional spaces. Paper presented at the annual conven- tion of the Mathematical Psychology Association, Purdue, Indiana, 1975. Wood, J. R., Weinstein, E. A., and Parker, R. Children's interpersonal tactics. Sociological Inquiry, 1967, El, 129-138. Zedeck, S. Problems with the use of "moderator" variables. Psychologi- cal Bulletin, 1971, 22, 295-310.