flfiw%glla'n-—---. _, _ A STUDY OF THE VALUES AND VALUE SYSTEMS . 1; OF COLLEGE RDDNIIVIATES IN ' CDNELIDT AND NDNDDNFLICT SITUATIDNS, AND AN INVESTIGATION TO DETERMINE WHETHER NDD'NIMATE CONFLICT CAN BE ATTRIBUTED TD DlFFERING VALUES AND VALUE SYSTEMS Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY ANDREW FRANK SIKULA 1970 ...... This is to certify that the thesis entitled A STUDY OF THE VALUES AND VALUE SYSTEMS'OF'COLLEGE ROOMMATES IN CONFLICT AND NONCONFLICT SITUATIONS, AND AN INVESTIGATION TO DETERMINE WHETHER ROOMMATE CONFLICT CAN BE ATTRIBUTED TO DIFFERING VALUES AND VALUE SYSTEMS. presented bg ANDREW FRANK S I KU LA has been accepted towards fulfillment of the requirements for ("I 11/1 7) degree in Mx/WVT L. Major professor Date .1/2’15/1; I III IIIIIIIIII III II I I III III II I. I ABSTRACT A STUDY OF THE VALUES AND VALUE SYSTEMS OF COLLEGE ROOMATES IN CONFLICT AND NONCONFLICT SITUATIONS, AND AN INVESTIGATION TO DETERMINE WHETHER ROOMMATE CONFLICT CAN BE ATTRIBUTED TO DIFFERING VALUES AND VALUE SYSTEMS BY Andrew Frank Sikula Problem The problems of business organizations are basically managerial problems many of which involve interpersonal conflict situations. By better understanding individual conflict situations, business and managerial situations of peer conflict, group (departmental) conflict, line-staff conflict, and labor-management conflict can all be better understood. The basic "problem" is twofold; first, there is a lack of understanding of the phenomenon of conflict; secondly, there is an underemphasis on and a lack of recog- nition of the importance of values and value systems as explanations of human behavior (such as conflict). Hope- fully, this study and subsequent research to follow will Andrew Frank Sikula serve as a new approach which adds insight and understand- ing into the human phenomenon of interpersonal conflict. Procedure The main purpose of this study was to determine whether or not interpersonal conflict could be attributed to differing individual values and value systems. To investigate this idea, approximately twelve hundred stu- dents were screen-interviewed to obtain a final sample of 200 students (100 student pairs). Each student and each pair in the final sample met various "entry conditions" and accordingly qualified for either the "conflict" or the "nonconflict" group. Each student and each pair in the final sample also filled out and returned Form E of the Rokeach Value Survey. The final 200 students formed 100 pairs, 50 pairs of conflict students and 50 pairs of non- conflict students. This aggregation en masse constitutes the "entire sample." The fifty pairs in each group were equally subdivided into an "original sample" and a "cross- validation sample." Data Analysis Existing computerized statistical operations ("routine oneway" and "routine stable") designed specifi- cally for use with the Rokeach Value Survey instrument were modified and applied to the three samples. From such data as applied to the original sample, a "prediction process" Andrew Frank Sikula was formulated and later subjectively and indirectly applied to the cross-validation sample. Findings For the sake of brevity and generalization, it is sufficient here to limit the discussion to the "entire sample" and to only the findings from "routine stable." "Value systems" were analyzed basically by routine stable in the form of pair agreement correlation coefficients (Spearman) between roommate l and rooomate 2 of each pair in each group for both the "terminal" and "instrumental" value scales (considered separately and as entities). In the entire sample, it was found that conflict and non- conflict roommate pairs collectively and significantly had diverse and different rankings for the terminal value scale but not for the instrumental value scale. The termi- nal value scale (conceived as an entity) distinguished be- tween conflict and nonconflict pairs at the .01 level of significance. The null hypothesis in question is whether the conflict and the nonconflict pairs came from the same or identical population. This hypothesis can be rejected at the .01 level of significance. In short, conflict and nonconflict roommate pairs do have significantly different "value systems." Furthermore, conflict (or its absence) between college roommates can be attributed to differing (similar) "value systems" (terminal but not instrumental). Andrew Frank Sikula "Individual values" were analyzed basically by routine stable in the form of Pearson correlation coef- ficients for both conflict and nonconflict pairs for both terminal and instrumental values. In the entire sample, for conflict pairs, three significant individual terminal values were determined. For nonconflict pairs, six sig- nificant individual terminal values and one significant individual instrumental value were discovered. In addition, the "Wilcoxon" test revealed that significant differences for the ranking of the individual values between conflict and nonconflict group pairs occurred in five out of a possible thirty-six instances. Again, the null hypothesis which assumes that conflict and nonconflict pairs come from the same population was significantly rejected (at least for and in regards to these five individual values sur- viving the Wilcoxon test). In sum, some individual values were ranked differently by conflict and nonconflict room- mate pairs. However, the idea of using individual values and/ or value systems as "determinants" or "predictors" of con— flict and nonconflict situations was not supported as evidenced by the failure of the "prediction process" which was derived from the original sample and subjectively and indirectly applied to the cross-validation sample. Al- though much of the reported routine stable data is "sig- nificant," it is not necessarily "determinant." Individual values and value systems may be characteristic of and r"\l' ' Andrew Frank Sikula significantly correlated within some samples-~but not necessarily all subsamples. Although individual values and value systems relationships are not always consistent among subsamples, across many samples (and subsamples) some individual values and value systems relationships are consistently noteworthy (for examples, the "individual values" of Wisdom, Forgiving, et al., and the terminal "value system" in this study). In sum, the data reveal that some individual values and value systems are more important than are others as "characteristics" (but not necessarily "determinants" or "predictors") of conflict and nonconflict situations. Having viewed the results of this research project, the author envisions a parallel between value theory and the personality traits concept. For example, although, in general, successful executives may have certain per- sonality traits in common (e.g., the need for achievement), each successful executive has a unique set of traits any one of which will not assure nor prevent his successful- ness. Similarly, although, in general, conflict and non- conflict roommates may have certain values in common) each roommate has a unique set of values any one of which will not assure nor prevent his compatibility. A STUDY OF THE VALUES AND VALUE SYSTEMS OF COLLEGE ROOMMATES IN CONFLICT AND NONCONFLICT SITUATIONS, AND AN INVESTIGATION TO DETERMINE WHETHER ROOMMATE CONFLICT CAN BE ATTRIBUTED TO DIFFERING VALUES AND VALUE SYSTEMS BY Andrew Frank Sikula A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Management 1970 §G72Vé ACKNOWLEDGMENTS The author gratefully acknowledges the help of his Dissertation Committee: Dr. Dalton McFarland; Dr. Carl Frost; and Dr. Winston Oberg. The author especially wants to thank Dr. Oberg for his advice, suggestions, and en- couragement. Recommendations from Dr. Oberg have helped to both mold and refine this research project throughout its various stages. The author also wishes to express his appreciation and gratitude to Mr. John Teeter for his computer program- ming assistance, to Dr. James Stapleton for his advice con- cerning statistical matters, and to Mrs. Dorothy Johnson and Mrs. Carolyn Piersma for their clerical, proofreading, and miscellaneous endeavors. ii .TABLE OF CONTENTS Chapter Page I. PROBLEM DESCRIPTION AND THESIS OBJECTIVE . . 1 Introduction. . . . . . . . . . . 1 Problem . . . . . . . . . . . . 2 Objectives . . . . . . . . . . . 5 Semantic Difficulties. . . . . . . . 6 Definition of Key Terms . . . . . . . 13 Importance of Values . . . . . . . . 15 Related Research . . . . . . . . . 21 II. GENERAL PROCEDURE AND METHODOLOGY . . . . 28 Introduction. . . . . . . . . . . 28 Evolution of the Present Research Study. . 29 DeSign. O O O O I O I O I O O O 36 Population and Sample. . . . . . . . 41 Data and Instrumentation. . . . . . . 43 Rokeach Value Survey . . . . . . . 73 Pilot Study and Subsequent Changes in the Initially Planned Procedures. . . . . 76 III. ANALYSIS OF THE DATA. . . . . . . . . 80 Introduction. . . . . . . . . . 80 Statistical Techniques of "Routine Oneway". 82 Statistical Techniques of "Routine Stable" . 90 Group Prediction of Unclassified Pairs . . 97 IV. FINDINGS. . . . . . . . . . . . . 103 Introduction. . . . . . . . . . . 103 Original Sample--Routine Oneway . . . . 103 Original Sample--Routine Stable . . . . 124 Group Prediction of Unclassified Pairs . . 151 Cross-Validation Sample-~Routine Stable. . 163 Entire Sample--Routine Oneway . . . . . 188 Entire Sample--Routine Stable . . . . . 201 iii Chapter Page V. SUMMARY, CONCLUSIONS, AND IMPLICATIONS . . 218 Introduction . . . . . . . . . . 218 Procedure . . . . . . . . . . . 218 Results. . . . . . . . . . . . 219 Limitations of the Study . . . . . . 226 Contributions of the Study . . . . . 228 Implications and Future Research . . . 230 BIBLIOGRAPHY . . . . . . . . . . . . . 233 APPENDICES APPENDIX A. Interview-Questionnaire . . . . . . . 238 B. Rokeach Value Survey, Form B . . . . . 241 C. Program Description and User's Manual for: Valutest . . . . . . . . . . . 244 D. Statistical Supplements . . . . . . . 316 iv Table LIST OF TABLES Original Sample--Routine Oneway--Frequency Distributions, Medians, and Quartile Deviations for Terminal and Instrumental Values (N = 100) . . . . . . . . . Original Sample--Routine Oneway--Medians and Ranks for Terminal and Instrumental Values (N = 100) o a o o o o o o o o 0 Original Sample--Routine Oneway--Frequency Distributions, Medians, and Quartile Deviations for Conflict and Nonconflict Groups (N = 50) . . . . . . . . . Original Sample--Routine Oneway--Termina1 Values, Medians, and Ranks for Conflict and Nonconflict Groups (N = 50) . . . . Original Sample--Routine Oneway--Instru- mental Values, Medians, and Ranks for Conflict and Nonconflict Groups (N = 50) . Original Sample--Routine Stable-~Agreement Coefficients for Each Pair: Terminal and Instrumental Value Coefficients for Con— flict and Nonconflict Group Pairs (N=25)............ Original Sample--Routine Stable--Frequency Distribution, Median, Mean, and Standard Deviation of Agreement Coefficients for Terminal and Instrumental Values for Pairs in Both the Conflict and Nonconflict Groups (N = 25) . . . . . . . . . Original Sample--Routine Stable--Frequen- cies, Medians, and Quartile Deviations of Differences for Terminal Values for Conflict Group Pairs (N = 25). . . . . Page 105 108 109 113 114 125 129 133 Table Page 19. Cross-Validation Sample--Routine Stable-- Frequencies, Medians, and Quartile Devi- ations of Differences for Terminal Values for Conflict Group Pairs (N = 25) . . . . 170 20. Cross-Validation Sample--Routine Stable-- Frequencies, Medians, and Quartile Devi- ations of Differences for Terminal Values for Nonconflict Group Pairs (N =.25) . . . 171 21. Cross-Validation Sample--Routine Stable-- Median Difference, Roommate l to Roommate 2 Correlation, and Ranks for Terminal Values for Conflict Group Pairs (N = 25) . . . . 174 22. Cross-Validation Sample--Routine Stable-- Median Difference, Roommate 1 to Roommate 2 Correlations, and Ranks for Terminal Values for Nonconflict Group Pairs (N = 25) . . . 175 23. Cross-Validation Sample--Routine Stable--Fre- quencies, Medians, and Quartile Deviations of Differences for Instrumental Values for Conflict Group Pairs (N = 25) . . . . . 177 24. Cross-Validation Sample--Routine Stable--Fre- quencies, Medians, and Quartile Deviations of Differences for Instrumental Values for Nonconflict Group Pairs (N = 25) . . . . 178 25. Cross-Validation Sample-~Routine Stable-- Median Difference, Roommate 1 to Roommate 2 Correlation, and Ranks for Instrumental Values for Conflict Group Pairs (N = 25). . 180 26. Cross-Validation Sample--Routine Stable-- Median Difference, Roommate 1 to Roommate 2 Correlation, and Ranks for Instrumental Values for Nonconflict Group Pairs (N = 25). 181 27. Cross-Validation Sample-~Prediction Process . 183 28. Entire Sample--Routine Oneway--Frequency Distributions, Medians, and Quartile Devi- ations for Terminal and Instrumental Values (N = 200). . . . . . . . . . 190 vii Table Page 29. Entire Sample--Routine Oneway--Medians and Ranks for Terminal and Instrumental Values (N = 200) O O O I O I O O O O O O 193 30. Entire Sample--Routine Oneway--Frequency Distributions, Medians, and Quartile Devi- ations for Terminal and Instrumental Values for Conflict and Nonconflict Groups (N = 100) . . . . . . . . . . . . 194 31. Entire Sample--Routine Oneway-~Termina1 Values --Medians and Ranks for Conflict and Non- conflict Groups (N = 100). . . . . . . 198 32. Entire Sample--Routine Oneway--Instrumenta1 Values-~Medians and Ranks for Conflict and Nonconflict Groups (N = 100). . . . . . 199 33. Entire Sample--Routine Stable--Agreement Coef- ficients for Each Pair: Terminal and Instrumental Value Coefficients for Conflict Group Pairs (N = 50) . . . . . . . . 202 34. Entire Sample--Routine Stable--Agreement Coef- ficients for Each Pair: Terminal and Instrumental Value Coefficients for Non- conflict Group Pairs (N = 50) . . . . . 203 35. Entire Sample--Routine Stable--Frequency Distribution, Median, Mean, and Standard Deviation of Agreement Coefficients of Terminal and Instrumental Values for Pairs in Both the Conflict and Nonconflict Groups (N = 50) . . . . . . . . . . . . 204 36. Entire Sample--Routine Stable--Frequencies, Medians, and Quartile Deviations of Differ- ences for Terminal Values for Conflict Group Pairs (N = 50) . . . . . . . . 207 37. Entire Sample--Routine Stable--Frequencies, .Medians, and Quartile Deviations of Differ- ences for Terminal Values for Nonconflict Group Pairs (N = 50) . . . . . . . . 208 38. Entire Sample--Routine Stable-~Median Differ- ence, Roommate 1 to Roommate 2 Correlation, and Ranks for Terminal Values for Conflict Group Pairs (N = 50) . . . . . . . . 209 viii Table Page 39. Entire Sample--Routine Stable--Median Differ- ence, Roommate 1 to Roommate 2 Correlation, and Ranks for Terminal Values for Noncon- flict Group Pairs (N = 50) . . . . . . 210 40. Entire Sample--Routine Stable--Frequencies, Medians, and Quartile Deviations of Differ- ences for Instrumental Values for Conflict Group Pairs (N = 50) . . . . . . . . 212 41. Entire Sample--Routine Stable--Frequencies, Medians, and Quartile Deviations of Differ- ences for Instrumental Values for Noncon- flict Group Pairs (N = 50) . . . . . . 213 42. Entire Sample-~Routine Stable--Median Differ- ence, Roommate 1 to Roommate 2 Correlation, and Ranks for Instrumental Values for Con- flict Group Pairs (N = 50) . . . . . . 214 43. Entire Sample--Routine Stable--Median Differ- ence, Roommate 1 to Roommate 2 Correlation, and Ranks for Instrumental Values for Non- conflict Group Pairs (N = 50) . . . . . 215 44. Summary of Findings . . . . . . . . . 221 D-2. Probabilities Associated with Values as Extreme as Observed Values of z in the Normal Distribution . . . . . . . . 322 D-3. Critical Values of Chi-Square. . . . . . 323 D-4. Critical Values of t . . . . . . . . 324 ix LIST OF FIGURES Figure Page 1. Design of Research . . . . . . . . . 39 2. Screening Process for Nonconflict Group Members . . . . . . . . . . . . 64 3. Screening Process for Conflict Group Members . . . . . . . . . . . . 66 4. Size of 2 Scores Necessary for Rejection of Null Hypothesis at Various Levels of Significance (Two-Tailed Test)--Wi1coxon Technique . . . . . . . . . . . 119 5. Computation of R- and ZR- in Formula for Kendall Concordance Coefficient . . . . 122 6. Relationship of t and r for Three Values of N and Three Levels of Significance (t-Test) O I O O O O O O O O O O 126 7. Original Sample--Routine Stable--Number of Pairs in the Conflict and Nonconflict Groups Which Had Meaningful Correlations at Various Significance Levels for the Terminal and Instrumental Value Scales (N = 25). . . . . . . . . . . . 127 8. Cross-Validation Sample--Routine Stable-- Number of Pairs in the Conflict and Non- Conflict Groups Which Had Meaningful Correlations at Various Significance Levels for the Terminal and Instrumental Value Scales (N = 25) . . . . . . . 164 D} II ’ Figure Page 9. Entire Sample--Routine Stable--Number of Pairs in the Conflict and Nonconflict Groups Which Had Meaningful Correlations at Various Significance Levels for the Terminal and Instrumental Value Scales (N = 50) . . . . . . . . . . . . 205 xi CHAPTER I PROBLEM DESCRIPTION AND THESIS OBJECTIVE Introduction The purpose of this chapter is to introduce the reader to the subject matter of this study by presenting and discussing topics which will serve as background material for ideas that will be discussed in greater de- tail in later chapters. To obtain an adequate grasp of and an intuitive feel for the underlying rationale of this study, it is necessary initially to present and to discuss a number of preliminary topics. Accordingly, in this chapter, the initial task is to identify the general "problem" which will be examined and analyzed in this study. Then the "objectives" of the current study are presented and immediately following the "objectives" is a discussion of the "semantic difficulties" encountered when dealing with such abstract concepts as attitudes, values, value systems, conflict, and motivation. After the "semantic difficulties" have been mentioned, the "definition of key terms" section of the chapter defines the key concepts which will be examined in this research project. The "importance of values" portion of this chapter illustrates why a shift from the study of atti- tudes to the study of values may often be necessary in order to more adequately understand human behavior in various situations. The final section of this chapter, entitled "related research," briefly discusses in a general way the academic literature which deals indirectly with the subject matter of this study. Problem The problems of business organizations are basically managerial problems many of which involve interpersonal con- flict situations. Organizations are a combination of groups which in turn are a combination of individual human beings. By better understanding individual conflict situations, business and managerial situations of peer conflict, group (departmental) conflict, line-staff conflict, and labor- management conflict can all be better understood. Hope- fully, this study and subsequent research to follow may serve as a new approach to studying these more complex business and managerial situations. There are basically two vital questions about con- flict to be asked and answered. First, what are the nature and causes of the conflict; and, secondly, how can such conflict be prevented or resolved? This study lays out a format which analyzes and answers basically only the first of these two questions. Possible answers to the second question may be apparent after the initial question has been examined. This second question will also be examined somewhat in the final chapter of this report which deals with the implications of this research study. The basic "problem" is twofold: first, there is a lack of understanding of the phenomenon of conflict; secondly, there may well be an underemphasis on and a lack of recognition of the importance of values and value systems as determinants of human behavior (such as con- flict). Conflict is a phenomenon which appears to be in- creasing in frequency, intensity, and duration throughout human society. Conflict exists between and within nations, social classes, races, religions, political parties, fami- lies, and individuals. Obviously, a better understanding of conflict can be fruitful for all human beings indi- vidually and collectively. Hopefully, this study will add insight into the understanding of the human phenomenon of conflict. Before proceeding further in this analysis of con- flict, two precautions or assumptions need to be pointed out. First, it is usually assumed and generally accepted that conflict per se is not necessarily an evil that can or should be avoided or eliminated at all costs. Each human being is unique, thus, no two persons are entirely alike and, accordingly, some conflict or incompatibility is inevitable. It can also be argued that a certain degree of conflict is beneficial, for example, competitive conflict may promote individual excellence and thus lead to a greater number of human achievements. As is usually the case, however, the extremes of this or any other human phenomenon should be avoided. Too much conflict can lead to hostility and hatred; while too little may result in a passive, indifferent, apathetic, and undynamic state of affairs. The second precaution is to recognize that any phenomenon is caused by an infinite number of interacting variables. This fact is widely known and accepted and currently is a partial explanation for the rebirth in pOpularity of the "systems approach" in the social sci- ences. To say that human behavior in general or that conflict in particular is determined solely by individual values or value systems is, of course, an oversimplifi- cation. However, the human mind is unable to understand an infinite number of interacting variables. Furthermore, the vast majority of these variables have such a minor effect on any phenomenon that they are usually ignored when attempting to explain any event. Time, energy, money, and other resources are limited so that only key variables or determinants are studied in relation to any phenomenon. Accordingly, this study of individual behavior and conflict deals only with the variables of values and value systems. Objectives This study was undertaken to fulfill, explore, and examine many objectives, most of which can be reduced to and are contained within the following four main stated objectives which are listed in their respective order of importance: 1. To determine whether conflict (or its absence) between college roommates can be attributed to differing (similar) "value systems" by: a. Identifying interpersonal conflict and nonconflict situations between college roommates. b. Studying, comparing, and contrasting the "value systems" of college roommates in conflict and nonconflict situations. To determine whether or not some "individual values" are more important than are others as possible determinants of conflict and non- conflict situations. Possibly to express and argue for the importance of "individual values" and value systems" as the primary determinants of individual be- havior. To examine the interrelatedness of individual values, value systems, individual behavior, and interpersonal conflict. Many related topics about values and value theory will spontaneously present themselves during the course of this study. Most of these additional problem areas, how- ever, are beyond the scope of this research. For examples, the questions of whether or how values change, and also the matter of whether value changes precede behavior changes or vice versa--these and other related topics are very inter- esting but will not be dealt with directly by this study. This study was designed to explore basically only the four main stated objectives as listed above. Specific hypotheses related to the above stated objectives can be found and will be analyzed in Chapter IV of this thesis which involves reporting the "Findings" revealed by analysis of the research data. It should be noted that these hypotheses will be stated in a manner which will serve directly only the examination of the first and second stated objectives of this study. The con- clusions and implications arrived at after testing these hypotheses should indirectly serve as material and evi- dence for the substantiation or refutation of the third and fourth stated objectives. Semantic Difficulties Studies in the discipline of communications have revealed that the meanings of words are in people. That is, the same word can mean different things to different people. To add to the confusion is the fact that the same word experienced (read or heard) by the same person very often means different things to this person depending upon a number of situational factors such as the tone of voice in which the word might have been delivered, or the state of mind of the person experiencing this word. Each human being is unique and practices selective perception, i.e., he sees what he wants to see and hears what he wants to hear. Meanings of words especially vary over time, that is, the longer the interval of time between experiencing a word and eventually re-experiencing it, the greater the likelihood that the meaning of the word has changed for any person. Thus all words are subject to different meanings. However, some words are more ambiguous than are others. Unfortunately, the terms value and attitude are of the ambiguous variety. In the management and business litera- ture the term "value(s)" is especially confusing for it is a homonym usually meaning monetary value but increasingly used today to also denote social value. It is only in this latter context that the word value(s) will be used in this study. Such terms as attitudes, values, and value systems are confusing enough for the layman, but, unfortunately, even the experts in the fields of psychology and sociology do not agree as to their meaning. The conceptual bound— aries between an attitude and a value, and between a value and a value system, are unclear and are today still being debated. For the purposes of this study, we will consider an attitude to be an organization of several beliefs focused on a specific object (physical or social, concrete or abstract) or situation, predisposing one to respond in some preferential manner.l Some of these beliefs about an object or situation concern matters of fact while others concern matters of evaluation. Thus, an attitude is a package of beliefs consisting of interconnected assertions to the effect that certain things about a specific object or situation are true or false, and other things about it are desirable or undesirable.2 On the other hand, values, again as defined for the purposes of this study, will have to do with modes of conduct and end-states of existence. To say that a person "has a value" is to say that he has an enduring belief that a specific mode of conduct or end-state of existence is personally and socially preferable for him to alternative modes of conduct or end-states of existence.3 Once a value is internalized it becomes, consciously or unconsciously, 1Milton Rokeach, "A Theory of Organization and Change Within Value and Attitude Systems," Presidential address to the Society for the Psychological Study of Social Issues, The American Psychological Association, September 2, 1967, Michigan State University, Department of Psychology reprint, pp. 2-3. Hereafter referred to as Rokeach, "Theory of Organization." 2 3 Ibid. Ibid. a standard, guide, or criterion for monitoring action, for developing and maintaining attitudes toward relevant ob- jects and situations, for justifying one's own and others' actions and attitudes, for morally judging self and others, and for comparing self with others. Also, one's own value- standard is employed to influence the values, attitudes, and actions of others, for example, our childrens'. The preceding explanation of value(s) is largely taken from the works of Milton Rokeach, but it is highly compatible with explanations of value(s) advanced by other leading social psychologists.4 So explained, values differ from attitudes in several important respects. While an attitude represents several beliefs focused on a specific object or situation, a value is a single belief which transcendentally guides action and judgments across spe- cific objects and situations, and even beyond immediate goals to more ultimate end-states of existence.5 In addition, a value, unlike an attitude, is an imperative to action. A value is not only a belief about the preferable, 4For examples, see Clyde K. M. Kluckhohn, "Values and Value Orientations in the Theory of Action," in Toward A General Theory of Action, ed. by T. Parsons and E. A. Shils (Cambridge: Harvard University Press, 1951); M. Brewster Smith, "Personal Values in the Study of Lives," in The Study of Lives, ed. by R. K. White (New York: Atherton Press, 1963); and/or Robin M. Williams, "Values," in International Encyclopedia of Social Sciences (New York: Macmillan, 1967). 5 Rokeach, "Theory of Organization, p. 6. 10 but it is also a preference for the preferable.6 Finally, unlike an attitude, a value is a standard or yardstick to guide actions, attitudes, evaluations, justifications, and comparisons of self and others. The distinction between preferable modes of con- duct and preferable end-states of existence, according to Rokeach, is a more or less familiar one in the literature on values. It is a distinction between values representing means and ends, that is, between instrumental and terminal values.7 An instrumental (means) value is therefore de- fined as a single belief which always takes the following form: "I believe that such-and-such a mode of conduct (for example, honesty, or courage) is personally and socially preferable in all situations with respect to all objects."8 A terminal value takes a comparable form: "I believe that such-and-such an end-state of existence (for example, sal- vation, or a world at peace) is personally and socially worth striving for."9 Only those words or phrases that can be meaningfully inserted into the first sentence are instrumental values (means), and only those words or phrases 6A. O. Lovejoy, "Terminal and Adjectival Values," Journal of Philosophy, XLVII (1950), 593-608. 7Ibid.; A. L. Hilliard, The Forms of Value (New York: Columbia University Press, 1950). 8Rokeach, "Theory of Organization," p. 6. 91bid. 11 that can meaningfully be inserted into the second sentence are terminal values (ends). Now it is necessary to distinguish between a value and a value system. Many writers believe that values are organized into hierarchical structures and substructures. Operationally speaking, the concept of value system sug- gests a ranking of values along a continuum of importance. Recalling the aforementioned discussion, for Rokeach this implies two separate value systems, instrumental and termi- nal values, each with a rank-ordered structure of its own. Although Rokeach advocates two distinct and separate value systems, he recognizes that both systems are functionally and cognitively connected together, and that both systems are also inter-connected with various attitudes toward specific objects and situations. Rokeach hypothesizes that everyone has the same absolute values although the number of such values is subject to conjecture. However, individuals have different value systems, that is, the relative importance of or the priority attached to each value varies for different individuals. This study will also be dealing with the concepts of conflict and motivation. Both concepts, especially the latter, are ambiguous, abstract, and subject to defi- nitional difficulties. Although many alternative defi- nitions are available, for the purposes of this study, conflict will be defined as a perceived state or condition 12 of incompatibility, opposition, antagonism, and/or dis- cord. Since this study deals directly with individual behavior, it must also then deal at least indirectly with the concept of motivation. However, this is one term that will not be defined since it will be dealt with only in- directly and because there is no common, widely-accepted definition of the term--not even among specialists in the area of motivational psychology. Indeed, as Cofer and Appley have stated: . Unfortunately, in the field of motivational psychology, we are rather in the position of blind men. Each in- vestigator explores an area of the subject most readily available to him or amenable to his tools or methods. His conclusions may derive from his methods and fit his facts.lo As an example of the confusion involved with the concept of motivation, the term is often used in reference to con- scious feeling of desire and the whole complex of ideas and feelings which seem to constitute the conscious ante- cedents of behavior. However, just as often, the term motivation is used to refer to the unconscious determi- nants of behavior. The debate ultimately is over the question of whether or not all behavior is motivated be- havior. Experts differ in their answer to this question, and for every different answer there is usually also an accompanying different theory. 10C. N. Cofer and M. H. Appley, Motivation: Theory and Research (New York: John Wiley and Sons, 1964), p. 809. 13 Thus, there are semantic difficulties involved when dealing with the concepts of attitude, value, instru- mental value, terminal value, value system, conflict, and motivation. In order to obtain a common understanding of these terms for the purposes of this dissertation, summary definitions follow. Definition of Key Terms Attitude.--An attitude is an organization of several beliefs focused on a specific object (physical or social, concrete or abstract) or situation, predisposing 11 An attitude one to respond in some preferential manner. is a package of beliefs. An attitude is neither a standard to guide action nor an imperative to action. Conflict.--A perceived state or condition of in- compatibility, opposition, antagonism, and/or discord. Instrumental Values.--Va1ues representing "means." An instrumental value is a single belief which always takes the form: "I believe that such-and-such a mode of conduct (e.g., honesty, courage) is personally and socially prefer- able in all situations with respect to all objects."12 Only those words or phrases that can be meaningfully in- serted into this sentence are instrumental values. llRokeach, "Theory of Organization," p. 5. lzIbid., p. 6. 14 Interpersonal Conflict.--Conflict existing between two individual human beings. Terminal Values.--Va1ues representing "ends." A terminal value is a single belief which always takes the form: "I believe that such-and-such an end-state of existence (e.g., salvation, a world of peace) is personally and socially worth striving for."13 Only those words or phrases that can be meaningfully inserted into this sentence are terminal values. Values.--Values have to do with modes of conduct and end-states of existence. To say that a person "has a value" is to say that he has an enduring belief that a specific mode of conduct or end-state of existence is personally and socially preferable to alternative modes 14 A value is a of conduct or end-states of existence. single belief which transcendentally guides actions and judgments across specific objects and situations, and beyond immediate goals to more ultimate end-states of existence. Values serve as standards to guide action and thus are imperative to action. Value System.--A rank-ordering of values along a continuum of importance. The pattern by which values l3Ibid. l4Ibid., p. 5. 15 are organized into hierarchical structures and sub- structures.15 Importance of Values Conflict is a form of human behavior. Mankind's atomistic and superficial knowledge about conflict stems from its inadequate understanding of human behavior. Part of the problem is that over the past fifty years empirically oriented social scientists have paid considerably more attention to the theory and measurement of attitudes than to the theory and measurement of values. The "importance of values" has too long been ignored as a possible expla- nation of human behavior. Over the last half century the study of social attitudes has been the central problem area of social psy- chology. The attitude concept has occupied a dominant place in theory and research, and it has been assigned prominent space in the textbooks and handbooks of social psychology. In one of these handbooks, Gordon Allport has written: The concept of attitude is probably the most dis- tinctive and indispensable concept in contemporary American psychology. No other term appears more 16 frequently in experimental and theoretical literature. lsIbid., p. 7. 16Gordon Allport, "Attitudes," in A Handbook of Social Psychology, ed. by C. Murchison (Worcester, Mass.: Clark University Press, 1935), p. 798. 16 Today, a few pioneer social psychologists are questioning whether the attitude concept should continue to occupy the central position it has enjoyed for so long. This is not to say that the attitude concept is no longer important nor that it is any the less important now than it has been in the past. The key question to ask is whether there is now, and indeed whether there always has been, an even more deserving concept for this central position. Milton Rokeach, Michigan State University social psychologist, believes that the value concept should be placed in importance ahead of the attitude concept for the following reasons: First, value seems to be a more dynamic concept since it has a strong motivational component as well as cognitive, affective, and behavioral components. Second, while attitude and value are both widely assumed to be determinants of social behavior, value is a determinant of attitude as well as of behavior. Third, if we further assume that a person possesses considerably fewer values than attitudes, then the value concept provides us with a more economical analytic tool for describing and explaining similar- ities and differences between persons, groups, nations and cultures. . . . Consider, finally, the relative ubiquitousness of the value and attitude concepts across disciplines. While attitudes seem to be a specialized concern mainly of psychology and sociology, values have long been a center of theoretical attention across many disciplines--in philosophy, education, political science, economics, anthropology, and the- ology, as well as in psychology and sociology. All these disciplines seem to share a common concern with the antecedents and consequents of value organization and value change. Around such a shared concern we may more realistically anticipate genuine interdisci- plinary collaboration.l7 l7 Rokeach, "Theory of Organization,‘ pp. 2-3. 17 It might seem somewhat paradoxical, in view of the more central theoretical status generally accorded the value concept, that we should have witnessed over the years a more rapid theoretical advance in the study of attitude rather than value. Rokeach has commented on the reasons for this: One reason for this, I suspect, was the more rapid development of methods for measuring attitudes, due to the efforts of such men as Bogardus, Thurstone, Likert, and Guttman. A second reason, perhaps, was the existence of a better consensus on the meaning of attitude than value. A third possible reason is that attitudes were believed to be more amenable to experi- mental manipulation than values. In any event, the ready availability of quantitative methods for measur- ing attitudes made it only a matter of time before experimentally-minded social psychologists would seek to determine whether the attitude variable could be fruitfully fitted into the classical pretest-treatment- posttest paradigm. And upon finding that attitudes were indeed susceptible to experimental influence, a demand was created for theories of attitude organi- zation and change to explain the results rather than for theories of value organization and change. . . . I would like to draw special attention to one conse- quence of the fact that it was the attitude rather than the value variable which thus came into the experimentalist's focus. Bypassing the problem of values and their relation to attitudes, we settled perhaps a bit too hastily for studies that I will call problems of persuasion to the neglect of what I will call problems of education and re-education. We emphasized, for example, the persuasive effects of group pressure, prestige, order of communication, role playing and forced compliance on attitudes. But we neglected the more difficult study of, say, the more enduring effects of socialization, educational innovation, psychotherapy and cultural change on values. laIbid., pp. 3-4. 18 Rokeach has summarized the superiority of the value concept as an alternative to the attitude concept by stating: It was therefore, our hope that in shifting from a concern with attitudes to a concern with values we would be dealing with a concept which is more central, ‘more dynamic, more economical; a concept which would invite a more enthusiastic interdisciplinary collabo- ration, and which would broaden the range of the social psychologist's traditional concern to include problems of education and re-education as well as problems of persuasion.19 Thus, according to Rokeach, the past emphasis on attitudes has not arisen out of any deep conviction that man's attitudes are more important determinants of social behavior than man's values. Rather it has been forced upon us, or has evolved out of more rapid development of methods for measuring attitudes, combined, perhaps, with a certain vagueness of understanding about the conceptual difference between values and attitudes and about the relation between values and attitudes. Apparently, this overemphasis on attitude, according to Rokeach, is con- ceptually indefensible given the increasingly popular view that values, however difficult to define and measure, play a more central and more dynamic role than attitudes within a person's cognitive-effective system. Thus to better understand such human behavior as conflict, Professor Rokeach would argue that initially it is necessary to better understand the nature of values and 19Ibid. 19 value systems. The vital importance of the concept of values is evidenced in the following quotation: It is difficult to conceive of a human problem that would not be better illuminated if relevant and reli- able value data concerning it were available. Differ- ences, for example, in culture, social class, occu- pation, sex, religion or politics are all equally translatable into questions concerning differsnces in underlying values and value systems . . .2 If it is assumed that values are important determinants of individual behavior, questions that immediately come to mind are--"why" and "how" do values actually affect behavior? Psychologists tell us that man lives in a universe of events and objects which do not have intrinsic meaning for human experience and behavior.21 Rather, man records past perceptions which contain infor- mation concerning the events he experiences. These past perceptions are then formulated into categories of infor- mation which take the form of values (or attitudes) about events.22 These categories, in turn, guide or filter future perceptions and interpretations of encountered 20Milton Rokeach, "The Measurement of Values and Value Systems," pre-publication copy, Michigan State Uni- versity, Department of Psychology reprint (Winter, 1970), p. l. Hereafter referred to as Rokeach, "Measurement of Values." 21George A. Kelly, A Theory of Personality (New York: W. W. Norton, 1963), pp. 105-84; Wilbur Schramm, The Science_of Human Communication (New York: Basic Books, 1963): pp. 54-61. 22Kelly, Theory of Personality, pp. 45-56; Schramm, Science of Human Communication, pp. 61-75. 20 objects. In short, man's attitudes and values determine his view of the world. Values and value systems are important for at least three reasons. First, information an individual gains as a result of an encounter with one object should apply to other objects in that category. This suggests that an individual can make an inference about his relation- ship to an object without encountering that object, but simply by encountering another object which he conceives to be in the same class as the first. Second, such cate- gories provide individuals with expectations about those objects that they believe to be members of the same cate- gories. Values can direct perception, causing an indi- vidual to notice certain characteristics of an object and to react to the object on the basis of those character- istics. Third, when a person develops a system of values, it opens up for him channels of choice along which he is able to move. But without such a system of categories to store information, man cannot effectively develop a differentiated repertoire of responses; he cannot long for things, for he has no idea of what he is missing, nor even any sense of missing at all. A system of values constitutes a ready-made format for future thinking and habitual responses.23 A system of values thus serves as a "frame of reference." 23George Miller, Eugene Galanter, and Karl H. Priebram, Plans and the Structure of Behavior (New York: Holt, 1960), PP. 21-24. Ib’IIII'I II) I.. [I 4 I'll Ill-Ill L . 21 The importance of values and value systems as determinants of conflict can thus readily be seen. When two individuals, a group, a society, or a culture has (have) common values, men think in a similar manner, have common expectations about events, and respond to events in a similar manner. However, when two groups of indi- viduals have different values, then conflict "may" arise due to differences in perceptions, expectations, and re- sponses. Conflict need not always occur, however, even if two individuals or two groups have different values. Richard McKeon suggests that men can have differing values (opinions, attitudes) and still be of one mind.24 His point is that men can come to agreement and thus be of one mind if they attempt to understand and appreciate the values of others. Under such conditions men can be said to be of one mind without having common opinions, atti- tudes, and values. Related Research Much has been written about values, individual be- havior, and conflict. However, the author has been unable to find any study which links these three phenomena. In fact, there apparently is no research literature available even to explain the relationship between the two key con- cepts of values and conflict with which this study is 24Richard McKeon, "Communication, Truth and Society," Ethics (1956), 99. 22 25 Some literature is available specifically interested. which mentions the phrase "conflict of values" (or its equivalent, e.g., values in conflict); but in all cases discovered the concept of values is not defined and/or no actual attempt is made to measure the values involved in the conflict situation.26 Values have been discussed by many writers and researchers. However, values and value systems are usually defined in similar terms such as beliefs and attitudes. These attitudes and beliefs are often claimed to determine an individual's orientation or frame of reference. Values, 25This conclusion has been reached after thoroughly examining and reviewing: (1) various books and periodicals listed in the Michigan State University card catalogue under the topics of "values," "conflict," and ”behavior"; (2) the numerous past editions of Dissertation Abstracts; (3) the past six editions of BurosTTMental Measurement Yearbook; (4) several conversations and discussions with various professors of psychology and other disciplines; and (5) the results received from DATRIX (Direct Access to Reference Information: A Xerox Service) which was to report any studies which dealt collectively with the two topics of "values" (social) and "conflict." 26Actually, often when the phrase "conflict of values" has been used in the past, the writer meant con- flict of attitudes for such studies virtually always dealt with attitudes but, for the sake of verbal diversity, the authors occasionally mentioned the word values and thus appeared at times to use the terms interchangeably, that is, as synonyms. Authors often spoke of "conflicting atti- tudes and values" but in explaining such a phrase, only the affects of attitudes were actually mentioned. The preced- ing explanation assumes that the distinction between atti- tudes and values as set forth in the "Definition of Terms" section of this chapter is valid. It must be recognized, however, that other authors have less clear cut definitions of such terms and apparently, also, a few writers consider the terms to be interchangeable. 23 however, are not usually considered to be motivators or determinants of individual behavior. Most supposed experts in the area believe that values can only indirectly in- fluence one's behavior. Much of the research done in the area of values centers around the measurement rather than the determination of individual value systems. Little re- search has attempted to relate values directly to indi- vidual behavior. No research has attempted to relate the concepts of values and conflict. Hopefully, this study may help to fill the academic void. Individual behavior has been explained in almost an infinite number of ways. Some current and popular attempts to explain individual behavior have used such concepts as "needs," "drives," or "expectancy-values." The "needs" approach claims that any and every individual has various needs. An individual allegedly acts or be- haves in a fashion which will satisfy these needs. These needs are usually, in the management literature, classi- fied in a hierarchy similar to the following: physio- logical, safety, love-social, self-esteem, and self- 27 actualization. Under one theory, an individual initially 27A. H. Maslow, Motivation and Personality (New York: Harper and Brothers, 1954);TFA Theory of Human Moti- vation," Psychological Review, L, No. 4 (September, 1943). In addition, see the works of Chris Argyris, Personality and Organization (New York: Harper and Brothers, 1957); Integrating the Individual and the Organization (New York: John Wiley and Sons, Inc., 1964); Understanding Organi- zational Behavior (Homewood, Ill.: The Dorsey Press, 1960). 24 attempts to satisfy basic needs (physiological, safety) after which he tries to fulfill higher level needs (self- esteem and self-realization). Thus "needs" allegedly can explain individual behavior. Others claim that needs cause "drives" and it is really this drive which is at the heart of the explanation of human behavior.28 Man attempts to eliminate or alleviate this drive and in doing so acts or behaves in a manner characteristic of human beings. There are many theoretical variations of the 29 Common to all such theories "expectancy-value" theory. is the notion that the strength of the tendency to act in a certain way is a product of the strength of expectancy that the act will be followed by a given consequence and the relative importance of that consequence to the indi- 30 vidual. Some authors refer to the "valence of the goal," others refer to this concept as the "demand for the goal," 28K. B. Madsen, "Theories of Motivation: An Overview and a Synthesis," in Human Motivations: A Symposium, ed. by Marshall R. Jones (Lincoln: University of Nebraska Press, 1965), pp. 49-70; Theories of Motivation (Copenhagen: Munksgaard Press, 1959), passim. See also C. N. Cofer and M. H. Appley, Motivation: Theory and Re- search (New York: John Wiley and Sons, 1964). 29John W. Atkinson, An Introduction to Motivation (Princeton, N.J.: D. Van Nostrand Company, Inc., 1964), p. 274. 301n the management literature, a popular version of this approach can be found in Victor Vroom, Work and Motivation (New York: John Wiley and Sons, Inc., 1964). 25 the "utility of the consequence,‘ or as "the product of motive and incentive."31 Research examples of "needs," "drives," and "expectancy-value" approaches to explain individual be- havior are too common and too numerous to mention specifi- cally. These approaches are the more popular attempts to explain individual behavior today. In addition to "needs," and "expectancy-value" theories are numerous 32 "drives,' "interaction" and/or "systems" theories of behavior. Other approaches and theories about behavior are also available but have not been mentioned here.33 Those approaches examined have been mentioned so as to give the reader some feel for the current state of affairs in moti- vation theories. Such theories are only indirectly related to the subject matter of this thesis, and, consequently, will not be examined in any greater detail. Conflict, similarly, has been written about and researched extensively in the literature, but conflict is usually studied and explained in terms of conflicting needs, 31Atkinson, Introduction to Motivation. 32In the management literature, popular versions of such theories can be found in George C. Homans, The Human Group (New York: Harcourt, Brace and Company, 1950), pp. 81-155; and in William Foote Whyte, Man and Organizations (Homewood, Ill.: Richard D. Irwin, Inc., 1959). 33See the works of Atkinson, Introduction to Moti- vation; Madsen, Theories of Motivation; and Cofer and Appley, Motivation: Theory and Research. 26 expectations, roles, strategies, goals, interests, ob- jectives, or beliefs and attitudes.34 No study that this researcher is aware of attempts to explain the conflict phenomenon in terms of conflicting values. Conflict can be and is also studied along many dimensions usually labeled as intrapersonal conflict, inter- personal conflict, conflict between or among groups, or conflict of an individual or a group with the organization itself as an entity. Intrapersonal conflict, that is, con- flict which is internal and within any individual human being, is the province of psychologists and psychiatrists, and accordingly, will not be dealt with in this study. Because organizations are composed of groups which in turn are comprised of individuals, the key dimension of conflict is interpersonal conflict, that is, conflict between one individual and another. This study is concerned with in- vestigating and analyzing interpersonal conflict. The present study will contain three unique aspects. First, the study will focus only upon interpersonal con- flict. Second, conflict will be (tentatively) attributed to differing values and value systems. And third, the study will attempt to show the degree of interrelatedness 34In the management literature, for examples, see Dalton E. McFarland, Conflict and Cooperation in Personnel Administration (New York: American Foundation for Manage- ment Research, American Management Association, 1962); or Maynard Toussaint, "Line-Staff Conflict--Its Causes and Cure," Personnel, XXXIX, No. 3 (May-June, 1962), 8-20. 27 of the three phenomena of values, individual behavior, and conflict. CHAPTER II GENERAL PROCEDURE AND METHODOLOGY Introduction The purpose of this chapter is to explain the general procedures and methodology used in both the setting up of the study and in the subsequent gathering of the data. Initially, a discussion of the "evolution of the present research study" is presented in order to demon- strate how the current study is actually an adaptation of several former research proposals. This discussion not only places the current research project in its proper historical perspective, but it also points out areas in which a similar type of research project might be under- taken. After the "evolution of the present research study" has been presented, a discussion of the current research "design" will follow. Both a verbal and an illustrated explanation of the design will be presented. The "popu- lation and sample" from which the research information has been gathered is then discussed after which is presented a discussion of the "data and instrumentation" employed when gathering information from sample individuals. 28 29 Two important information gathering instruments were used in this study. The first device is the interview- questionnaire which appears in Appendix A and which will be explained and examined in the "data and instrumentation" section of this chapter. The second instrument is the Rokeach Value Survey which is briefly described in this chapter and which appears in Appendix B of this report. A pilot study was initially undertaken in order to explore the feasibility of the current study. This "pilot study and subsequent changes in the initially planned pro- cedures" will be discussed in the final section of this chapter. The pilot study description is presented at the end rather than at the beginning of this chapter because the subsequent changes resulting from the pilot study are best understood after rather than before, the reader has obtained an adequate initial comprehension of the "design," "population and sample," and "data and instrumentation" utilized within the current study. Evolution of the Present Research Study The current study grew out of the author's con- viction that values and value systems are possibly a better explanation of human behavior--especially if we are considering adult, individual behavior rather than child or group behavior--then are the theories and concepts of 30 ' or “systems." This "needs," "drives,“ "expectancy-value,‘ conviction was instilled when the author asked himself the following question: "Why do you, as an adult human being, behave in the manner that you do?" Granted that "needs" and "systems" and other variables and theories explain some types of behavior given certain circumstances; but the bulk and vast majority of adult individual behavior, at least in this author's estimation, could be determined by and are de- pendent upon one's values. To try to test this idea, the author looked for a topic in the management literature (since management is this author's major discipline area) which would lend it- self to a study testing the importance of values as a determinant of behavior. The concept of "conflict" pre- sented itself and seemed to fulfill all the necessary re- quirements. The main focus initially was to illustrate the relationship between values and behavior; conflict is a type of behavior, accordingly, the author hypothesized a relationship between values and conflict. Indeed, it will be noted that the emphasis of this study here began to switch from being a study of the relationship between values and behavior, to being more specifically a study of the relationship between values and conflict behavior. Several preliminary research proposals grew out of a desire to investigate this alleged value-conflict relationship. The first such proposal tentatively in- volved analyzing conflict between and among various 31 hierarchical groups within a large-scale organization. This proposal was designed to look at conflict between departments, conflict between the first line supervisor and his boss, conflict between line and staff positions, and any other type of conflict which might present itself be- tween one level of an organization and another level of the same organization. After discussions with various faculty members, it was decided that such a study was too broad in scope and that the exact study matter ought to be more narrowly and distinctly defined. The second research proposal was designed so as to concentrate only on labor versus management conflict rather than conflict at various levels within, between, and among various individuals and groups in a large-scale organi- zation. The intent was to study situations where labor- management conflict existed and w ere it was nonexistent, and to attempt to explain such phInomena via different and similar values and value systems. Whereas the hierarchical study could have been made within one large organization, this study would necessitate the examination of several labor-management situations. Again, this proposal was criticized as being too broad in scope, somewhat difficult to implement, and perhaps somewhat overly optimistic (optimistic in the sense that the researcher expected to show labor-management conflict via values when all modern writers and experts in this field were using a systems approach to explain such a complex, dynamic relationship). 32 The suggestion was again made to more narrowly and dis- tinctly define the problem area and subject matter. The suggestion to concentrate on inter-individual behavior (conflict) rather than upon group or organizational be- havior (conflict) was made. This marked the second major shift in the author's final research design; the first major change was to shift the emphasis from the study of the relationship between values and behavior to a study of the relationship between values and conflict. This second shift in emphasis was a change from the concern for con- flict (behavior) in general to a concern for only inter- individual conflict (behavior). Thus it was decided that the more narrowly defined subject matter area would now be individual versus indi- vidual conflict situations rather than group or organi- zational conflict problems.1 It was also decided at this point that the focus of the study should be interpersonal conflict situations rather than intrapersonal conflict situations. Intrapersonal conflict (that is, internal 1Other types of conflict situations such as indi- vidual versus group, individual versus organization, group versus group, group versus organization, or organization versus organization were also considered but rejected, basically again because they were rather broad in scope. It was rationalized that since organizations are comprised of groups which in turn are comprised of individuals, the central focus of the study should be the individual. After a better understanding of individual versus individual conflict situations has been obtained, then it will be possible later to branch out and apply such findings to other more complex types of conflict situations. This matter will be discussed in greater detail in Chapter V of this report. 33 conflict within any one individual) might perhaps also be explainable in terms of conflicting values, but, neverthe- less, intrapersonal conflict could not and will not be dealt with directly in the scope of this report. The main subject matter at this point has now been reduced to a study of interpersonal conflict (that is, conflict be- tween one individual and another individual). The third tentative research proposal was to be a study investigating the relationship of purchasing agents with other members of an industrial organization. The purchasing agent often has conflict with numerous other positions. Hopefully, such a study could deal with and add insight into the persistent managerial problem of line versus staff conflict. This study could have been carried out without too much difficulty and would have perhaps pro- vided the necessary information needed to investigate the author's convictions about values and conflict. However, this study was not undertaken for a number of rather un- related reasons, the most influential of which was that another research proposal seemed to be a better alter- native.2 2Another consideration was the possibility of re- ceiving financial backing from a national purchasing association if a definite proposal could be submitted be- fore a certain date. This deadline could have been met probably, but an unfavorable written comprehensive exami- nation result made it necessary for the author to re- schedule the allocation of scarce studying and research time so that the possibility of receiving research funds for this project was forfeited. An additional complication 34 A fourth tentative study was to investigate marital conflict situations within the Michigan State University married housing units. This alternative seemed more favorable than the purchasing agent alternative because the estimated cost would be less and thus the project could be financed by the researcher, and also the time lag associ- ated with applying for the financial assistance could be avoided. In addition, the time lag associated with getting the necessary information from the purchasing agents and their co-workers could be avoided since with the married housing alternative, the researcher could obtain the neces- sary data virtually as soon and as fast as he was willing to put forth the necessary effort. However, this proposal was also subsequently abandoned in favor of a better one. The main drawbacks of this proposal were: technically, the sex variable, male versus female, would influence values and conflict situations and thus would somehow have to be accounted for in explaining the relationship between values and conflict;3 and practically, this researcher would have was that there also seemed to be some question about whether the type of research that the author wanted to do would be of economic utility to the national purchasing association and thus there was no assurance of financial support. 3Individuals might have different values and value systems simply because of their sex (inter alia). Accord- ingly, a married couple might have different values and value systems merely because one of them is a man and the other is a woman. An underlying assumption of this study is that "birds of a feather flock together." The converse of this assumption is that "opposites attract." This study 35 felt uncomfortable trying to uncover marital conflict situ- ations, and, also, the number of couples willing to volun- teer such information to an uninvited outsider was, accord- ing to this author's estimation, somewhat limited. This proposal, however, brought about a third major shift in emphasis involving a shift from an industrial-working setting to a habitat-living environment. The research proposal eventually chosen involved analyzing conflict and nonconflict college roommate situ- ations.4 To facilitate the gathering of information and to avoid distractions, the decision was made to analyze only male college roommate situations. Specific aspects of the current research design will be discussed in length in the next sections of this chapter. To summarize the "evolution of the present research study,‘ we have seen how the current research project evolved out of former research proposals dealing with the conflict situations of a hierarchical organization, labor will not deal with these assumptions as they relate to married couples, but we might suspect that the "opposites attract" theory is relatively and probably more meaningful in explaining marriage partners than in explaining same- sexed college roommates. Absolutely, however, we might suspect that the "birds of a feather flock together" theory is the most meaningful of the two theories as an expla— nation of both married-couple and same-sexed college room- mate attraction. 4From this point on, if not already so perceived, the reader should assume that when reference is made to college roommates, these roommates are of the same gender. All roommates studied were males. 36 versus management, purchasing agents, and married couples.5 During this evolution, three major transitions appeared in changing the focus of the study. First, a change from being concerned about the relationship between values and global behavior to a concern about the relationship between values and conflict. Second, a change from a concern for conflict in general, to a concern for only inter-individual conflict. And, third, a change in focus from an industrial- working setting to a habitat-living environment. Design This study analyzes the values and value systems of male college roommates who have lived in conflict or nonconflict situations. Before focusing finally upon a study of interpersonal conflict, the research project went through the aforementioned evolutionary stages. After the study of conflict at the interpersonal level was chosen, other changes occurred both before and during the actual collection of data. Those changes occurring before the data gathering began will be described now but changes occurring after the data gathering began will be described in the section of this chapter entitled "pilot study and subsequent changes in the initially planned procedures." 5Actually, quite a few other research proposals at _one time or another were considered but none of them were planned out to any great extent and thus were not mentioned in the above explanation of the evolution of the current study. 37 The decision to look at interpersonal conflict in terms of male college roommates was easy enough. How actually to go about getting the necessary information, however, was a much more difficult problem to solve. The researcher talked with various academic and administrative individuals within the educational hierarchy and then eventually and originally planned to use the following procedures. Conflict could be identified by referring to the reasons given by individuals for desiring to change roommates. The resident assistants and/or the managers' offices of all the male dormitory complexes on campus keep a record of all room changes which have occurred during the last three years. After a period of three years such records are destroyed. The records are in the form of "room change cards" and one question on the room change card asks why the request for a room change is being made. According to estimates by "Head Advisors" of several dormi- tory complexes, roughly 30 per cent of individuals request- ing a room change do so because of conflict with one's current roommate. Thus, conflict information could have been obtained in this manner. A measurement of values was initially planned to have been obtained by looking at three-year-old Value Surveys which current juniors had filled out when they were freshmen. However, many complications developed which pre- vented this research design from materializing further. First of all, the Dean of Students at Michigan State 38 University thought that for the researcher to look at a student's room change card without the student's per- mission would be a violation of the student's rights. This problem could have been overcome by getting the students to sign a statement allowing the researcher to use such infor- mation. Another major problem was that the three-year-old Value Surveys also could not be at the researcher's dis- posal unless the proper approvals were obtained from the researchers and sponsors of the former study. This probably could have been obtained but an unusually large amount of red tape and loss of time would have been involved. Another predicted obstacle was that even if the researcher discovered that a conflict situation existed between persons A and B, and even if a Value Survey for either person, say person A, was available; these conditions would not assure the researcher that a Value Survey for per- son B was also available; it also would be impossible to determine how person B felt about the room change since only one room change card per roommate pair was required in order to obtain permission to change roommates. To find out how person B felt, the researcher would have had to contact him in order to ask him about a three-year-old room change situation. Evidence reveals that after three years, approximately only one-half of the original freshman class is still enrolled as students. These and other complications led to the abandonment of the original room- mate study design. 39 The ultimate design utilized entirely first-hand information. It eliminates the types of problems en- countered with the original interpersonal conflict room- mate study design, and, in addition, it serves better to provide the type of information needed in order to fulfill the stated objectives of this study. Schematically, the research design is as shown in Figure l. , Set A , CONFLICT NONCONFLICT GR UP GR P I T T I I I l ' 25 pafirs 25 pairs 25 pairs 25 pairs I I Set B Set C Figure 1. Design of Research Two groups of roommates have been identified, a conflict group of roommates and a nonconflict group of roommates. Complete data on 200 students were obtained. Fifty roommate pairs (100 students) belong to the conflict group, and fifty roommate pairs (100 students) belong to the nonconflict group. Set A is the "entire sample"; Set B is the "original sample"; and Set C is the "Cross- Validation" sample. Set B and Set C are of equal size (twenty-five pairs of conflict roommates and twenty-five 40 pairs of nonconflict roommates) and collectively they com- prise Set A (the entire sample). Computerized data analyses in the form of "routine oneway" and/or "routine stable" which will be explained in greater detail shortly have been obtained for all three "sets" (samples) of roommates. The data to be reported consist of information about both "individual" members in each group and "pair" members in each group. The design of the research permits comparisons and contrasts to be made for values and value systems of both "individuals" and "pairs" from either (both) the conflict or the non- conflict group.6 Data about the "entire sample" will initially be reported. The entire sample was split in order to test various hypotheses and to attempt to formulate a "prediction process." Computerized routines were performed upon the "original sample" and the results later were applied to the "cross-validation" sample. More specifically, the findings and relationships revealed from analyzing the data of Set B were used to "predict" whether roommate pairs of Set C belonged to either the conflict or the nonconflict 7 group. 6The exact nature of the analysis of the data will be explained fully in the next two chapters. 7Specific "findings" will be reported in Chapter IV of this report. 41 Population and Sample The population consists of male undergraduate college students living in the on-campus dormitory com- plexes of Michigan State University. Only male students were asked to participate as it was easier for the re- searcher, a male, to gather only information from said gender. The information was gathered initially from the population by means of personal interviews, and the limited open hours in some of the women dormitories would have been an obstacle to such interviews. The main reason for not including female participants, however, was that going door-to-door doing personal interviewing in a women's dormitory would have been an embarrassing procedure for the researcher involved. Only undergraduate students were examined because usually graduate students do not live on campus and of those who do, their number is small and their time is scarce. Only on-campus students were asked to participate for two basic reasons. First, they are all in locations which are easily accessible to the researcher. Second, Value Surveys could be returned via campus mail from on-campus dormitories thus saving the cost of return postage. The vast majority of students interviewed were living in one of three dormitories.8 These dormitories were chosen because they are all close to one another; they 8Case, Wilson, and Wonders Halls. 42 are close to the researcher's home; they have been subject to fewer questionnaires than have other dormitories re- cently; and they house basically upperclassmen. Upper- classmen were desired because they have been on campus for a longer period of time than have underclassmen, and as such, they were the students who most likely would have had the experience of having had a roommate in a conflict and/or nonconflict situation. Before interviewing began in any of these three major dormitory complexes, the researcher met with the Head Advisor and/or the resident assistants of each of these dormitories so that the students could be notified in ad- vance that someone would be in their dorm interviewing students as part of a research project. Students who did not want to participate obviously did not have to.9 Some students in other dormitories besides these three main ones were also interviewed. Often an individual would identify a former roommate as a person with whom he had experienced conflict or nonconflict. If this was the case, and if this former roommate was still living on campus, he would also be interviewed by the researcher 9This solved the problem of student "rights" which complicated an earlier research design. If a student agreed to be interviewed, then the researcher, using discretion, could use the information thus obtained for research purposes. The Dean of Students at Michigan State University believed that the agreement by a student to be interviewed tacitly and indirectly gave the researcher the "right" to use such information. 43 (although usually this was a "telephone interview" which will be explained later in greater detail). A total of 1,200 students were interviewed.10 How- ever, the final analysis of the data included information from only a sample of 200 students. To understand fully the reasons why only completed data were obtained and analyzed from 200 students when 1,200 students were interviewed, we must combine our explanation of the "population and sample" with a discussion of the "data and instrumentation" used in this study. Data and Instrumentation The main goal during the data collection period of this research was to find fifty pairs of students who could satisfy the criteria of the conflict group and fifty pairs of students who could satisfy the criteria of the noncon- flict group. The criteria for admittance into the conflict group were as follows: 1. A student must answer "yes" to the following question: "Have you had any roommates at MSU with whom you did not get along very well, that is, with whom you have experienced a high de- gree of incompatibility?" 10The exact total was 1,194. 44 2. A student must have lived with the roommate in question for two academic quarters or less. 3. The living arrangement or setting in which the incompatibility existed must have occurred within living accommodations shared by a total of four students or less. 4. The (former) roommate that the student has "in mind" and with whom he has experienced incom- patibility must still be a student living on campus in the university sponsored dormitory system. 5. The student must identify by name the incom- patible roommate he has "in mind." 6. The student must answer "yes" or "yes and no" to the following question:11 "Would you say that the primary reason that you are no longer rooming with him is because you two could not get along very well together?" If a student fulfilled all the above conditions, he became a "potential" conflict pair group member. Before a "potential" member became an "actual" conflict group mem- ber, two additional requirements were necessary: 1 1 II II ' A yes and no answer, wh1ch was common, meant that this statement along with another reason or two were jointly the cause of the separation, and that the relative importance of each reason was about the same. 45 7. A student's pair-mate, that is, the individual identified in fulfilling condition number 5, must "mutually" (although interviewed separ- ately) meet all of the above entrance require- ments also. 8. Both members of a pair must fill out (accord- ing to the instructions given) and return their respective Value Surveys. A pair became a member of the final group of fifty pairs of conflict roommates only if all of the above eight conditions were satisfied. Each of these conditions needs additional explanation. The first condition is that each student must answer "yes" to the following question: "Have you had any roommates at MSU with whom you did not get along very well, that is, with whom you have experienced a high degree of incompatibility?" This question contains many key phrases. First, anyone identified had to be at one time or another an MSU student. This was necessary to check out the "mutuality" condition to be explained later. Notice also that although these are potential conflict group members answering this question, the word conflict is not mentioned. This is because the researcher felt that the word conflict was too "strong." Many students asked about how they should interpret the word "high" and were told to interpret it in any manner they wished to. Accordingly, some students would answer the question "no" and others would answer it "yes." If the student answered 46 the preceding question "yes," the student was then asked if he had more than one person in mind and, if he did, he should then answer the remaining questions in reference to the most incompatible and/or the most recent roommate. The second entry condition is that a student must have lived with the roommate in question for two academic quarters or less. This condition was set up so that con- flict between any two roommates was more than merely "per- ceived" conflict. As noted in the interview-questionnaire (Appendix A), only roommates who actually separated or were in the process of separating were, accordingly, included in the conflict group. In the final sample of fifty pairs of conflict roommates, forty-seven pairs had already split up and three pairs were in the process of separating. Many students claimed that they had had incompatible roommates and the incompatibility was such that it led to their eventual separation. However, they were often not admitted into this conflict group because they had lived with their incompatible roommate for over two academic quarters. It was very common to find incompatible roommates who lived together an entire year and then separated. Such students were not included because the cutoff point was purposely set at two academic quarters or less so that the conflict was to be of such a degree that the roommates refused to live a full year together. It was assumed that if two individuals would tolerate each other the entire year, the case was unlikely to have been characterized by a high 47 enough degree of incompatibility for which the conflict group was intended. No minimum amount of living time was officially set by the researcher but, nevertheless, the researcher using his own discretion excluded individuals who experienced conflict situations lasting one week or less. Actually, this was a very limited number of cases for usually a student would not even remember his former roommate if he had only lived with him for that period of time. It should also be noted that only bona fide room- mates were included in the sample. Suitemates (that is, the two roommates next door with whom one shares lavatory facilities) were not considered to be bona fide roommates.12 Neither were students who actually lived in one room with an individual most of the time but who were officially assigned to different rooms. Such cases were very few in number. In addition it should be noted that the majority of conflict situations arose out of a situation usually at the beginning of a new academic year when often three stu- dents were assigned at least temporarily to a two-man room until a vacancy became available. 12Virtually all of the students interviewed were currently living in rooms designed for two individuals. However, many of them had had conflict with former room- mates when living in different dorms, some of which are designed so that four, five, or six students often would share the same living facilities. Another complication was that even within the dormitories designed with two-men rooms in mind with roommates and suitemates sharing a com- mon lavatory, on occasion four students would decide to sleep in one room and thus to have the other room serve as a recreation and/or study lounge. 48 The third entry prerequisite for admittance into the conflict group was that the living arrangement or setting in which the incompatibility existed must have occurred within living accommodations shared by a total of four students or less. This condition was initiated so as to prevent the analysis of cooperative housing living arrangements. It will be remembered that the focus of this study is upon interpersonal conflict, not group conflict or individual versus group conflict. Well over 90 per cent of the pairs included are from former two or three men roommate situations. It is estimated that over half of the conflict pairs were initially three-man roommate situ- ations. This entry requirement only eliminated one or two pairs of roommates who could meet all other group entry requirements except this one. The fourth entry condition was that the (former) roommate who the student has "in mind" and with whom he has experienced incompatibility must still be a student living on Campus in the university sponsored dormitory system. This condition was put in as a matter of economic convenience to the researcher. Since the former roommate had to be contacted in order to see whether or not the "mutuality" condition was met, it was decided to eliminate those students whose former roommates were now off campus or out of town. It would be uneconomical to seek out and interview these geographically dispersed former roommates in order to secure the necessary information. Often, also, 49 former roommates were now married or living in a fraternity and to go to such places claiming to be doing research for a study of on-campus roommates would be looked upon with suspicion. Such former roommates might have been able to guess that they had been identified by someone on campus, and the researcher wanted to prevent such a thing from happening. Enough sample individuals could more easily and more economically be obtained by merely interviewing more students until enough students were found on campus who met all of the entry conditions. Again, a large per- centage of students, estimated to be about 40 per cent, who would have met all other entry conditions were dis- qualified because of this criterion. The fifth entry requirement for admittance into the conflict group was that the student must identify by name the incompatible (former) roommate whom he has in mind. This information was necessary in order to satisfy the "mutuality" entry condition and also in order to find out to whom a Value Survey was to be issued. The supplying of this information by a student was often given only after a hesitation and/or an explanation was given to the student as to why this data was needed. This information, however, was almost invariably given (that is, only a few students refused to reveal their former roommate's name after the reason it was needed was explained, and also after confi- dentiality of all data was reassured to each student). 50 The sixth entry requirement was that a student must answer "yes" or "yes and no" to the following question: "Would you say that the primary reason that you are no longer rooming with him is because you two could not get along very well together?" The crucial word in the pre- vious statement is the word "primary." Many roommates split up and had at least some degree of incompatibility, but usually the "primary" reason for the subsequent separ- ation is some situation other than extreme incompatibility. Common reasons for splitting up were: graduation, one roommate joining a fraternity or desiring to move off campus, former roommates getting married, dropping out, flunking out, joining the armed services, getting drafted or moving in with other students who were closer friends, roommates changing dorms because of the location and/or different atmosphere of other dorms, and also individuals changing roommates just for the sake of variety. These and other reasons were usually found along with some de- gree of incompatibility as explanations as to why roommates split up. Thus, usually roommates who split up had more than one reason. However, to be included in the conflict group, the "primary" or one of the main reasons for the separation was to have been roommate incompatibility. This condition eliminated many potential conflict pairs, es- pecially because of lack of mutuality in response to this condition. 51 The seventh entry condition is the all-important requirement that any individual identified by another stu- dent as a person with whom incompatibility had existed, must also meet all of the entry requirements into the con- flict group. This entry condition actually served as a verification of the answers given by the first roommate in the potential pair. True vertification of responses was impossible since individual perceptions vary, but fic- titious individuals and outright lies could usually be de- tected.13 The key conditions to be mutually verified are conditions 1, 5, and 6 although questions about each condition were asked to this person.14 This entry require- ment disqualified approximately 50 per cent of all potential conflict pairs. Often one of the roommates felt incom- patible with the other, but when the other roommate was questioned he often felt neutral about the relationship, or if incompatibility existed, he would not agree that it was present to a "high" degree or that it was the "primary" l3 0 Such 1nstances were very, very rare and more or less could be assumed to be nonexistent. l4Condition number 5 is crucial in that on rare occasions all conditions up to and through condition number 7 would be "mutually" answered, but the cross-check of names in condition number 5 revealed that former roommates had identified other former roommates rather than having identified each other! Obviously, such "pairs" were not match-merged and as such could not be included in the final conflict group. Thus the condition of "mutuality" also means that former roommates had each other respectively "in mind" when answering interview questions and meeting entry requirements. 52 reason for the separation. The mutuality condition was initiated so as to attempt to get at actual rather than merely perceived conflict situations. If both roommates independently (that is, interviewed privately) and mutually agreed that conflict existed and that a separation actually resulted because of the conflict, actual as well as per- ceived conflict could be assumed to have existed. The final entry condition for admittance into the eventual fifty pairs of conflict roommates was that both members of a pair must properly fill out and return their respective Value Surveys. This again eliminated a few potential pairs. Later in this chapter exact figures and percentages revealing subsample sizes and rates of return will be presented. I In like manner, eight entry conditions existed and had to be satisfied before an individual or a pair could be admitted into the nonconflict group: 1. A student must answer "yes" to the following question: "Have you had any roommates at MSU with whom you have gotten along very well, that is, with whom you have experienced little or no incompatibility?" 2. A student must have lived with the roommate in question for four full academic quarters or more. 3. The living arrangement or setting in which the compatibility existed must have occurred within key words. 53 living accommodations shared by a total of four students or less. The roommate that the student has "in mind" and with whom he has experienced compatibility must still be a student living on campus in the uni- versity sponsored dormitory system. The student must identify by name the com- patible roommate he has "in mind." The student must answer "yes" or "yes and no" to the following question: "Would you say that the primary reason that you two have (had) roomed together so long is because you two get (got) along very well together?" A student's pair-mate, that is, the individual identified in fulfilling condition number 5 above, must "mutually" (when interviewed pri- vately and separately) meet all of the above entrance requirements also. Both members of a pair must fill out (accord- ing to the instructions given) and return their respective Value Surveys. Many of these conditions need further elaboration. In entry condition number 1, again the question contains The types of problems involved with the inter- pretation of the analogous question for the conflict group are once again encountered here. Former or current room- mates had to be from MSU. Students had difficulty often in 54 interpreting the phrases "very well" and "little or no in- compatibility." Notice also the absence of the word non- conflict. Once again students often identified more than one person and if they did they were asked to consider the most compatible and/or most recent roommate. Entry condition number 2 requires a student to have lived with the roommate in question for four full academic quarters or more. A fourth quarter together usually means the start of a second year together, and if two individuals have roomed together that long, it was assumed that this 15 It should could be a physical measure of nonconflict. also be noted that these four quarters had to be successive academic quarters (possibly and usually excluding summer quarter). The final sample pairs in this group were com- prised of roommates who had lived together anywhere from four to eight full academic quarters, the average estimated to be slightly greater than five quarters of consecutive living together. No roommates were found out of the 1,200 students interviewed who had lived together with the same person for over eight quarters and who met the other conditions of entry (the crucial one here being on campus residency). 15A few roommate pairs made the four full quarters together cutoff by going to school and rooming together during the summer thus one academic year rather than two academic years was involved. 55 The basic rationale for entry condition number 3 (maximum four-man room requirement) was explained earlier for the conflict group and the same reasons and expla- nation given previously are applicable now for the non- conflict group. Again, similar to the conflict group situ- ation, this condition virtually eliminated no one who was otherwise qualified for entry. Almost all of the non- conflict pairs (probably about 95 per cent) are from two- man room situations.16 The remainder being from three- or four-man living arrangements. The fourth nonconflict group entry requirement was the on-campus residence requirement whose basic rationale was explained earlier also. As explained earlier, this condition arose out of economic necessity. In sharp con- trast to the same condition as an entry requirement for the conflict group, this condition eliminated very few potential nonconflict group pairs. In fact, probably 90 per cent or more of the roommate pairs included in the non- conflict group were individuals who were currently (during the time of the study, that is) rooming together. The fifth entry requirement is the identification by name condition. Again, in sharp contrast to the 16However, often a temporary third man was involved due to the university practice of frequently having tem- porary three-man rooms (in rooms that were actually de- signed with only two roommates in mind) at the beginning of each year and later correcting the situation as the year progressed and students left the dormitory system for one reason or another. 56 conflict group, there was no hesitation in supplying this information since the roommate situation is or had been a friendly rather than a hostile one. Usually all that was required was for a student to mention the name of his cur- rent roommate. This entry requirement did not eliminate any potential nonconflict group members who were qualified otherwise. The sixth entry requirement for admittance into the nonconflict group required a student to answer "yes" or "yes and no" to the following question: "Would you say that the primary reason that you two have (had) roomed together so long is because you two get (got) along very well together?" Once again, in marked contrast with an analogous question in the conflict group, this condition did not eliminate very many individuals or pairs. Only about five potential pairs failed to meet this condition. Such pairs stated that they had roomed together so long because they never saw each other, they were from the same high school area and thus often went home during weekends together, or because one or both of the students was (were) not willing to go through the red tape and/or physical effort required by any room change. For the vast majority of nonconflict group roommates, the primary reason that they stayed together so long was because they enjoyed each others' company. The seventh entry requirement is the mutuality condition. The use of the mutuality condition as a 57 validation and cross-checking technique was explained earlier. Still once again, in tremendous contrast to the same requirement for conflict group members, this pre- requisite virtually eliminated no one from the nonconflict 17 Almost group who could meet the other entry conditions. invariably, if one roommate claimed that their relationship was a compatible one and the relationship was one lasting four full quarters or more, not surprisingly, the other roommate would agree (when interviewed privately) that the relationship was of the nonconflict variety. The final requirement for admittance into the non- conflict group was that both roommates had to properly fill out and return the Value Surveys issued to them. Ob- viously, the entire effort to find students meeting the first eight conditions was all in vain unless condition number 8 was also satisfied. This condition eliminated a few potential pairs and statistics revealing the figures and percentages involved will be presented later in this chapter. So much for the entry conditions for admittance into the two groups. It is necessary now to look in greater detail at the interviewing procedure, the inter- viewing-questionnaire, the use of the Value Survey, the students' rate of return of the Value Survey, and the employment of follow-up procedures. 17The author recalls only two such cases. 58 The interviewing technique was developed to pro- vide the easiest method of gathering the information neces- sary to begin the research project. The researcher planned to establish a good working relationship with the students so that their cooperation could be obtained. The re- searcher simply began by going from door-to-door in all the halls in each of the dorms and talking with any and all students who were in their rooms at the time. The re- searcher began each interview with a personal (by name) introduction of himself, and an explanation that he was a Ph.D. candidate in the Department of Management who was conducting a survey in order to gather data for an eventual dissertation. The researcher then went on to mention that he was doing a study of male, college roommates and that the study had been approved by the Dean of Students, their Head Advisor, and their resident assistant. The author mentioned also that his main objective was to analyze why some roommates do, while other roommates do not get along very well. The interviewer then asked them if he could take a few minutes of their time and ask them a few questions and thus give them the opportunity to partici- pate in the study if they cared to. Students were then assured that although the Dean of Students, the Head Ad- visor, and the resident assistants all knew that this author was doing the study, no one except the author would 59 see any of the information which the students revealed.18 The above introduction was delivered in such a manner that, although it was basically memorized, the author attempted to be as sincere and as appreciative of their participation as possible. Only four students refused to participate when contacted initially, but many would and/or could participate only if the interviewer would return at a more convenient time. Each dorm repeatedly was visited until the researcher had interviewed almost everyone in the dorm before he went on to a different dorm. Sometimes this meant knocking on the same door six times or more before finding a certain roommate at home. After a few days of interviewing, students were anticipating the arrival of this researcher and many of them looked forward to being interviewed. One problem area was that often many students would be in the same room and after introductory statements, this author would mention that it would be advantageous if he could talk to each person alone, that is, one at a time. On occasion, this caused resentment, but, usually the stu- dents consented without much hesitation. If the students seemed to resist the idea, the interviewer offered to come 18Most of the students commented that they did not care who saw the data. I assumed that if they were willing to tell me about former roommates, they would be willing to tell anyone else also. The only concern that was sometimes mentioned was that a few students did not necessarily want their former or present roommate to know how they had answered the questions, especially if the person being interviewed was a potential conflict group member. 60 back at a time when it could be more convenient for them in order that he might talk with them individually. Such a tactic proved to be successful. Another problem was that frequently girls were in the rooms and the request to talk with a student privately under such circumstances usually proved to be an awkward situation. Without hesitation, the interviewer again offered to return at some other time. The "girl in the room" problem was intensified not only because of the winter weather but also because the researcher found that evenings and weekends were the best time to interview since most students were in their rooms at this time. Actually, each interview began as a screening interview. Although this researcher talked with a total of some 1,200 students, he often talked with as many as fifty students in one day. Certain interviewing short- cuts were revealed so that the author could within a minute or two decide upon whether or not any student was a potential member of either the conflict or the nonconflict group. At this point an understanding of the design of the questionnaire is necessary in order to comprehend the interviewing technique that the author utilized. A copy of the interview-questionnaire appears in Appendix A of this report. In the upper left-hand corner of the interview- questionnaire form is an outline of the introduction state- ments that this researcher went through with each person 61 contacted--as explained above.19 The information in the upper right-hand corner is simply a method devised by the author to denote student cooperation. Essential identifi- cational information then follows in the form of data con- cerning a student's name, student number, room number, etc. Part II of the interview-questionnaire concerns the nonconflict group whereas Part III concerns the conflict group. The researcher purposely planned initially to ask questions about nonconflict roommates so as to establish some interviewer-interviewee rapport before going on to more personal questions concerning possible roommate con- flict situations. Part IV of the interview-questionnaire form is an outline of the comments the researcher used to conclude an interview. No matter how many questions an individual answered or how brief or lengthy the interview, each student was thanked for his cooperation, and often, when an interview was very short (reasons for which will be explained shortly), a brief explanation was given to the student as to why it would not be necessary to ask him any additional questions. Before ending the longer interviews (the reasons for lengthy interviews to also be explained 19Each interview, although initially planned to be of the "directed" type, turned out to be at least to some extent "nondirected." Often the researcher had to be an attentive listener for quite some time before a student eventually got around to answering a "directed" question. Often, also, a student would ask me why I needed infor- ' mation of one type or another, and in such cases I would endeavor to explain the basis for the question without revealing exactly what I was going to do with the data. 62 shortly), students were told that they might be contacted again, and if they were, it would involve not asking them more questions, but instead it would involve the filling out of a Value Survey which was explained as being a rather game-like survey which only required about fifteen minutes to complete. Also, these students were told that if they were contacted again, it was because they were randomly selected, but in actuality they were recontacted if the mutuality entry condition was satisfied. If they eventually did receive a Value Survey, they were instructed to fill it out as soon as possible and to return it via campus mail to the researcher's office. The services of the campus mail were to be utilized, it was explained to students, so that the cost of postage could be averted.20 We now want to examine Parts II and III of the interview-questionnaire more thoroughly in order to obtain a better understanding of what actually transpired during an actual interview. Both Parts II and III are actually a series of screening questions designed to either get the necessary information required or to end the interview as quickly as possible. 0Fourteen students were to eventually return the Value Surveys via the United States mail rather than by campus mail. Thus, the goal of avoiding all costs of re- turn postage was not entirely accomplished. The slight mix-up was due to student confusion and lack of knowledge about the operation and procedures of the on campus mailing system. 63 In order to understand this screening process, the questions in Part II of the interview-questionnaire form must be examined (see Appendix A). These questions relate to potential nonconflict group members. The first screen- ing question is question II-l, which asks a student if he has had any roommates with whom he has experienced little or no incompatibility, that is, someone with whom he has gotten along very well.21 If a student answered " no" to this question, then all of the remaining questions of Part II could be eliminated. If he answered "yes" to this question, the researcher then went on to question II-4, skipping questions II-2 and II-3 temporarily and usually also permanently.22 Question II-4 inquired about the length of residency condition. If this question were answered with data revealing that the student had lived with the roommate in question for less than four full 21Actually, if a student was a freshman (known from identificational data in Part I of the interview-question- naire), all of the questions of Part II could be eliminated since being a freshman, a student would not have lived with another student for four academic quarters or more. Al- though the three dorms in which most of the research was done were basically upperclassmen dormitories, nevertheless, many freshmen lived there also. 22The pilot study which will be explained later in the chapter revealed that a certain sequence to the questions could be thus arranged so as to serve as a screening process. The first few students interviewed in the study were asked the questions in the order that they appear on the interview-questionnaire but the rela- tive superiority of the sequence-screening process became immediately apparent. 64 academic quarters, the researcher could then skip the re- maining questions in Part II. If the answer to this question revealed that the student had lived with this roommate "in mind" for four full academic quarters or more, the interviewer then went on to question II-S. Rather than proceeding through this step-by-step screening process in greater detail, the reader should simply note that information about the necessary entry conditions for the nonconflict group was obtained from key questions in the screening process. As soon as a person failed to meet any of the entry conditions in any of the screening process questions, no other questions in Part II of the interview-questionnaire were asked. The following figure points out the entry conditions, the respective questions on the interview-questionnaire, and the sequence in which these key questions were asked: NONCONFLICT GROUP Entry Respective Question on Condition Interview-Questionnaire Sequence II-l II-4 II-5 II-6 II-2 II*12 mU'IDbLAJNl-J mUlobbJNH Figure 2. Screening Process for Nonconflict Group Members. 65 Whether or not a student would fulfill, or be dis- qualified because of, entry conditions 7 and 8 could not be ascertained during an interview. A student might per- haps have known at this time whether or not he would fill out a Value Survey (if he got one), but he could not know nor would the interviewer be able to predict whether or not the other student in the roommate pair would answer the questions "mutually" and then take time to properly fill out and return the Value Survey. Thus, whether or not a student or roommate pair would be ultimately ad- mitted into the final nonconflict group sample was deter- mined at a much later point during the course of the re- search project. It should be pointed out that although the above mentioned sequence of the screening process was usually the procedure used, often other question sequences occurred. A student often answered some questions before they were actually asked, and often also the interviewer would frequently encounter a particularly timid or an especially gregarious person. In such cases the author, using his discretion, altered the sequence of questions trying to fit yet unanswered questions into the natural flow of the discussion. Much additional and interesting information was obtained from the potential nonconflict group members via the non-screening process questions. However, since this research project does not further attempt to analyze such 66 data, Part II questions 3, 7, 8, 9, 10, and 11 and their subsequent answers will not be explained and discussed. Non-screening process questions were asked only if and after all screening process questions had been answered favorably, that is, in a manner which did not eliminate the potential nonconflict member from the nonconflict group. In a similar manner, a screening process existed for potential conflict group members. Part III of the interview-questionnaire contains the questions asked to these students who were potential members of the conflict group. Again, as soon as a person failed to meet any of the entry conditions in any of the screening process questions, no further questions in Part III of the inter- view-questionnaire were asked to the student. The follow- ing figure points out the entry conditions, the respective questions on the interview-questionnaire, and the sequence in which these key questions were asked: CONFLICT GROUP Entry Respective Question on Condition Interview-Questionnaire Sequence III-1 III-4 III-5 III-6 III-2 III-12 ChU'lthNH ONU'IAUJNH Figure 3. Screening Process for Conflict Group Members. 67 Again, whether or not a student or a roommate pair were ultimately to be admitted into the conflict group de- pended upon whether or not entry conditions 7 and 8 would eventually be satisfied and this could not be determined until a much later date. Again, also, the sequence was often altered depending upon the situation and the unique circumstances involved with each interview. Once more, interesting and additional information was gathered in the form of non-screening process questions (3, 7, 8, 9, 10, and 11 of Part III), but data received from such questions will not be reported in this study. Thus, within the screening process for each of the two groups (nonconflict and conflict), students were usually eliminated at each step (successive question) in the process. Accordingly, some students may have been asked all the questions stated on the interview-question- naire, but most of them were only asked a few questions.23 Time-wise, the length of most interviews varied from be- tween three to twelve minutes. 23Many students answered questions in both Part II and Part III for since most of them were upperclassmen and had had many roommates, it was not uncommon for a student to have had both a compatible and an incompatible roommate. Thus, many students were both "potential" conflict and non- conflict group members. However, only two students out of the 1,200 interviewed met all eight entry conditions of both groups. The researcher decided to place these two students and their respective pair-mates into the conflict group since conflict pairs were more scarce than non- conflict pairs. 68 As mentioned before, approximately 1,200 under- graduate male students were interviewed. However, not all of the information was gathered via personal (person-to- person, face-to-face) interviews, although it is estimated that over 90 per cent of the information was so obtained. The rest was obtained via telephone conversations. Tele- phone conversations followed the same format as mentioned above utilizing an introduction, the gathering of identifi- cational data, a screening process and closing comments (Parts I-IV of the interview-questionnaire). The telephone alternative was used in the latter stages of the infor- mation gathering when basically the mutuality as well as other entry conditions had to be checked out. Each person telephoned, unknown to him, had already been identified by a former roommate as a potential conflict or nonconflict (invariably the former) half of a roommate pair. Students telephoned were told that a random sample of students was being telephoned in order to obtain data for a study of college roommates. The telephoning alternative was used because those students telephoned were scattered all over campus and did not live in one or the other of the three main dormitories studied. It was impractical to attempt to interview them privately in their dormitory rooms since this would necessitate a special visit to their respective dorms and rooms, and more than likely they would not be in their room when the researcher attempted to interview 69 them privately.24 The telephoning alternative proved to be successful and was a pleasant change of technique for the interviewer since the data could be gathered from an office rather than by roaming through the dorms. The only problem encountered was the requirement that each individual be interviewed alone. The researcher would explain the neces- sity of this requirement to the student over the phone. The problem was solved by either calling the person back later (often recalling him several times because other stu- dents were always in the room when the researcher called), or more commonly, by calling the student at a different phone number--usually the phone number of his suitemates. Thus, using the above procedures and methodology, this researcher interviewed 1,194 students and ended up with 62 pairs of conflict roommates and 62 pairs of non- conflict roommates. The nonconflict roommates were much easier to find than the conflict roommates, and, accord- ingly, the nonconflict group of 62 pairs was completed well in advance of finding 62 pairs of conflict roommates. After finding the 62 pairs of nonconflict roommates, Part II of the interview-questionnaire was eliminated from the discussion when interviewing students. It was pointed out 24The author was going to call them in order to make an appointment to interview them personally, but a few attempts to do this revealed that often schedules were incompatible and/or that the duration of time and effort required before actually receiving any data was not worth the bother. Besides, the telephone interviewing technique (after it began) seemed to be working out very well. 70 to them that the researcher was analyzing also a noncon- flict group in addition to a conflict group, but that he already had enough members in this group so that it would not be necessary to ask them questions in this area. The number of sixty-two pairs of roommates in each group was not set arbitrarily. The researcher planned on a good rate of return since rapport seemed to have been established during the interviews. The author estimated and expected an 80 per cent return, 80 per cent of sixty- two pairs in each group would give him the necessary fifty pairs in each group. As soon as sixty-two pairs in each group were obtained, the researcher did not do any more interviewing. These sixty-two pairs in each group satis- fied all entry conditions up to and including entry con- dition number 7 (mutuality). The next problem was to get these pairs to fulfill entry requirement number 8, that is, the proper filling out and returning of the Value Surveys. Thus, at this point there were sixty-two pairs of potential conflict group members and sixty-two pairs of potential nonconflict group members. Each of these 248 students was given Form B of Rokeach's Value Survey (see Appendix B) plus a slip of paper giving the following introduction and instructions: You have been randomly selected to be included in the study of roommates which I am conducting. In our initial meeting, I informed you that I might be getting in contact with you again. Would you please fill out the following Value Survey as soon as possi- ble and return it in the envelope provided to the main lobby of your dormitory where the mail is distributed. 71 The person in charge of mail distribution will then return the Survey to my office via campus mail (no stamp is necessary). If you have not returned the Value Survey within a week, I will call you and re- mind you that your prompt filling-out of the Survey will be greatly appreciated. Thank you for your cooperation. Sincerely, Andrew F. Sikula After ten days 62 per cent of the Value Surveys had been returned. Those students who had received a Value Survey but who had not yet returned them were telephoned as a follow-up, reminded of the Survey, and pleasantly asked to return the Survey as soon as possible. Some students were given another Value Survey because they claimed that they either had not received or had misplaced the Survey.25 Within seven days, another 19 per cent of the Surveys had been returned. However, only about thirty-five pairs of complete data for each group was accumulated (along with thirty half pairs, that is, only one person of a potential pair to data had returned the Value Survey). It was ap- parent that an 80 per cent return did not assure that information for fifty complete pairs in each group would be attained. This was because for every individual who 25The researcher personally delivered all Value Surveys by going to the rooms of the students and sliding the Surveys under their doors. If an additional Survey was delivered, the researcher knocked on each door and attempted to place the Survey directly into the hands of the poten- tial group member. If no one was home, the Survey was again slid under the door. 72 failed to return a Survey, the possibility of one pair was destroyed. At this point the researcher was confronted with another decision. Either additional interviews could be undertaken in an attempt to add to the sixty-two pairs in each group who satisfied the first seven entry conditions of their respective groups, or additional follow-ups attempting to get more of the members of the original sixty-two pairs in each group to return their Value Surveys could be undertaken. The latter course of action was decided upon basically because it involved only telephoning rather than doing additional personal interviewing. Second, third, and occasionally even fourth follow- up phone calls were made usually spaced about three days apart. Often, again, additional Value Surveys were re- distributed. Eventually 93 per cent of the Value Surveys were required and returned before complete data from fifty pairs of roommates in each group were obtained. As soon as the fifty-pair goal in each group was obtained, all follow- up procedures ended. The final tally showed that 230 out of 248 students (93%) had eventually, often due to much persistence and persuasion from the researcher, returned the Value Surveys.26 After all final Value Surveys had 26Of the eighteen students who did not return Value Surveys, all but two of them, according to their telephone conversations, were eventually planning to return them. One student said that he did not want to fill it out, and the other student who did not return his Survey hung up on the researcher during a third follow-up phone call. 73 drifted in, actual data on fifty-three pairs of nonconflict roommates were available, but one of these pairs had in- complete data and the last two pairs to be completed were not considered as part of the final nonconflict group since only data on the first fifty pairs to have complete data were to be analyzed.27 Actually, also, fifty-two pairs of conflict roommates returned their Surveys, but two of the individuals did not fill out the Survey properly and thus complete data existed for only fifty pairs. Thus, 105 pairs of roommates returned their Value Surveys, these 210 people and 20 half-pairs comprised the 230 individuals out of the 248 students who returned their Value Surveys. At last, complete data from fifty pairs of conflict roommates and fifty pairs of nonconflict roommates were obtained. These fifty pairs of roommates in each group satisfied all eight entry conditions required by their respective groups. Rokeach Value Survey Two basic data gathering instruments were used in this study. The interview-questionnaire which has pre- viously been described and explained at length, and the Rokeach Value Survey. The Value Survey (see Appendix B) is made up of a set of eighteen terminal (ends) values and a 27Incomplete data on the Value Surveys consisted of ranking the eighteen values in each scale as high, medium, or low rather than on a scale from 1 to 18 (one case), or consisted of refusing to fill out the Survey be- cause it involved too much mental effort (two cases). 74 set of eighteen instrumental (means) values. The student is asked to rank these values in each set from 1 to 18 in order of their importance to the student. It may surprise the reader to learn that this author was unaware of the existence of the Rokeach Value Survey until well after many of the author's current ideas, convictions, and research plans were formulated. In fact, remembering the discussion concerned with the evolution of the present research study, the proposals about the hier- archical organization, labor versus management, and the purchasing agents were all presented with plans to use the Allport-Vernon-Lindzey Study of Values device. After talk- ing with various colleagues, professors, and Dr. Rokeach himself, the author became convinced that for the purposes of this study, the Rokeach instrument would be superior to the AVL technique. Thus, it was a coincidence that a type of measuring device that this researcher was looking for was, in fact, available and, indeed, had originated but a few blocks from the researcher's office. Even more to this author's amazement was the fact that he found out that many of the ideas that he had had about values and behavior were far from being new ideas, and indeed, Rokeach had already written many documents and conducted various research projects which examined many of the ideas and theories possessed by this author.28 28See Milton Rokeach, Beliefs, Attitudes and Values (San Francisco: Jossey-Bass, 1968); and The Open and Closed Mind (New York: Basic Books, 1960). 75 Actually, if one wished to measure social values up until less than five years ago, about the only instru- ment available was the AVL Study of Values. Many acknowl- edged experts, however, are skeptical about the reliability and validity of the AVL device.29 Rokeach and his assistants have statistical data which tend to support the use of the Value Survey.30 A review of the strengths and weaknesses of both the AVL and the Rokeach devices convinced the author to use the latter measuring device for the purposes of this study. It is beyond the scope of this report to go into detail to rationalize or support the use of one technique versus another. However, since the Rokeach instrument is so new and relatively unknown to most behavioral science researchers, the author strongly encourages all readers to consult the two Rokeach citations mentioned earlier. The two articles describe the Rokeach approach to the measurement of values and value systems. Statistical 29Oscar Krisen Buros, Mental Measurements Year- book (6th ed.; Highland Park, N.J.: Gryphon Press, 1965), pp. 384—.850 30For examples, see R. P. Beech, "Value Systems, Attitudes, and Interpersonal Attraction" (unpublished Ph.D. dissertation, Michigan State University, 1966); C. C. Hollen, "The Stability of Values and Value Systems" (un- published Masters thesis, Michigan State University, 1967); and R. Homant, "The Meaning and Ranking of Values" (un- published Masters thesis, Michigan State University, 1967). 76 properties of his value scales along with a few illus- trative substantive findings in the area of political behavior are also contained within the manuscripts. Additional statistical support of the Value Survey is contained within the references noted in footnote number 30. I The author's original intent was to use Form D of the Value Survey which is a gummed label version and which has a greater reliability than Form E. However, Form B (see Appendix B) was used because Form D Surveys were not available when needed since Rokeach had recently switched contracts from one printing company to another and some legal red tape was slowing the printing down and had, in fact, brought the production of Form D Surveys to a halt. A backlog of orders existed so that no Form D Surveys were expected to be available for several months. Pilot Study and Subsequent Changes in the Initially Planned Procedures A pilot study initially occurred in order to esti- mate the feasibility of the current study. Ten pairs of conflict and ten pairs of nonconflict roommates were sought using the basic procedure and methodology that was ex- plained earlier. The pilot study revealed that only about one out of every five persons interviewed would meet all of 77 the entry requirements (excluding number 8) of one group or the other. This ratio proved to be correct in the long run, too, as 248 individuals (124 pairs) met all of the first seven entry conditions and thus potentially quali- fied for one group or the other out of the total of some 1,200 students who were interviewed. Thus, the pilot study indicated the size of the population which would be neces- sary in order to acquire the needed data. The interview-questionnaire form also changed as a result of the pilot study. A few questions were dropped and others were added. Most of these changes were too trivial to mention here. The main change was the addition of questions II-12 and III-12 which were described earlier.31 Also the sequence of the questions changed as a result of the pilot study. The screening process and the ultimate sequence in which questions were asked (both described earlier) resulted because of the pilot study. The pilot study also revealed that nonconflict pairs were easier to find than conflict pairs. Accord- ingly, minor adjustments in the wording of certain sen- tences, and the manner in which the researcher asked some 31This involves entry condition number 6. Original pilot study pairs had to be recontacted because questions II-12 and III-12 did not exist prior to the pilot study. The recontacting procedure was simple enough, individuals merely being told that the researcher wanted to clarify a point or two from the previous interview. The recontact- ing interview only took a few seconds and only had to be done for forty students (ten pairs of conflict and ten pairs of nonconflict roommates). 78 of the questions changed somewhat so that more potential conflict pairs could be identified. The "mini" study also revealed, however, that both groups could be found easily enough if the necessary amount of effort was exerted. Originally, off-campus students were going to be contacted, but since enough students meeting all the necessary group entry requirements could be found on campus, it was de- cided that only on campus students would be studied. As explained previously, economic considerations (the cost of postage for returning Value Surveys and the traveling ex- pense to reach off campus locations) also influenced this decision. Originally, all students were to be given a Value Survey when interviewed. This was not done, however, be- cause Form D was going to be used and it is relatively costly. Therefore, it was decided that only after both members of a pair met all the entry requirements (again excluding number 8) would they then each receive a Survey. Thus, only 248 Surveys instead of 1,200 were needed. How- ever, because of the unavailability of Form D Surveys when they were needed (as explained above), it became necessary to use Form B instead of Form D and the cost factor virtu- ally eliminated itself since Form E is easily reproduced. Another change that was not originally planned on was the use of the telephone interview described earlier and which was explained as being necessary due to the physical distances between dormitories and the scarcity of 79 the researcher's time. The necessity of making second, third, and occasionally fourth telephone follow-ups, similarly, was initially not anticipated. In conclusion, the pilot study showed that the research project was feasible. The "mini" study accounted for a few minor changes in methodology and procedure, most of which involved the wording, number, and sequence of questions asked during the interviews. Other changes which developed later during the study concerned the use of a different version of the Value Survey and the use of the telephone in conducting interview and follow-up pro- cedures. CHAPTER III ANALYSIS OF THE DATA Introduction Whereas the last chapter concerned the general procedures and methodology used in gathering data, this chapter will concern the specific statistical techniques which were used in analyzing the Rokeach Value Surveys. Also in contrast to Chapter II which assumed initially that the reader had no understanding of the general method- ology used by the author in this study, this chapter will be presented in a manner which assumes that the reader has a basic understanding of elementary statistical techniques. The purpose of this chapter is not to explain in great detail the exact operations and/or the statistical back- ground and properites of every measuring technique used in this study. Readers interested in such matters should consult the footnoted material of the previous chapter in addition to exploring the references cited in this chapter. The purpose of this chapter is to briefly explain what specific statistical tools were used to analyze the data. The questions as to why certain statistical techniques were 80 81 chosen and how such statistical tools operate are somewhat beyond the sc0pe of the main context of this report and, accordingly, are the subject matter of Appendix D (Sta- tistical Supplements) and of the cited references. In general, the why_question is answered in Appendix 0-1 which is entitled "Choice of Statistical Techniques"; the hgw question is the province of the footnoted references. The question as to what statistical measures were actually used is the subject matter of the three sections of this chapter. The design of the research and the analysis of the data is divided into three parts corresponding to and in accordance with the three sample sets, namely, the entire sample, the original sample, and the cross-validation sample. Existing computerized techniques in the form of "routine oneway" and/or "routine stable" (both to be ex- plained shortly) were performed on and applied to the three samples. The first section of this chapter entitled "Sta- tistical Techniques of 'Routine Oneway'" discusses sta- tistical procedures used to compare and contrast 12917 viduals in the conflict group versus individuals in the nonconflict group. The "Statistical Techniques of 'Routine Stable'" are presented in the second portion of this chap- ter. The same student roommates are again analyzed, but this time the emphasis or focus has shifted from indi- viduals to paigg of roommates in the conflict group versus pairs of roommates in the nonconflict group. 82 The final section of this chapter reports how in- formation from routines oneway and stable was used to "pre- dict" whether or not a non-classified roommate pair from the cross-validation sample belonged to either the conflict or the nonconflict group. The forecasting procedures which were used will be explained in the "group prediction of un- classified pairs" section of this chapter. Statistical Technigues of "Routine Oneway Two of the computer routines mentioned in Appen- dix D-1 and described in detail in Appendix C (Program Description and User's Manual for: VALUTEST) were used by this researcher.l In this section of this chapter we will discuss routine oneway and the next section of this chapter will concern routine stable. Routine oneway is designed to provide comparisons of the median value rankings of several groups, as defined by an independent variable. It is analogous to a one-way analysis of variance. It will be used in this study to 1Actually a third routine called routine "check- sums" was also used. Routine checksums offers a quick and easy method of double-checking one's data for errors before running it on the other routines. This routine merely adds together all the rankings in each set of values. If a per- son has properly ranked the values, 1 through 18, the total must be 171 for each scale of values. If this total is not obtained, an error has been detected in the data. This routine provides a simple method of double-checking the data for respondent, coding, and keypunching errors by adding together the sum of each respondent's value rank- ings and printing out his data for error correction if this sum is not correct. 83 compare individuals in the conflict group with individuals in the nonconflict group, the independent variable being the presence or absence of conflict. This routine is de- signed to compute the median rankings of each of k objects (two sets of eighteen values each) by two or more groups of individuals (conflict and nonconflict groups), and to test the significance of the differences between the groups for each object ranked. The results of routine oneway appear in two sections. First, for each category variable (conflict and nonconflict) and for each dependent variable (that is, each of the thirty-six values) there is computed and printed an overall frequency table. Medians, quartile deviations, ranks, and concordance coefficients are also reported for the total sample population.2 Second, a detailed breakdown of the frequencies is printed showing how many individuals in each subgroup (con- flict and nonconflict) ranked each value first, second, third, etc. Medians, quartile deviations, ranks, and con- cordance coefficients are again reported but now for each subgroup, namely the conflict and the nonconflict sub- groups. In its unaltered form as reported in Appendix C, routine oneway also provides either a value for chi-square or a value for a Kruskal-Wallis H statistic both (either) 2Specific illustrations of these statistics are given in the next chapter. 84 with accompanying probabilities of occurrence under a two- tailed test, given a certain "degree of freedom." However, neither of these statistical techniques will be described further since the author did not utilize information from either of these procedures. A Wilcoxon statistical test of significance was instead used and its nature and proper- ties will be described shortly.3 Although this study did not use the chi-square test, it did make use of the chi- square distribution when examining the significance of the Kendall coefficient of concordance.4 One-tailed and two-tailed tests differ in the location but not in the size of the region of rejection of any null hypothesis. That is, in a one-tailed test the region of rejection is entirely at one end of the sampling 3This choice was made after consultation with Dr. James Stapleton, Chairman of the Statistics Department at Michigan State University. The Kruskal-Wallis technique is preferable to the chi-square test but the Kruskal-Wallis technique, when properly applied, is designed for use when analyzing many samples or subgroups. This research in- volves only two subgroups, the conflict group and the non- conflict group, and thus the Wilcoxon technique is the proper test of significance. The Wilcoxon and Kruskal- Wallis procedures are statistical analogues. The Wilcoxon procedure utilizes a z score and a normal distribution; the Kruskal-Wallis procedure utilizes an H statistic (one-way analysis of variance) and a chi-square distribution. The 2 value of the Wilcoxon technique when squared equals the H statistic of the Kruskal-Wallis technique. 4The author assumes that the reader is familiar with the distinction between a chi-square test and a chi- square distribution. Confused readers should consult Sidney Siegel, Nonparametric Statistics for the Behavioral Sciences (New York: McGraw-Hill Book Company, 1956), p. 43. 85 distribution. In a two-tailed test, the region of re- jection is located at both ends of the sampling distri- bution. The size of the region of rejection is expressed by the assumed and stated level of significance. A two- tailed test is used when the hypothesis being tested does not indicate the direction of the predicted difference. If a .05 level of significance is chosen, a one-tailed test has a region of rejection of 5 per cent all at one end (tail) under the curve in the sampling distribution. A two-tailed test has a region of rejection of 2.5 per cent at each end under the curve in a sampling distribution. The hypotheses tested in this study did not indicate the direction of the predicted difference and thus only two- tailed test procedures were used. There are a number of different sampling distri- butions, one for each value of df (df refers to degrees of freedom). The size of df reflects the number of obser- vations that are free to vary after certain restrictions have been placed on the data. For example, if the data for 100 cases (individuals) are classified in two cate- gories (conflict and nonconflict), then as soon as we know that, say, fifty cases fall in one category, we also know that fifty must fall in the other. For this example, df=l, because with two categories and any fixed value of n, as soon as the number of cases in one category is ascertained, then the number of cases in the other category is determined. 86 The Wilcoxon W test may be used to test whether two independent groups have been drawn from the same population. This is one of the most "powerful" of the nonparametric tests, and it is a most useful alternative to the para- metric t-test when the researcher wishes to avoid the t- test's assumptions, or when the measurement in the research is weaker than interval scaling (as is the case in this study).5 In this study the Wilcoxon test is used to test: frequency distributions of individuals in the conflict and nonconflict groups as analyzed via routine oneway; and, the amount of differences between and among roommate pgigp of the conflict and nonconflict groups as analyzed via routine stable. The exact statistical procedures performed during the Wilcoxon test are difficult to understand with- out an accompanying example. Thus, a more detailed expla- nation of this significance test is postponed until the next chapter is presented.6 Routine oneway also makes use of a (Kendall) coef- ficient of concordance. This statistic is a measure of the 5The "power" of a test is defined as the probability of rejecting the null hypothesis when it is in fact false (and thus should be rejected). 6The Wilcoxon test is described in full detail in F. Wilcoxon, "Individual Comparisons by Ranking Methods," Biometrics Bulletin, I (1945), 80-83. See also the dis- cussion of the Mann-Whitney U Test in Siegel, Nonpagametric Statistics for the Behavioral Sciences, pp. 116-27. The Mann-Whitney U statistic is easily converted into the Wilcoxon W as will be explained fully in the next chapter. 87 degree of agreement among all of the members of the group on the entire set of dependent variables. In terms of Rokeach's Value Survey, it provides an index of homogeneity of the group's values. Thus, it is a measure of the extent to which conflict and nonconflict group members agree among themselves on their ranking of all eighteen terminal and all eighteen instrumental values. Methodologically, the concordance coefficient is an index of the degree of corre- lation between the "value profiles" of all group members. It is mathematically equivalent to the average of the Spearman rank-order (rho) correlations between all possible pairs of group members. In general, the more homogeneous are the groups, that is, the larger is the concordance coefficient, the more reliable are the data and conse- quently the more reliable are the conclusions and impli- cations drawn from the analysis of the data. A high or significant value of the concordance coefficient may be interpreted as meaning that the observers or judges are all applying essentially the same standard in ranking the objects (values) under study. The inquisitive reader should once again consult the references cited below.7 7M. Friedman, "A Comparison of Alternative Tests of Significance for the Problem of n Rankings," Annals of Mathematical Statistics, XI (1940), 86-92; M. G. Kendall, Rank Correlation Methods (London: Griffin Publishing, 1948); and B. Willerman, "The Adaptation and Use of Ken- dall's Coefficient of Concordance (w) to Sociometric-type Ranking," Psychological Bulletin, LII (1955), 132-33. 88 Thus, routine oneway as modified for the purposes of this study makes use of the following statistical tech- niques: sample and subsample frequency distributions, medians, quartile deviations, ranks, and concordance coef- ficients; and the Wilcoxon W test of significance of the frequency distributions between conflict group and non- conflict group individuals for the individual values in each value scale (terminal and instrumental). All of these techniques were briefly explained above and confused and/or interested readers are urged to consult the references cited in this chapter.8 Specific data illustrating the above statistical techniques will be given in the next chapter. Perhaps there is some disadvantage in presenting theoretical statistical procedures without accompanying data. Hopefully, this loss is offset or even overcome by the advantage of separating methodology from findings. It is cumbersome to present the "findings" of any research project in a manner which necessitates the explanation of statistical terms during the course of the discussion whose stated purpose is to explain only the "findings" of any 8Besides the aforementioned references, other help- ful resource books are: Merle W. Tate, Nonparametric and Shortcut Statistics in the Social, Biological and Medical Sciences (Danville, Ill.: Interstate Printers and Pub- 1ishers, 1957); Charles Kraft, A Nonparametric Introduction to Statistics (New York: Macmillan, 1968); Donald Fraser, Nonparametrig_Methods in Statistics (New York: John Wiley and Sons, 1957); John Edward Walsh, Nonparametric Sta- tistics (Princeton, N.J.: Van Nostrand, 1962); and Q. McNemar, Psychological Statistics (New York:. John Wiley and Sons, 1962). 89 research project. Hopefully the reader will have a general understanding of the statistical techniques used in this study by understanding the material in this chapter so that the subject matter of the chapter to follow can deal exclusively with research "findings." Before ending our discussion of routine oneway, the user "options" available to the researcher using this routine should be pointed out. The first option is a choice of either of two significance tests, the Kruskal- Wallis H test or the Median Test for k independent groups. As explained above, the author chose the Wilcoxon test to replace both of these alternatives. The author utilized an "echo check" option also which specifies that the data be printed out after being read in and recoded. This option is a safeguard technique to insure that the data is being read in and recoded correctly. A third "automatic recoding option" was not used by the researcher since the number of variables in this project is limited. The "en- tire sample option" was used so as to obtain frequency distributions and medians for the entire sample. The re- searcher did not use the "identification variables" Option since only one card per subject was needed. The final option, the "controls variable" option, similarly was not used because the design of this research project does not involve the analysis of many "subgroups" (only two groups, namely conflict versus nonconflict). Specific reader inquiries concerning the substance and use of these 90 options should be directed toward and can probably be answered by reference to Program Valutest which is Appendix C of this report. Statistical Techniques of "Routine Stable" Routine stable was originally designed for use with individual subjects who had been tested twice but with no intervening influence attempt, that is, for data in which the experimenter's primary interest was in the degree of value stability from Time 1 to Time 2. This researcher modified this routine so that it could be used with the data gathered in this research study. Instead of Time 1 and Time 2, the researcher substituted Individual 1 and Individual 2--both of whom were pair-mates in either the conflict or the nonconflict groups. This researcher's primary interest at this stage of the data analysis was in measuring the degree of value agreement between individual members of a pair. Routine stable provides results per- taining to both the degree of total "value system" profile correlations for each pair, and the degree of agreement on each "individual value" for each roommate pair. Thus, routine stable was used to compare and contrast the pairs of conflict group roommates with pairs of nonconflict group roommates. The only major technical change in the com- puterized routine involved reading each pair of data into the computer as if it were one subject (this operationally substituted Individual 1 and Individual 2 for Time 1 and 91 Time 2). Also, actually, routine stable had to be run twice for each sample tested; once for the pairs of con- flict roommates and once for the pairs of nonconflict roommates. For our purposes, the question of value agreement may be divided into two separate experimental questions: (1) What is the degree of "value system" agreement of each pair in each group, that is, who are the people (conflict or nonconflict group members) with the most agreement on "value systems" and who are those with the least agree- ment? and (2) What is the degree of pair agreement of each individual "value," that is, for which "individual values" is agreement greatest and for which ones is agreement least in this particular pOpulation of conflict and non- conflict group members? Routine stable provides results to answer both of these questions. First it computes and prints an index of value system agreement for each pair, for each of the rank- ing scales (terminal and instrumental). Each pair, accord- ingly, has a terminal and an instrumental value correlation (agreement) coefficient. These coefficients are the Spear- man rank-order (rho) correlations between Individual 1 and Individual 2's (pair-mates) rankings of the two sets of values. If pair-mates rank each scale in exactly the same manner, the correlation-agreement coefficient will be 1.00 (a perfect score). To the extent that Individual 1 and Individual 2 rankings are different, the agreement 92 coefficient will be reduced, and theoretically may go as low as -1.00. Of all the statistics based on ranks, the Spearman rank correlation coefficient was the earliest to be de- veloped and is probably the best known today.9 This sta- tistic, often called "rho," is a measure of association which requires that both variables be measured in at least an ordinal scale so that the objects (values) under study may be ranked in two ordered series. The basic rationale and methodology of the Spearman rank correlation coef- ficient is known to most researchers and will not be ex- plained in greater detail here. Individuals wishing to review the mathematical procedures involved with the use of this technique should refer to Siegel or any other appropriate statistical text.10 The value of rho obtained can be tested for signifi- cance. If the subjects whose scores are used in computing rho are randomly drawn from some population (which is assumed to be the case in this research study), we may use those scores to determine whether the two variables are associated in the population. That is, we may wish to test the null hypothesis that the two variables under study are not associated in the population and that the observed rho 9Siegel, Nonparametric Statistics for the Behavioral Sciences, p. 202. lOIbid., pp. 202-13. 93 differs from zero only by chance. Various techniques for testing the significance of rho are available and the choice of technique is usually dictated by the size of the sample. This study utilizes the t-test to determine the significance of rho. Concerned readers should again refer to the Siegel or Kendall references cited previously in this chapter.11 Thus, routine stable will initially compute and print each pair's "value system" agreement coefficients (Spearman rank-order correlation coefficient). Routine stable also provides frequency distributions, medians, means, and standard deviations of these sets of agreement scores. Such devices should be familiar to the reader so no additional time or space will be devoted to the expla- nation of these statistical techniques. Referring back and in regard to the second question --the pair agreement on each individual value--routine stable offers two different measures of agreement (or dis- agreement) for one individual in a pair as compared to the other individual in the same pair.12 First, a frequency 11In addition to the Siegel and Kendall references, other worthwhile discussions of the Spearman rank-order correlation can be found in: H. Hotelling and Margaret R. Pabst, "Rank Correlation and Tests of Significance Involv- ing No Assumption of Normality," Annals of Mathematical Sta- tistics, VII (1936), 29-43; and E. G. Olds, "The 5% Sig- nificance Levels for Sums of Squares of Rank Differences and a Correction," Annals of Mathematical Statistics, XX (1949), 117-18. 12The unaltered version of routine stable provides three such measures. The third measure which this author 94 distribution of the differences between Individual 1 and Individual 2 (pair-mates) is given. This frequency distri- bution information is reported for both conflict and non- conflict group pairs and for both terminal and instrumental individual values. In addition, for each individual value the median amount of difference will be given. Second, routine stable provides a display of two different measures of agreement for each of the individual values.13 These are median difference (mentioned above) and a Pearson product-moment correlation between Individual 1 and Indi- vidual 2's (pair-mates) scores for each individual value (again reported for both conflict and nonconflict group pairs and for both terminal and instrumental value scales). Each difference measure will be ranked from least to most difference. The Pearson product-moment correlation technique should be familiar to the reader. This sum of the squared differences technique is similar to the Spearman rank corre- lation coefficient discussed earlier and an explanation of has not reported in this study is the results of a Median Test which was eliminated because when this author adapted routine stable so that it could be used with his data, the significance and interpretation of the Median Test infor- mation became ambiguous and to some extent even meaning- less. This was due to the fact that Individual 1 and Indi- vidual 2 were substituted for Time 1 and Time 2 as ex- plained previously. 13Again, the unaltered version of routine stable contained an additional statistic namely the "mean change" figure which again was inapplicable to this author's data because of the substitution of individuals for times. 95 mathematical manipulations involved can be found in most every modern statistical textbook on the market today. At this point, all the statistical techniques of routine stable have at least been mentioned although, per- haps, not thoroughly explained to or understood by the reader. This author wishes to comment on only one addi- tional topic as an attempt to clarify a matter that might be confusing to the reader. This research project utilizes both parametric and nonparametric statistics. The Spear- man rank-order correlation coefficient and many other techniques described previously are nonparametric sta- tistics and are used with data in ordinal (rank order) form. Nonparametric statistics cannot make use of arith- metic operations (adding, multiplying, etc.) but instead utilize medians, frequency distributions, and other devices. If the data are in the form of an interval scale, para- metric statistics can usually be used (see the "assumptions" or "conditions" involving the use of parametric statistics presented in the "choice of statistical techniques" appen- dix of this report). Accordingly, after Spearman coef- ficients have been determined from ordinal ranked value data, a parametric test such as a t-test can be applied to the Spearman coefficients because such coefficients now represent interval measurements. Similarly, a parametric Pearson product-moment correlation test may be applied to differences in individual rankings because although the 96 rankings themselves are ordinal measurements, the "differ- ences" between rankings are interval measurements. The methodology involved with computing the Spear- man coefficient is analogous to the Pearson coefficient-- both are sum of the squared differences techniques. How- ever, the Spearman technique is used with ordinal data (nonparametric statistics) and the Pearson technique is used with interval data (parametric statistics). The re- spective tests of significance for each of these corre- lation coefficients usually involve the use of chi-square with the Spearman coefficient and the use of a t-test for the Pearson coefficient.l4 Hopefully, the above expla- nation points out why the use of parametric statistical techniques may be justified (however, usually at a later stage) in analyzing data received from an ordinal (ranking) measuring instrument (such as the Rokeach Value Survey). If this matter or any other statistical technique mentioned in this chapter is still confusing to the reader, the reader is urged to consult the references cited in this chapter before going on to the "findings" chapter which is soon to follow. 14In this study, however, a t-test was applied also to the Spearman coefficients. The manner and rationale of this application will be explained in the next chapter. 15Along with an understanding of the aforementioned statistical procedures, the author also assumes that the reader is familiar with the use of chi-square distributions, normal distributions, t-tests, z scores, and other standard statistical techniques. 97 Some mention again should be made of the user options available to any researcher who uses this routine. Most of these options are exactly the same, although there are less of them, as the options associated with routine oneway. Again, an "echo check" option is available. In this case this researcher did not have his data printed after it was read in and recoded because this was done previously when the echo check option was utilized during routine oneway. The "test Option" permits the user to specify whether he wishes to obtain results pertaining to just the subjects' (pairs') stability (agreement) on each value system, just the stability (agreement) on each indi- vidual value, or both. This researcher requested results of "both." Both the "identification variables" Option and the "control variables" option in routine stable are identi- cal to the same options in routine oneway. Again, because of the nature of this research project and the character- istics of the data gathered, the author did not attempt to utilize these options. A more detailed explanation of these user options is contained within Program Valutest which is Appendix C of this report. Group Prediction of Unclassified Pairs To review, routine oneway has analyzed data com- paring and contrasting individuals in the conflict group versus individuals in the nonconflict group. Routine stable has compared and contrasted information about "value systems" and individual "values" for pairs of roommates 98 in the conflict group versus paapa of roommates in the non- conflict group. Now, using the information obtained from routines oneway and stable, the next goal is to predict into which group (conflict or nonconflict) each and every pair from an unclassified sample of roommate pairs belongs. Recalling the design of this research project, three samples exist: the entire sample, the original sample, and the cross-validation sample. Routine oneway was applied to the entire sample and to the original sample. Routine one- way was not applied to the cross-validation sample because of financial and time constraints and the predicted rele- vance (irrelevance) of such information. Routine stable was applied to all of the three samples. Actually, this in- volved two routine stables, one for the conflict group and one for the nonconflict group, for each of the three samples. Chronologically, the sequence of computer programs was as follows: routine oneway applied to original sample, rou- tine stable(s) applied to the original sample, routine stable(s) applied to the cross-validation sample, routine stable(s) applied to the entire sample, and routine oneway applied to the entire sample. As initially conceived, routines stable and oneway were to be applied only to the original sample and the results obtained were to be formulated into a "prediction process" which would enable the author to classify un- analyzed and unidentified (by group) pairs of the cross- validation sample. However, neither a "systematic" nor a IMF”. a L I151 Au." :2 d-‘w- 411‘! 99 "clinical" prediction process proved to be available and/or reliable and as an attempt to explain the reasons for this, routine stable(s) was (were) later applied to the cross- validation sample. Both routines oneway and stable(s) were applied finally to the entire sample so as to strengthen the general findings and conclusions of this research project. The next chapter reports the actual "findings" from these various routines as applied to the three samples. No "systematic" prediction process was revealed after routines oneway and stable(s) were applied to the 16 O 0 O I However, a "c11n1cal" pred1ct1on process original sample. was formulated and applied by the author in the following manner. Roommate pairs in the cross-validation sample were identified by an impartial individual in such a manner that the researcher did not and could not know to what group any pair belonged. Not only were special identifying numbers given to the individuals and the pairs, but also the names of the students on the Value Surveys were eliminated be- cause in a limited number of special cases, the author may have been able to remember whether or not an individual identified by name belonged to one group or the other. Only after the researcher had made his predictions were the identifying numbers assigned by the impartial indi- vidual revealed to the author. 16By "systematic" is meant statistically valid and statistically reliable. 100 An unclassified pair was studied in two basic ways in order to make a prediction as to the group in which it belonged. Each "individual" in the pair was studied, that is, the value rankings of these two individuals were com- pared with the results from routine oneway. In addition, the two individuals taken together as a "pair" were studied via the use of information from routine stable. Thus, by comparing the unclassified pair individually and collec- tively with respective data from routine oneway and routine stable, predictions of group membership were made. This forecasting may appear to the reader to be easier than it actually was.17 For one thing, it is known that the pair has to belong to either one ”extreme" group or to the other. It conceivably and probably might be more difficult to obtain Value Surveys from random pairs of room- mates who we did not know belonged to one group or another and then to try to predict whether or not this pair might belong to one group or to the other. However, if the sta- tistics presented previously in this report prove true, four out of five such pairs would not belong to one "ex- treme" or the other, but would comprise the vast area be- tween these two extremes. Such an endeavor is interesting and illustrates the direction future research of this author or others might take, but such an endeavor currently 17The "ease" of the forecasting is one aspect of the problem; the "accuracy" of the forecasting, however, is quite a different matter as will be explained in greater detail later. 101 is beyond the scope and design of the present study. The prediction process might also appear to be easy because, not only can individuals and pairs be put into a group be- cause their Value Surveys are congruent with members of that group, but, also, these individuals and pairs may be assigned to a group because their Value Surveys are apt congruent with members of the other group. In addition, later predictions about group membership conceivably may be easier to make because the author knows that the final tally must be twenty-five pairs of roommates in each group (the prediction process being applied to the cross- validation sample). Whether or not the prediction pro- cess proved to be easy and/or successful will be reported in the next chapter. Before attempting to "clinically" predict the un- classified pairs of the cross-validation sample, the author predicted the analyzed pairs from the original sample. That is, after obtaining original sample results from routines oneway and stable the author used this infor- mation to predict group membership of pairs from which the initial data was gathered. These fifty "old" pairs from the original sample were mixed up and assigned an identifi— cation number again by an impartial outsider. After pre- dictions were made, the author observed the true group membership of the pairs. The author then compared his predicted group membership to the actual group membership, and the entire prediction process described above was 102 modified until the author, hopefully, found a reliable "clinical" technique of predicting the group membership of each pair. After an alleged reliable formula for predict- ing group membership of pairs had been discovered via experimenting on the "old" analyzed pairs from the original sample, the formula and the prediction process was applied to the "new" unclassified pairs of the cross-validation sample. The results of the prediction process are ex- plained in the next chapter. CHAPTER IV FINDINGS Introduction The purpose of this chapter is to report the find- ings revealed by the analysis of the data. In general, the findings will be reported as they chronologically occurred (the only exception being the switch in presentation of the last two sections of this chapter). Accordingly, the sectional sequence of this chapter is as follows: "Origi- nal Sample--Routine Oneway," "Original Sample--Routine Stable," "Group Prediction of Unclassified Pairs," "Cross- Validation Sample--Routine Stable," "Entire Sample--Routine Oneway," and "Entire Sample--Routine Stable." Original Sample--Routine Oneway The original sample consists of twenty-five pairs of conflict roommates and twenty-five pairs of nonconflict roommates. Routine oneway, it will be remembered, is de- signed to analyze individuals (not pairs) in one group versus individuals in the other group. Before analyzing the groups separately, however, first routine oneway analyzes the total sample--inc1uding both conflict and 103 104 nonconflict individuals. Since our sample is fifty con- flict individuals (twenty-five pairs) and fifty noncon- flict individuals (twenty-five pairs), the total sample is 100 students. By analyzing the total sample first, later comparisons and contrasts of subgroups (conflict and non- conflict groups) can be more meaningful. The data on the total sample follow and are reported in terms of frequency distributions, medians, quartile deviations and ranks (see Table 1 and Table 2). In the data reported thus far, the most important thing to observe is the final total sample medians and ranks of both the terminal and instrumental values. Having looked at the total sample, let us now look at the same data but now reorganized into two subsamples or subgroups, namely the conflict and nonconflict groups (see Tables 3, 4, and 5). For our scrutiny, the most important statistics are again the group medians, and the group ranks of both the terminal and instrumental values. The group medians and group ranks indicate that there is little difference in the manner in which individuals in either the conflict or the nonconflict group rank the terminal and instrumental values. The medians and ranks of the subgroups are similar and correspond accordingly with medians and ranks of the total sample of 100 (compare Tables 2, 4, and 5). The conflict and nonconflict frequency distribution (of ranks) can be tested in order to determine whether the 1()5 momm.N ONmm.mH OOH mH 5H OH OH n O O m O N N v N N H O O O Hmuoe N msouu NwHusomm HmcoHumz HNvN.m mmmm.m OOH O H N N v H v O O m O O O NH n MH HH a Hmuoe N msouu m>oq madam: mnNO.m OOON.~ OOH O v v m m m v m m e m m m m m m O OH Hmuoe N msouu Ncosum: umccH HOmm.m Omm0.0 OOH O H m O m e v n v O m n O m m m O OH Hmuoa N macaw mmmchmmm hOHm.N vaH.O OOH O H O m m O O O N n v O OH O NH O m m Hmuoe N msouu Eoowmum OOOh.m OOOM.O OOH O H v n v O O O n m m 5 HH O b m m m HmuomIN msouo NuHusomm NHHEmm HHHm.v OOmn.O OOH H O m S OH O m m m O O b m O O m O H Hmuoa N Qsouu NuHHmsmm vOOm.m OH>O.MH OOH OH HH MH 5 O n O m O n v v m N H m H H Hmuoe N Qsouu spammm mo pHuoz 4 ONmH.O OOON.O OOH O m w m O O O O O O OH v m O m m m h Hmuom u msouu momma um pHuoz < HmOO.m OOOH.O OOH N O m m N O O m n n m m h m n O O m Hmuoe N msouo ucmESmHHmEoo Io<.Oo mmcom d ooom.s eHeo.OH OOH m NH v m e m O m A O a v m A v a m s Hpuoe n msouo I «OH; chUHoxm ca mmmo.v OOOH.HH OOH O O O OH R m m O O n N m e H v m m h Hmuoe N msouo OOHH mHnmuHOMEou d OH OH OH OH OH MH NH HH OH m w b O O O m N H mHnmaum> ucwpcmmmo .Q.O cmemE Hmuoe mmHocmsvmum .HOOH N 2O mmsHm> HmucmEdHDOCH can HmcHEumu MOO mcoHumH>mo wHHuumsv cam .mcmHUwE .wcoHuanuumHU xocmsqmuulI>m3mco mcHusouIImeEdm HmchHHOII.H mqmHmm OOOO.N OON0.0H OOH v OH OH O OH O OH NH v H N O .O O H H O O HMUOB u QDOHU mnemmmHm OH NH OH OH OH OH NH HH OH O O N O m v O N H OHQMHHM> ucmocmmwa .o.o CMHOOE Hmuoa mmHocmskum OmscHuCOUII.H mqmdfi 107 HOH0.0 OOOO.N OOH O O O O O O O O O N O O N O NH O N Hmuoe N msouo OmHHouuCOUIOHmm ONO0.0 OOO0.0 OOH O H O O H O O N O O O O N N NH OH NH Hmuoa N msouo mHnHmcommmm OOH0.0 ONNN.OH OOH O O OH NH O O O N H O O O H O O N H Hmuoa N msouu muHHom OOOO.N OOO0.0H OOH ON NH HH OH N O O O N O H H N N H H O Hmuoe N msouo ucvamno NOO0.0 NOO0.0 OOH N H O O N O N O O O HH O O OH N HH HH HauOB N msouu mcH>oq ONO0.0 HNO0.0H OOH O O O O N O O O N O O O NH O O O O Hmuoa N msouu HMOHMOH OON0.0 OOO0.0H OOH O HH O O O O O O O O O O O O N N O Hmuoa N msouw HmsuomHHmucH OOO0.0 OOOO.N OOH N O O N O O OH O O O O O O O OH O OH Hmuoa N moon ucwwcw mvcH NOO0.0 OOOO.HH OOH HH O O OH OH O N OH O O O N O O O O O Hmuoe N msouo m>HumchmEH HH0.0 ONNN.O OOH N H H H H H O O N O O N N HH O O ON Hmuoa N msouu ummcom OOO0.0 NOON.O OOH N N O O O O O N N O N O OH O H O O Hmuoa N msouv H5O Hmm NOH0.0 OOO0.0 OOH O N O O O O O N O O O NH H O O O O Hmuoe N macs GH>H mom OH NH OH OH OH OH, NH HH OH O O N O O O N H .Q.O CONGO: HmuOB HMWQMWMMW mmHocosvmum O OmscHuCOOII.H NHmde 108 TABLE 2.--Original sample--routine oneway--medians and ranks for terminal and instrumental values (N = 100). Terminal Values Instrumental Values Total Total Med. Rank Med. Rank (N=100) (N=100) A Comfortable Ambitious 10.00 10 Life 11.70 13 Broadminded 6.79 4 An Exciting Life 10.07 12 Capable 8.50 7 A Sense of Ac- complishment 8.10 8 Cheerful 10.33 12 A World at Peace 8.20 9 Clean 13.93 17 A World Of Courageous 10.50 14 Beauty 13.07 14 Forgiving 9.00 8 Equality 8.75 11 Helpful 9.79 9 Family Security 3-30 10 Honest 4-23 1 Freedom 6.14 3 _‘___' Imaginative 11.00 15 Happiness 6.06 2 Independent 7.90 6 Inner Harmony 7.25 5 Intellectual 10.00 11 Mature Love 5 . 33 1 Logical 10.36 13 National.Secur- ity 15.55 17 Loving 6.67 3 Pleasure 13.58 15 Obedient 16.05 18 Salvation 17.13 18 Polite 13.28 16 Self-Respect 7.36 6 Responsible 4.50 2 Social Recogni- Self-Controlled 7.83 5 tion 13.64 16 Concordance True Friendship 7.10 4 Coeffs. 0.19 Wisdom 8.00 7 Concordance Coeffs. 0.23 I109 ONNN.N oooo.N ON N N o N H N uoHHOcoucozumsouo OOO0.0 OOON.O OO O N O H O N OOHchouuasouo mmmchmmm NNNN.N NOHN.N NN o N N O H N uoHHNcoocozumsouo NNov.N OHNo.N ON 0 N N N N N uoHHOcooumsouo Eonmmum NNoN.N NOON.N ON H N uoHchoocoznmaouo NONN.O NOOH.O OO N O OOHHOCOONQOOHO NOHHOOOOINHHEOO OONH.O OOO0.0 OO O O O O OOHchoocozumsouO OOO0.0 NOOH.O OO O O N O OOHHOCOONQOOHO NuHHmsmm OOON.O OOOO.NH OO O O uOHHOcoocoznmsouo ONNN.O OOO0.0H OO O H uoHchouumsouo Nusmmm mo OHuoz a OOOH.O OOO0.0 OO N O uOHHOcoocozumsouo ONO0.0 OOOH.O OO O O OOHHOOOONOOOHO momma um OHHOS < ONOH.O OOO0.0 OO N N H O O N O O O uOHHOcoocoznmsouu OON0.0 OOO0.0 OO O O H O H H O O O OOHHmeounmsouO ucmEHOHHQEOO and mo mmcmm < OOON.O OOON.OH OO N O O N N N O N O O O N N O O N N H uoHHOcoocozumsouo OOOH.O OOO0.0 OO O O H O O O O H O N O N O O H N H O OOHHOOOONOOOHO mNHH OcHuHoxm ca NNNo.N NNNN.NH ON 0 N N NH N N N N N N H o N N .N N H N uoHHNcoocozuasouo NHNN.N ONNN.OH ON 0 N O N N N N N O N H N N H N o N O uoHHmcooumsouo . mNHH mHnmuuoNsoo m OH NH OH OH OH OH NH HH OH O O N O O O O N H memHum> ucmOcmmmo .o.o OOHOmz Hmuoa mmHocmsUmum .HOO N zO masoum HOHchoococ Ocm . uOHHucoo HON mcoHumH>mO mHHuumsw new .OCMHOmE .mcoHuanuumHO Oocmsvmuwllam3mco mcHusouIImHQEOm HmcHOHHOII.O mqm<9 110 OOON.O OOO0.0 OO O N H H N O O O O N H O O O O O N O uoHHmcoocozumsouo OON0.0 OON0.0 OO N O H O O O O N O O O H O O H O O H uoHHmcooumsouu Eoanz NOOH.O OOON.N OO O H O H N O O O N O O O O O O O O O uoHchoocozumoouo ONOO.N OOO0.0 OO O O O O N H N N N O O O O O O O O H uoHHmcooumaouu mwnmvcmHum mama ONOO.N OOON.OH OO O O O O O O O N O H H H O H H H O O uoHHmcoocozumnouu ONOH.O OOO0.0H OO O O O N O O N O O O O N H O N H N H uoHHmcooumsouu :oHuHcOoomm HmHoom OOON.N NOOH.N OO O O H N N O N N H O O O H O O O H O uoHHucoocozumaoHo NHO0.0 OOOO.N OO H H N H H O N O O N O O O O O O O O uoHchooumsouo uoommwqume OOH0.0 OOO0.0H ON ON H O O N O N N N H H O O O H N O O uoHHmcoocozumaouu OOO0.0 OOHN.NH OO ON N O H N O H O N H O O H O O N O O uuHHucooumsouu coHum>Hmm OOON.N OOHN.OH OO N O O N N O O O H O O N O O O O N O uoHHmcoucozumsouo NOOH.O OOOO.NH OO N O O H O H O N O H N H O O H H O O uUHHmcouumsouo unammmHm NHOO.N OOON.OH OO O OH O O O O O N O H N O N H H O O O poHchoocozumsouu OHNO.N OOON.OH OO OH N OH O O O H H O H O O O H O O O O uoHHucoonmsouw NUHusomm HmcoHumz OON0.0 OON0.0 OO O H H H O O N H O N O O O O H O N O uoHHmcoocozumsouu OOOH.O OOON.O OO O O H H H H N O O N N O O O O O O O uoHHmcooumsouo . w>oq wuzumz OOO0.0 OOON.O OO O H O N H N O N O N H O O O O O O O uoHHmcoocozumoouu ONN0.0 OOON.N OO O O H H N H O N H N O O N O N O O O uomHmcouumsouo NmOEum: uwccH OH NH OH OH OH OH NH HH OH O O N O O O O N H .O.O cmHomx Hmuoe anmHmm> meocmswwum ucwvcm mo Umchucounu.O mqmHmuom OOOH.O OOO0.0 OO N O O O O O O O N O O O O H N O O H uoHchoocoznmzouo NOO0.0 OOO0.0H OO O O O O O O N O O N O O H O O H O H uoHHmcoonmzouo msoommusou OOH0.0 OOO0.0H OO N O O O O O O O N O N O N O O O O O uUHHmcoocoznmaouu NONN.O OOO0.0H OO HH O O O O O O O H O N H O H H H H O uoNHmcooumsouw :mmHU NON0.0 OOO0.0H cm H m N O N N O O O N H H H O O o N N uUHHmGOOGOZflmfiouu ONOH.N OOOO.OH ON H N H N H O O H N N H N N N N N N O uoHHucouumsouu Hsmummno NNNN.N OOOO.O ON H O N H N N N O N O O N N O O O N N uoHHOcoucoznmsonu NNHN.N OOOH.O ON O O H H N N O N O H N N O O N N O N uoHHucoun aouu mHnmmmu NNNN.N OOON.N ON H O O H N H N N H O N O O N H N N O uoHchoocozuasouo ONNN.N OOON.N ON H H O N O H N N N H N N O O O O N O OUHHOcooumsouu OmncHsOmoum NNHN.N ONNN.N ON N N N N N O H N N N O O N N O N O H uoHHOcoucozumaouo NONN.O OONN.OH ON O N N H N N N O N N H H N O N N N N uoHHmcooumaouu msoHanem OH NH OH NH OH NH NH HH OH O O N N N O N N H . . mHQMHum> O o cmHomz Hmuoa ucwncmmwo .mmHucwsvmum CUSCHUCOUII.O mam<fi 1112 OOO0.0 ONO0.0 OOOH.O OOO0.0 ooom.N OOO0.0 HHOO.H OOO0.0 NOOH.O NOO0.0 omN0.0 ONO0.0 ONO0.0 oooo.O HNO0.0 OON0.0 ONH0.0 NHOH.O OOON.O ooom.N oooo.m ooom.O ooom.OH OOOO.NH ooom.OH OOON.OH ooom.N ooom.m NOOH.O ooom.HH ooom.0H ooom.O OOO0.0 ooom.O NOOH.HH OON0.0H om om om om om om om om om om om om om om om om om Vfl' 0‘)" Hm N40 Nffi Nrn «wH oao Nnn N\D Q'N MP4 NH HM HO HO NO NM HN NO V0 \DN Mv-I M\O [‘0 NM HN MN HM NH MN NeH ofu P-M wvo uoHchoocoznmsouw uoHHmcouumsouu OmHHouucouamem uoHchoucozumsouu uoHHmcouumnouo mHnHmcommmm uoHHmcoocozumsouo uoHHmcooumaouo muHHom uoHHmcoocozumsouu uoHHmcoonmaouu ucmHvono uoHchoocozumsouU uoHHmcounmsouo mcH>OH uOHHmcoocozumsouu uoHHmcooumaouu HmoHOoq uoHchoucozumsouu uUHHmcoouasouo HmsuomHHmucH uoHHmcooaoznmsouo uoHHmcoonmsouu unmocmmwncH uoHchoocozumnouw uOHchouumsouu m>HumcHOmEH cmHOmz Hmuoe meocmswmum mHQMHum> ucmvcwmma Gm5CHUCOUII.m mqmdB 113 ON.O NN.O mucmHOwamou womanhoocoo O O0.0 OH O0.0 EOUOHS N OO.N O O0.0 mHzmvcmHHm mans NH NN.OH NH OO.NH coHuHcOoomm HmHoom O NH.N O OO.N pommmmmlemm OH O0.0H OH HN.NH coHum>Hmm OH HN.OH OH OO.NH whammem NH ON.OH NH ON.OH OuHHsomm HmcoHumz N O0.0 H ON.O m>oq musumz O ON.O O ON.N mwofiumm umccH H O0.0 O ON.O mmmchmmm O HN.O N N0.0 Eovmmuh OH O0.0 O NH.O OuHHsomm OHHEMM HH O0.0 HH NH.O NuMHmsmm OH OO.NH OH O0.0H Ousmmm mo OHuoz d O O0.0 N OH.O wommm um OHHOZ d O OO.N O ON.O unmaanHmaooum Oo mmcmm-< NH NN.OH NH NN.O mm qucwaoxm :4 OH NO.NH OH O0.0H mMHH mHQmuHOMEOU « AONuzO AOanO xcmm GOHUmz xcmm cmHUmz mmsHm> uOHHmcoocoz OUHHmcoo .HOO u zv mmsoum uOHHmnoococ Ocm uoHHmcoo mom mxcmn cam .OGMHUOE .mmsHm> HMGHEHmullmm3mco mcHusonllmHmEmm HMCHOHHOII.O mqmma 114 HN.O OH.O muanonmmoo mocmouoonou O ON.O O OO.N UmHHouucOUImHmm N O0.0 N O0.0 mHnHmcommmm OH O0.0H OH OO.NH mefldm. OH O0.0H OH ON.OH MQMHUMQG O OO.N O O0.0 mcH>0H N NH.O OH OO.HH HOOHNOH OH O0.0H O O0.0 Hmmmmwflflmmmfl N NN.O N ON.O usmozmmmOcH OH NH.HH OH O0.0H m>HumcHOMEH H O0.0 H O0.0 mmmmmm NH ON.O OH O0.0 HammHmm O OO.N O O0.0 mcH>Hmuom HH O0.0 OH O0.0H msommwusou NH O0.0H NH O0.0H cam OH NN.OH HH OO.OH flaw O O0.0 O OH.O mHQm mo O O0.0 O O0.0 OmUcHEUmomm OH ON.O NH NN.OH a xcmm cwHOmz xcmm cmHOmz mm5Hm> uoHHmcoocoz uOHHmcou .HOO u zv museum HOHHmcoococ Ocm uoHHmcoo Mom mxcmu can .mcchmE wasHm> HmucwEduumcHllmmzmco mcHusoullmHQEwm HMQHOHHOII.O mamma 115 groups are drawn from the same population. The Wilcoxon statistical technique will be used throughout this chapter as the appropriate measure of the significance of the distribution. Because the Wilcoxon test is used throughout this chapter, it is imperative for the reader to understand the statistical operations involved. It is somewhat un- orthodox to present lengthy statistical analysis in the "findings" chapter of a thesis. But because the Wilcoxon technique is so unfamiliar to most people and so often used in the analysis of this research data, the author has chosen an actual example to illustrate this technique to the reader. Accordingly, the following example is pro- vided. It is easiest to first determine the Wilcoxon W value and then to compute the Mann-Whitney U value via the following formulas: n1(n1 + l) a. U = W - 2 where n1 is the size of the conflict group 1As explained in Siegel, Egnparametric Statistics for the Behavioral Sciences, pp. 75-83 (Wilcoxon), and pp. 116-27 (Mann-Whitney). Actually, the method explained above does not exactly adhere to the formulas presented in Siegel but the results are the same. The author found the above example procedure the easiest to use. This re- searcher is expecially indebted to Dr. James Stapleton, Chairman of the Statistics Department at Michigan State University for his assistance in and explanation of the relationship between W and U. 116 nn 1 2 . 2 where n2 18 the b. Mean U or expected value of U is size of the nonconflict group. 2 / nlnz N3-N_2T NN-l) 12 where N is nl + n2 and ET is the sum of all tied c. Standard deviation of U = t3 - t observations and where T = __T§—— . U - mean U standard deviation of U d. 2 value is Observe the example on the following page. The Wilcoxon W is obtained by summing the values of each rank multiplied by its respective number of individuals in either the conflict group or the nonconflict group. Either the frequencies of the conflict or the nonconflict group may be used, the ultimate 2 score and its interpre- tation will be the same. We will find the W value only for the conflict group. First, the values for each rank should be noted. The initial rank of 4 is obtained by taking the average rank of l, 2, 3, 4, 5, 6, and 7, that is, by add— ing the numbers 1 through 7 and dividing by 7 (a short cut is to add the first and last numbers and to divide by 2). The second rank of 10 is obtained by taking the average rank of 8, 9, 10, 11, and 12, that is, by adding the numbers 8 through 12 and dividing by 5, the number of ties 117 EXAMPLE: A COMFORTABLE LIFE Frequencies l 2 3 4 5 6 7 8 9 Conflict Group 4 4 0 2 1 2 5 1 2 Nonconflict Group 3 l 3 2 0 2 O l 5 Rank 4 10 14 17.5 20 22.5 27 30.5 35 Number of Ties (t) 7 5 3 4 1 4 5 2 7 t3-t T= _fi— 28 10 2 5 O 5 10 .50 28 Frequencies 10 ll 12 13 14 15 16 17 18 Conflict Group 4 3 2 3 2 8 4 3 0 Nonconflict Group 2 2 3 6 5 10 2 3 0 Rank 41.5 47 52 59 67 79.5 91.5 97.5 Number of Ties (t) 6 5 5 9 7 18 6 9 0 t-t T=[-—i—2—] 17.5 10 10 60 28 484.5 17.5 60 o 118 (a short cut is to add 8 and 12 and to divide by 2). The third rank of 14 is obtained by taking the average rank of 13, 14, and 15, that is, by adding the numbers 13 through 15 and dividing by 3, the number of ties (short-cut method is to add 13 and 15 and divide by 2) . . . etc. After rank values have been obtained, W may be com- puted. Using the conflict group frequencies and the rank values: W = 4(4) + 4(10) + 0(14) + 2(17.5) + 1(20) + 2(22.5) + 5(27) + 1(30.5) + 2(35) + 4(4l.5) + 3(47) + 2(52) + 3(59) + 2(67) + 8(79.5) + 4(91.5) + 3(97.5) + 0(18) W = 2408 Now n (n + 1) a. U = w - 1 12 = 2408 — éflééll = 1133 nan b. Mean U or expected value of U is 2 := 1250 c. Standard deviation of U O = nlnz 53:5 _ 2T = //’ 50(50) 1003-100 _ 776 u EYNTTT 12 100(100-1) 12 The ET value is obtained by adding 28 + 10 + 2 + 5 + 0 + 5 + 10 + .50 + 28 + 17.5 + 10 + 10 + 60 + 28 + 484.5 + 17.5 + 60 + 0 = 776 standard deviation of U = approximately 144.5 or 145 119 It should be noted that the standard deviation figure can be computed much easier and quicker if the ET value is eliminated from the formula. The ZT value is a correction for ties figure and is not large enough in this 'study to be of any importance. The standard deviation figure from the above formula less the ET value is again 145. In future reference to the Wilcoxon W, the reader should assume that the ET correction was not included. U - mean U = 1133 - 1250 standard deviation of U 145 d. 2 value is -.807 In the conversion of W to a 2 value, the z score may be either positive or negative, but the sign is actually irrelevant. From the properties of a normal distribution, the following 2 scores and their respective levels of significance for a two-tailed test can be derived: (see Figure 4).2 Level of Significance Reject null .10 if z ‘1 1.645 < -l.645 hypothesis at .05 value is 1 1.960 or < -1.960 under a two- .01 3 2.575 < 2.575 tailed test Figure 4. Size of z scores necessary for rejection of null hypothesis at various levels of signifi- cance (two-tailed test)--Wilcoxon technique. 2The required size of the z score can be found from Appendix D-2. For example, the .05 level of significance for a two-tailed test means that an area of .025 exists at 120 The Wilcoxon technique tests the following null hypothesis: Ho The samples (groups) come from the same popu- l lation or from identical populations. Figure 4 indicates that the -.807 2 value for A Comfortable Life is not large enough to reject the null hypothesis. There is no significant difference in the ranking of this value for individuals in the conflict group as compared to individuals in the nonconflict group. All individual values, both terminal and instru- mental, which had a median difference of 2.00 or more be- tween the conflict and the nonconflict groups were tested for significance. Accordingly, as Tables 4 and 5 indicate the individual values of A Comfortable Life, A World of Beauty, Pleasure, Forgiving, Independent, Logical, and Loving were all tested for significance via the Wilcoxon technique described earlier. Also, the Wisdom value was tested, although the group-median-difference is less than two, because this value proved to be a group distinguish- ing value later in the research.3 either and both ends of the distribution. The .025 value is located from the context of Appendix D-2, and the associ- ated 2 value is found to be 1.9 plus .06 or 1.96. Similar required 2 values can also be found for the .10 and the .01 levels of significance. 3This will be explained later in greater detail when routine stable results are discussed. 121 Only the values Logical (2 value 2.14) and Egg: giving (2 value 1.69) were significant at the respective .05 and .10 levels. Thus, in general, we cannot reject the null hypothesis. The facts reveal that the conflict and nonconflict individuals come from identical popu- 1ations. For thirty-four of the thirty-six values, 1291? viduals in the conflict and nonconflict groups do not sig- nificantly rank terminal and instrumental values differ- ently. The idea of individuals in the conflict group having different "value systems" than individuals in the nonconflict group must accordingly be rejected. The idea of individuals in the conflict group having different "individual values" than individuals in the nonconflict group must similarly be rejected--at least for thirty-four of the thirty-six values. The last statistics in Tables 2, 4, and 5 to note are the concordance coefficients.4 Note that the 4The Kendall concordance coefficient is explained in detail in Siegel, Nonparametric Statistics for the Be- havioral Sciences, pp. 229-38 and in many other modern statistical texts. Briefly the computational formula is as follows: S l 2 3 fik (N N) concordance coefficient = where s = sum of the squares of the observed deviations from the mean of Rj' that is, [ ER.]2 S = Z R. — _l j N 122 coefficients are similar among the three groups (total, conflict, and nonconflict groups) for both the terminal and instrumental values. All of the coefficients are high enough to have a significant value which can be interpreted as meaning that the students are applying essentially the same standard in ranking the values under study.5 The null or more conveniently 2 ZR.2 _ [k(N + 1)] N j 2 k = number of sets of rankings (fifty in this sample and this routine) N = number of entities ranked (eighteen values) —l-k2(N3 - N) 12 maximum possible sum of the squared devi- ations, i.e., the sum 5 which would occur with perfect agreement among k rankings. To work through an actual example would require the author to report all of the value rankings of all of the individuals in one group (conflict or nonconflict) or the other for one or more of the individual values. Instead of doing this, the following figure illustrates how Rj and ZRj are obtained: (see Figure 5). Individual Individuals Values 1 2 3 . . . . . . . . 50 A Comfortable Life R1 An Exciting Life R2 . RANKINGS . Wisdom R18 ZRj = R + R + . . . R 1 2 18 Figure 5. Computation of R' and ZR- in formula for Kendall Concordance Coefficient. 5For example, the lowest concordance coefficient is .18. The significance of this value is found by using the following formula: 123 hypothesis associated with each concordance coefficient is as follows: Ho The (50) sets of rankings are independent or unrelated. N For each coefficient, the null hypothesis is re- jected (via the technique described in footnote 4). The rankings are related, that is, there is homogeneity among all the groups. Note, however, that the conflict group is the least homogeneous, the nonconflict group is the most homogeneous, and the total sample has concordance coef- ficients lying between the coefficient values of both the conflict and the nonconflict groups for both the terminal and instrumental values. x2=k(n-1)w where x chi-square the number of judges = 50 the number of objects ranked = 18 the concordance coefficient 225?? N llllllll If the value of x2 from the above formula equals or exceeds the value shown in Appendix D-3: for the assumed level of significance (.05) and a particular value of df = n-l, then the null hypothesis may be rejected at that level of significance. Even the lowest coefficient (.18) yields a value (153) - = k(n - l) w = 50(17) .18 2=153 significant at not only the .05 level of significance but also at the .001 level too. For a further explanation see Siegel, Nonparametric Statistics for the Behavioral Sciences, pp. 229-39. 124 Original Sample--Routine Stable Routine stable analyzes pairs of roommates (not individuals) in the conflict and nonconflict groups. Actually, two stable routines were undertaken, one routine for the conflict group and one routine for the nonconflict group.6 The author has taken the liberty of combining the data outputs from the two routines and reorganizing the data in a manner which should make it easier for a reader to comprehend, compare, and contrast the data outputs from the two computer runs of routine stable. The first table which follows gives the agreement correlation coefficients (Spearman rho correlations) of each roommate pair in each group (conflict and nonconflict) for each of the two value scales. The correlation coef- ficient measures the degree of relatedness between the value rankings of roommate 1 and roommate 2 who together constitute a pair in one group or the other (see Table 6). The significance of each correlation coefficient can be determined.7 The first thing to notice in the data 6Initially this author made the mistake of running only one routine stable but the data output accordingly mixed results of conflict and nonconflict group pairs, and thus the statistics presented were for the total sample (fifty pairs, twenty-five from the conflict and the non- conflict groups) rather than for subsamples (conflict ver- sus nonconflict groups). This output was of limited use and the statistics that the author really wanted could only be obtained by running separate routine stables for each grOup (conflict and nonconflict). 7See Siegel, Nonparametric Statistics for the Be- havioral Sciences, pp. 212-13. The significance of the 125 mmmo.o homo.o mmmm.o mvhm.o mm mmm©.o mowm.o ¢mmo.o mmHN.o vN mmmo.o moch.o momO.OI mmno.o MN Homo.0l omvm.o momH.o HNmN.o NN wMHN.o mhhm.o mmmo.o Hmmm.o HN oomv.o mdvm.o mmNo.o mmvm.o om Nva.o mmmm.o HmmH.OI ommo.o ma hmHH.o wwwm.o momo.o HNmN.o mH morm.o moam.o mmmw.o HHmH.o NH mMNN.o mvwm.o mmom.o HNmN.o ma mmha.o mmmm.o mmN®.o mmmH.OI ma ommo.OI hmwv.o HHmN.o hMNo.0l VH mmhv.o vam.o HhaH.o wwmm.0l ma m¢m~.o omNH.OI mmmo.o oomv.o NH NmmN.o hmmm.o wmmm.o mflhv.o HH ©MHN.O vHHo.o womm.o mmmO.OI 0H vmm¢.o O¢Hw.o MN¢O.OI hHmN.OI m m¢mm.o momN.o mmmm.o owmm.o m mHHN.OI mmmo.o mmmN.o wmov.o h hmvm.o vav.o m¢Hm.o omHm.o m hmvm.o hm¢m.0l mmv¢.o mmmN.o m mNmm.o maho.o «HH0.0! Noqo.0l v bvmo.OI «mmm.o ommH.o thH.OI m wwwm.o mHhm.o mem.OI Oth.o N NHON.O Hmmm.o vMVH.o mhhv.o H monam> mwsam> mwsHm> mosam> Honesz HmucmfidnumcH HMGHEHmB mem HmucmEduumcH Hmcwfiuma uwmm msouw uoHHmcoocoz moonw uoflamcou .Amm u zv muflmm msoum uoaamcoococ cam HUfiamcoo MOM mucmwoammmoo msam> HmucmEdHumcfl paw Hmcfiaump “Mama comm How mucmfiowmmmoo ucmfimmMOMIImHnmum mcfludonllmHmEMm HmchflHOIl.m mqmde 126 Spearman rank correlation coefficient (rs) under the null hypothesis can be tested by using the following formula: N-2 t = rS l - r 2 s where N = the number of subjects (values = 18) df = degrees of freedom = N - 2 = 16 r8 = Spearman rank correlation coefficient When N is equal to or greater than 10, the value defined by the above formula is distributed as t with df = N-Z. Thus the associated probability under HO (the null hypothesis) of any value as extreme as an observed; value of rs may be determined by computing the t associ- ated with that value, using the above formula, and then determining the significance of that t by referring to Appendix D-4. Using the above formula and by consulting Appendix D-4, Figure 6 has been constructed. Figure 6 re- ports the relationship between t and r (either Sprearman rs or Pearson r) for three values of N, namely, N = 18, 25, and 50. This information will be used throughout this chapter in connection with routine stable as applied to all three samples, namely, the original sample, the cross- validation sample, and the entire sample. Level of t r ___ Significance .10 1.746 .4000 N = 18 .05 2.120 .4683 .01 2.921 .5899 N _ 2 .10 1.714 .3367 t = r 2 N = 25 .05 2.069 .3962 l - r .01 2.807 .5052 .10 1.678 .2355 l N = 50 .05 2.011 .2790 .01 2.684 .3612 L Figure 6. Relationship of t and r for three values of N and three levels of significance (t-test). 127 is that there are twice as many negative correlations in the conflict group than in the nonconflict group (twelve versus six). The negative correlations mean that the association between the value rankings of two roommates is such that as one roommate tends to rank any value "high," the other roommate tends to rank this value low and vice versa. Recalling the data reported in Figure 6, notice that none of the negative correlations are significant at even the .10 level. The following figure summarizes the number of pairs in the two groups which had meaningful posi- tive correlations at various levels of significance: Required Significance Conflict Nonconflict Correlation Level Number of Pairs Number of Pairs .4000 .10 10(6 + 4) 18(12 + 6) .4683 .05 7(4 + 3) 13(10 + 3) .5899 .01 3(0 + 3) 7( 6 + 1) Note: Notice the recounting of pairs, that is, any pair counted in the .01 level of significance classifi- cation will also automatically belong to the .05 and .10 classifications also. Figure 7. Original Sample--Routine Stable-~number of pairs in the conflict and nonconflict groups which had meaningful correlations at various significance levels for the terminal and instrumental value scales (N = 25). PreviouSly, it was noted that negative correlations were twice as common in the conflict group than in the non- conflict group, although none of the negative correlations were significant. Figure 7 reports that individually and 128 collectively (total number of significant pairs) at various levels of significance, nonconflict pairs out- number the conflict pairs by approximately a 2:1 ratio. The parentheses report the respective number of pairs for the terminal and instrumental value scales. The 2:1 ratio also is indicative of the number of significant terminal versus the number of significant instrumental values. In summary, significant correlations are twice as frequent in the nonconflict group than in the conflict group, and within the terminal value scale than within the instru- mental value scale.8 The following table reorganizes the data from Table 6 and in addition reports the median, mean, and standard deviations of the agreement coefficients for terminal and instrumental values for pairs in both the conflict and the nonconflict groups (see Table 7). Notice that none of the median correlations are significant even at the .10 level although the nonconflict terminal value correlation of .392 is close to the required .400. Notice that, in respect to median correlations, the 8The validity and reliability of the computation of this data was questioned by Dr. James Stapleton (Chair- man of the Statistics Department at Michigan State Uni- versity) and others. The main concern is whether a t-test (parametric) can be applied to a Spearman correlation coef- ficient (nonparametric). Siegel, Nonparametric Statistics for the Behavioral Sciences. PP- 202-13, and others believe that such a technique may be appropriate as long as N is 10 or more. In this case, N = 18. The similarity between the Spearman rs and the Pearson r was noted in Chapter III, and everyone agrees that the t-test is applicable to the Pearson r. 129 m \D N N M O H O O mmsHm> Hmucwe asuumcH mam.o wm~.o mh~.o mm o o o H a v mom.o mnm.o Nam.o mm o o a v m N o N o m o a o H o mMSHm> HmcflEumB macho uuHHucoucoz mvm.o Ho~.o mwa.o mm o o o m o a v m v m m a o a o mosam> HmucmE IsuumcH Hmm.o mma.o Nv~.o mm o o o o N v m o a m m m N o o mosam> HmcHEHmB aneuu uoflawcou .Q.m com: andvmz Hmuoa m.o w.o h.o m.o m.o v.0 m.o N.o H.o 0.0 H.0I N.O| m.on v.ol m.on mam>u0ucH .Amm n zv mmsoum uoflawcoococ can uoflawcoo on» such CH muflmm uOM mmaam> Hmucmfiduumcfl paw Hmcflfiumu qu mucmfloqwmmoo unwEmwuom mo coflumfi>mp puwpcmuw Ucm .cme .cmHomE .coflusnfiuumflp >ocmsvouwllmabmum mcflusoulamamsmm Hmcfimwuo|n.h mqm

Hmm mmmm.v oooo.m mN o o o N o N H N H N o H m H H H m N musmmmHm mNmm.N ome.v mN o o o o o o H H N H N N v o N m v H meusomm HmcoHumz Humm.H mva.N mN o o o o o H H H H H H o H m N N m H m>oq madam: oooo.m omnm.v mN o o o o o H o m H H N v o v H H o H NGOEumm umccH mnmm.N omNm.v mN o o o o o o o N m H H N v H N N m H mmmmmmmmm MHmo.N omNH.m mN o o o o o H H H H o o v N H v m H m Eoommum MHmo.m 000N.v mN o o o o H H o N N N H H H m N v N H quusomm NHHEmm oomN.v noww.m mN o o N N H o H H N H m N H H H H v N NMMflmmmm mNHm.N mmmm.v mN o o o H H o o o H v o N m m N v m H wwdmmm Ho pHuoz < mNmm.N oooo.m mN o o o o H o v H o H H N m m N m N o momma um UHuoz d mmmm.N omNm.m mN o o o o H H H H m H H v N v H m H H unmeanHmEoo Io< mo mmcwm d boom.m oooo.m mN o H H o H N o m o N o H m v m o N o mqu mcHuHoxm ca oomN.m oomN.m mN o o H N H H o m N H H N N m N H N H mMHH mHnmuHONEou d 5H 6H ma 4H MH NH HH 0H m m a o m 4 m N H o .o.o cmHomz Hmuoe msHm> mmHocmswmum .AmN u zv muHmm msoum uoHHmsoo HON mmsHm> HmcHEHmu Ham mwocmumwch Ho mcoHumH>mU mHHunmsv can .mcmemE .mmHocmsvmuMIlmHnmum mcHusouulmHQEmm HMCHmHnoIn.m mamas JQ34 NNSN.H nmoo.m mN o o o o o o o o o o H v mmmmmm mmmm.N mmmm.v mN o o o o o o o m o o m v mwzmocwHHm mane comm.N mmmm.¢ mN o o o o o o o v o N H m chuHcmoomm HmHoom Hmmv.m oooo.v mN o o o H H o H H H N N N HommmmmuuHmm HHme.m omNN.N mN H o H o o H N o H o N N coHum>Hmm omNN.H omNH.N mN o H o o o o o m H o O NH musmmmHN mmco.H omNH.N mN o o o o o o H o o H N m wuHusuwm HmcoHumz oomN.N oomn.m mN o o o o o N o o o o N m m>oH musumz HHvN.m oomn.m mN o H o o o o H o m N N h NmOEHmm uwccH moom.N oooo.q mN o o o o o H H o H N N v mmmmmmmmm oomo.N mmmm.q mN o o o o o o H H H N H H aonmmum NNNm.N mmmm.m mN o o o o o o o H N H v m (NuHusomm HHHsmm omNm.N ooov.q mN o o o o o o H H N o N m HmmHmmmm mmmo.N oomN.N mN o o o o o o N o H N o m wammm Ho nHuoz a NNHN.N oomN.m mN o o o H o v H o H H N H momma um nHuoz a vmmq.m oomN.m mN o H o o o o H H N N N N unmecmHHmsou I04 mo mmcmm d HmHH.m oooo.m mN o o o o o m o o N v N m muHH mcHuHoxm cc mNHm.N ooo¢.N mN o H o o o H N o H o H m «HHH mHnmuuoNaoo a NH 6H mH «H NH NH HH OH a N N H .o.o cmHomz Hmuoe m=Hm> mocmummmHo mo unsosm .HmN u zv muHmm msoum uoHHmcooco: Haw mmsHm> . HmcHEHmu How mmocmumwan wo mcoHumH>m© wHHuumsq 6cm .mcmewE .mmHocmswmuwnannmum mcHusounumHmEmm HmchHuoll.m mamcs 135 of each individual terminal value by the two roommates in each and every pair in both the conflict and the noncon- flict groups (see Tables 8 and 9). In anticipation of a "prediction process" to be applied to the unclassified pairs of the cross-validation sample, the author made the following observations on this data. Six terminal values (namely, A Comfortable Life, A Sense of Accomplishment, Equality, National Security, Pleasure, and Wisdom) had medians which differed by 2 or more. The Wilcoxon technique which was described earlier was applied to each of these six values to determine whether the difference distributions between the conflict and non- conflict groups for these six individual values were sig- nificantly different. A z score of 2.17 for A Comfortable Life and a z score of 2.85 for Wisdom was obtained, both 2 scores being large enough to be significant at the .05 level and the Wisdom score also being large enough to be significant at the .01 level too. The reported quartile deviation figures are not too meaningful or helpful as a guide to prediction although small deviations indicate that scores above or below the medians were not scattered appreciably.lo Conversely, a loThe quartile deviation figure indicates that 50 per cent of the total sample had differences above or below the median by the stated amount of the quartile deviation figure. For example, in the conflict group, for the value A Comfortable Life, the median was 6.25, the quartile de- viation is 3.25 indicating that 50 per cent of the entire sample had differences of 6.25 1 3.25, that is, the range was 3.00 to 9.50 for 50 per cent of the sample. The figure 136 large quartile deviation for any value means that the difference scores of the roommate pairs varied considerably for this value. One might expect that the amount of differences would be considerably less for each individual value for members of the nonconflict group as compared with the amount of difference for each individual value by members of the conflict group. This, however, usually did not prove to be so. Especially at the extremes of little or much difference, one might expect nonconflict pairs to have less difference than conflict pairs. Occasionally this was true, but usually this idea also was not supported. For examples: roommates in the conflict group had thirty- three instances where roommates of a pair ranked a value identically (with the same rank). Roommates in the non- conflict group did only slightly better with twenty-five pairs of roommates each ranking eighteen values having but thirty~eight identical rankings. This is a partial eXp1a~ nation of why the Wilcoxon technique revealed that only two of the eighteen values distinguished significantly between the conflict and nonconflict populations (groups). Looking at each value individually rather than collectively also indicated no appreciable differences in the manner in also indicates that roughly 25 per cent (one-quarter) had‘ differences of 3.00 to 6.25 (the median), and that also 25 per cent (one-quarter) had differences of 6.25 (the median) to 9.50. 137 which conflict and nonconflict group pairs differed in their rankings of the values at this low extreme. For example, fourteen roommate pairs in the conflict group had amounts of difference of from zero to five for the value An Exciting Life. In the nonconflict group, thirteen room- mate pairs had amounts of difference of from zero to five for this value. In general, a similar phenomenon existed for all the individual values. This again helps explain the Wilcoxon results. The other end of the frequency distribution is more revealing, at least in a few cases. Usually, differences greater than ten for amounts of differences for any value in either group were not meaningful and/or significant. Accordingly, notice the number of differences greater than ten for each and every terminal value and for both the con- flict and the nonconflict groups. However, there are a few noticeable exceptions to this generalization. For the values of Wisdom, Pleasure, and Equality, the occurrence of a difference of ten or more in the roommates rankings of these values usually indicated that the roommate pair belonged to the conflict group rather than to the non- conflict group. Accordingly, three conflict pairs had a difference of over ten on the value of Wisdom but no non- conflict pairs had a difference this large. Similarly, five conflict pairs versus one nonconflict pair had a difference of ten or more on the value of pleasure, and six conflict pairs versus one nonconflict pair had a 138 difference of over ten on the value Equality. How this information was used in the "prediction process" will be explained again a little later in this chapter. The median difference figures along with one other measure of the agreement of each individual value are pre- sented next. The additional agreement measure is the Pearson correlation between value rankings for roommate l and roommate 2. Each of these two measures is ranked from least to most difference (see Tables 10 and 11). The first thing to notice in this data is that the ranks are considerably different for any individual value between the two measurement techniques. Thus, depending upon which technique one prefers, he may get a markably different value profile (ranking) than he would have ob- tained if he had chosen another measuring technique. The median difference statistic has already been discussed. It is now desirable and necessary to observe and analyze the correlation measure which is usually considered to be a more powerful tool. It is necessary to determine the significance of 11 the correlation coefficients. Referring to Figure 6 and 11Now being used is a Pearson product-moment corre- lation coefficient rather than a Spearman rank order corre- lation coefficient. Although the methodology involved in computing each statistic is identical, the former technique is conventionally used with parametric statistics and the latter with nonparametric data. It was previously ex- plained that the significance of the Spearman rho could be tested using the formula: 139 I, .mocmHmMMHp umofi Op ummmH Eoum Umxcmn musmmme numm« mH thv.0| mH oomN.m mmmmmm OH wome.on n mmmm.v mHsmpcmHHm mane N anm.o v mmmm.m coHuHcmoomm HMHoom m mmHo.o mH bmwo.m HommmmmIMHmm H Noov.o H omnm.H coHum>Hmm NH HHmm.OI mH oooo.m musmmmHm mH mNHN.on 0H omNm.¢ muHusomm HmcoHumz m mHmH.o N .BNNH.N m>oH manna: m MHNN.o m omhm.v acoEHmm HmccH v amON.o m omNm.v mmmchmwm m mmmH.o m omNH.m Eovmmnm NH HmNH.OI m OOON.v muHusomm meEmm mH OONm.OI NH nmmw.m NMMflmmmm n NNNo.o m mmmm.v Nusmmm mo UHH03 a m mmvo.ou NH oooo.m momma um UHHoz 4 HH mqmo.on vH omNm.m ucmficmHHm30004 mo mmcmm d vH thN.ou HH oooo.m mMHH mcHuHoxm ad mH hmmN.on NH oomN.m mMHH mwnmummmammld xcmm .Huoo xcmm cmHomz dem> «.38 u 5 9:3 mdoum uoHHmcoo How mmsHm> HmcHEnmu now mxcmu cam .COHHMHmHHoo N mumEEoon on H mumafioon .mocmummep GMHmeIImHnmum mcHusoulumHmEMm HmsHmHHOII.0H mqmda 140 .mocmHmMMHU umoE on ummmH Bonn Umxcmn mHSmmmE somm« H momm.o m Nmmm.m .ammmmm vH mNNo.o NH mmmm.v mwsmocmHHm mama NH mva.on mH mmmm.v coHuHcmoomm HMHoom NH mmmN.ou OH oooo.v mommmmmumem m HmHm.o m omNm.N :onm>Hmm NH Hmso.o N omNH.N mmmmmmmm N ommv.o H omNH.N NUHHsomm HMGOHumz m mmNm.o N oomN.m m>0H mnsumz NH HHOH.o N oomN.m Hmoaumm umcaH m Nmom.o HH oooo.v mmmmmmmmm mH HHoo.o 4H NNHN.N .mmmmmmm m mmON.o NH mmmm.m NMHusomm NHHEmm OH NHHH.o mH OOON.N .Hmmmmmmm v mmNm.o m oomN.m Nusmmm mo oHHoz 4 6H NNmo.on NH oomN.m momma mm mHuoz-« HH NOHH.o m oomN.m ucmEanHmEooo< mo mmcmm m N NNHN.o 6H oooo.m mHHHWmaHmHoxm am a mmNH.o H OOON.N mHHH mHnmmuomaoo a xcmm .nnou xcmm cmHomz mSHm> All! «.HmN u zv mHHmm moonm HUHHmcoococ How mmsHm> HmcHEHmu How mxnmu pnm.c0HumHmHuoo N mumEEoon o» H mumEEOOH .mocmHmMMHU GMHUmEIImHnmum msHusoullmHmEMm HmchHHOII.HH mqmfla 141 realizing that N is now 25, the following facts are ob- served. Disregarding the levels of significance, the con- flict group has ten negative correlations whereas only three negative correlations exist for the nonconflict group. For the conflict group, Pleasure is significant at the .10 level and Salvation and Wisdom are significant at the .05 levels. Not surprisingly, the Pleasure and Wisdom significant coefficients are negative. The Sal- vation agreement coefficient, however, is positive.12 For the nonconflict group, significant positive agreement correlation coefficients exist for National Security at the .10 level and for Wisdom at the .01 level. Thus, room- mates who get along well rank Wisdom similarly. It was The Pearson coefficient is tested similarly using the formula: The Spearman and Pearson methodologies are both sum-of-the-squared-differences techniques. For the pur- poses of this study, rS can be assumed to equal rxy so that the two above formulas are equivalent. Notice, how- ever, that the value of N is no longer 18 (the number of values) but is instead 25 (the number of subjects or pairs in each group). 12From routine oneway as applied to the original sample, it is known that the median of Salvation for the conflict group is 17.21 and that the median of Salvation for the nonconflict group is 16.50. Thus, in general, 142 noted previously that no nonconflict roommate pairs had a difference of over ten on the Wisdom value, and, in fact, referring again to the frequency distribution, it can be observed that only one nonconflict pair had a difference of seven or more on the ranking of this value. Explanations of and reasons for the reported corre- lations, both positive and negative and belonging in both the conflict and nonconflict groups, are not too difficult to hypothesize. Each person, however, might conceivably conjecture a different explanation for the observance of these phenomena. The correlations to date are not too puzzling, although later correlations will prove to be more difficult to explain and rationalize. In any event, when strong correlations exist, whether they appear to support common sense or to defy the laws of logic, this author will not devote much if any time trying to explain or rationalize the correlations. The objective is to report rather than to explain these correlations. Thus far in discussing and analyzing "individual values,‘ only data reported for the terminal value scale has been examined. It is necessary now to look at all of the previously described techniques which analyze the degree of pair agreement on each individual value as they apply to Salvation is ranked very low for each group but conflict group members apparently mutually rank it lower than do nonconflict group members. Conflict group members accord— ingly have more of a mutual disregard for Salvation than do nonconflict group members. 143 the instrumental value scale. Initially, again, it is necessary to observe and analyze the frequencies, medians, and quartile deviations of differences between roommates in pairs belonging to both the conflict and nonconflict groups. The following two Tables report this information (see Tables 12 and 13). A median difference of 2.00 or more for pairs in the conflict group versus pairs in the nonconflict group exists for only two individual values in the entire instru- mental value scale--namely, the values Ambitious and Self- Controlled. To be expected, nonconflict pairs agree on the importance of Ambitious much more than do conflict roommate pairs. However, it is somewhat surprising that conflict roommate pairs agree more on the importance of Self-Controlled than do nonconflict roommate pairs. The twenty—five roommate pairs of students in the nonconflict group (each roommate in each pair ranking the eighteen instrumental values) had identical rankings of an individual value on thirty-six occasions. This figure contrasts with a tally of seventeen for the analogous figure in the conflict group. However, the sums of the total amounts of differences for differences ranking zero to five reveal that such data in general are meaningless as an attempt to identify whether a pair belongs to either the conflict or nonconflict groups.l4 Only the individual 14It is interesting to note that identical "logi- cal" ranks occurred five times for nonconflict pairs but 144 mNHv.N mmmw.N mmmw.m oomN.m mNmo.m mmvm.N vvom.m Nomv.m oomN.N wmvw.N mNom.n wmo¢.N Nvoo.N NmNN.m mwNN.H mmmm.N mmmm.H oomv.m omNm.N ooov.N Nwom.m oooo.m mmmm.v omNm.m omNm.m oooo.m mmmm.m omNm.m mmmm.m omNH.m oomN.m oooo.m oomN.m 000N.v omNm.m oooo.N mN mN mN mN mN mN mN mN mN mN mN mN mN mN mN mN mN mN pmH ouucooumHmm mHQHmcommmm mmHHom mcmHmmno mcH>OH HmonoH HmsuomHHmucH ucmpcmmmvcH m>HumchmEH ummcom Hmmmmmm mmmmmmmmw mmmmmmmmmw flu Hmmmmmmw mmmmmmw 6mmcHemmoum wsowmdmafl .o.o cmHmmz Hmuoa NH oH mH vH MH NH HH 0H m m mocmumNMHQ Mo uGDOE< mnHm> .HmN u zv mHHmm msoum uoHchoo new mmSHm> HmucmESHumcH now mmocmumume mo mcoHumH>m© mHHuumsv ocm .mcmemE .mmHocmsvmuulannmum mcHusounlmHmEMm HmchHuonl.NH mqmda L145 NNNH.H ONNN.N NN o H o H o N H N N H H N H N o N N N omHHoumcoo-NHmm ONNN.N ONNN.N NN o o o o o o H N N H H H N H N NN N N mHnHNcommmN NNNo.N ONNN.N NN N o H o o H o o H H N o N H N H N N mmHHoN NNHN.N ONNN.N NN o o o o o o o H N o N N N N N N N N mmmmmmma ONNN.H NONN.N NN o o o H H o H o o N o N N N N N N H mmmwmm NNNN.N ONNN.N NN o N o o o o H o N o N N H N H H N N Hmmmmmm NNNH.N NNNN.N NN o H o H o N N H N o o H N N N N N H HmsummHHmmcH NONN.N ONNN.N NN o o o H o N N o N N o H N v H N N o mammcmmmmcH NHNo.N ONNN.N NN o H o o o H H H H v H H H N H N N H m>HmmchmsH NNNN.H ONNN.N .NN o o N o o o o o N o N o N N N N H N mmmmau NNNH.N NNNH.N NN o H o o N H H o o H o o N N v N N H Hmmmwmm NNNN.N NNNN.N NN o o o o H H H N N o H N N N N H N N Nmmmmmmmm NNHN.N NNNN.N NN o o o H o o N H o H N N N N H N N H mmmmmmmmmw NNNN.N NNHN.N NN o o o o o H H N H N o H H N N N v H .mmme NNNN.N NNNN.N NN o o o o o o o N H H N H N N N N N N Hmwmmmnu NNNN.N ONNN.N NN o o o H o o o o H N N o H N N N N H mmmmmmm NNHN.N NNNN.N NN o o H o o H o o o N H N H H N N N H nmncHamNoum NNNN.N ONNN.N NN o o o o H o N H o H N N N N N N N H msoHuHaaa NH NH NH NH NH NH HH NH N N N N N N N N H o .o.o cmHUmz Hmuoe msHu> mocmummmHQ mo u:50&4 .HmN n zv muHmm moonm uoHHmcoocos HON mmsHm> HmucmaduumCH now mmocmumwuHU wo mcoHuMH>m© mHHuHmsw can .mCMHomE ~mmHucdeumuwllmHnmum mCHudouIImHQEMm HMCHmHHOII.mH mqmfia 146 value of Ambitious shows any marked contrast as fourteen occurrences of amounts of differences from zero to five are reported for the nonconflict group and the corres- ponding figure for the conflict group is seven. For differences of greater than ten, again, con- sidered as entities, the conflict and nonconflict groups vary insignificantly. Individually, the values of Honest and Ambitious are perhaps noteworthy. No nonconflict pairs had a difference of greater than ten on the value of Honest, but two conflict pairs had differences this large. Seven conflict pairs had differences of greater than ten on the value of Ambitious, but only three nonconflict pairs had differences this large on this value. When the Wilcoxon test of significance was applied to each of the individual instrumental values, none of the values had a z score large enough to even be significant at the .10 level. The Ambitious value was the only value close to significance having a z score of 1.621 (a value of 1.645 being required for .10 significance). zero times for the conflict group. This helps account for the fact that Lo ical was the only value reported in rou- tine oneway Wthh Significantly (at the .05 level) dis— tinguished between individuals in the conflict group versus individuals in the nonconflict group. This may be a chance occurrence, but more than likely, the Logical value is in some way interpreted by the students to correspond to the value of Wisdom in the terminal value scale--the strong positive and negative correlations of the Wisdom value respectively for the nonconflict and conflict groups have already been noted. 147 The quartile deviation figures were again not very useful indicating only the general spread of differences above and below the reported median. The next two tables report the median difference along with the roommate l--roommate 2 correlation coef- ficients and the subsequent ranks revealed by these two measures of the pair agreement on each individual value within the instrumental value scale for each of the room- mate pairs in the two groups (see Tables 14 and 15). Again notice the wide differences in rankings corresponding to the two different statistical techniques. The median difference figures will not be analyzed further since they were presented in the previous discussion. In- stead, the analysis will focus upon the more meaningful correlation coefficients. Referring again to Figure 6, for the conflict group, nine negative correlation coefficients exist (disregarding statistical significance), and three significant positive coefficients exist for Broadminded (.10), Cheerful (.05), and Self-Controlled (.05). For the nonconflict group, six negative correlations exist (disregarding statistical significance), and two values, namely, Honest and Cheerful have positive and significant correlations at the .05 level. Since the Cheerful correlation was positively correlated at the .05 level for each group, this value cannot distinguish whether roommate pairs belong to either the conflict or the nonconflict group. 148 .QUCGHGMHHU UmOE OUN ummmH EOHM UMXCMH mhflmmmg SUMWNN. H NNNN.N N ONNN.N omHHoumcooumHmN N NNHN.N H ONNN.N mHnHNcommmm N NNNH.o m NNNN.N mmmwmm NH NNNo.o- N ONNN.N mmmmmmmm HH NNHN.N- NH NNNN.N mmmwmm NH NNoH.ou NH ONNN.N Hmmmmmm N NNNo.o NH ONNN.N HmsmomHHmmcH NH NNNo.o- NH ONNN.N mcmncmmmocH mH mmHH.on «H mmmm.m m>HumchmEH N NNNo.o N NNNN.N mmmmmm NH NNNH.ou NH NNNN.N HmmmHmm NH NNHN.N- NH NNNH.N NNH>HNHom N NHNH.o N ONNN.N msommmusoo NH HNNN.N- HH ooo.m mmMHm N NNNN.N N ONNN.N Hamummno N HNNo.o N ONNN.N mmmmmmm N NONN.O N ONNN.N ommaHemmmmm NH NNNN.N- NH ONNN.N NsonHnaa xcmm .HHOU xcmm cmHUmz mnHm> «.HmN u 29 mHHmm msoum uoHHmnoo How mmsHm> HaunmEdnumcH How mxcmu new .coHHMHmHHoo N mumEEoou on H mumEEoou .mocmnmmmHo CMHmerlmHnmum msHusonllmHmEMm HmsHmHHOII.vH mqmds 149 .QUHHOHGMHHU #mOE On. HmmmH EOHM UGMCMH whflmmwg £00W£ NH NNNN.N- NH ONNN.N omHHouucouumHmN N NNNN.N N ONNN.N mHnHmcomNmm N NNNo.o N ONNN.N mmmmmm NH HHNH.N- N ONNN.N mmmmmmmm N NNNH.o NH ONNN.N mmmwmm N NNNH.N NH ONNN.N .mmmmumm NH ONNN.N- NH NNNN.N HmsuomHHmuaH NH NNNo.o NH ONNN.N usmnammmmcH NH NNNo.ou NH ONNN.N m>HumcHNmeH N NONN.O N ONNN.N .mmmmmm NH HNNN.N- N NNNH.N mmmmmmm N NNNN.N NH NNNN.N NcH>HNmoN m vaH.o Ha bmmm.v msomwmnsou NH HNNN.N- N NNHN.N mmmmm H NHoN.o N NNNN.N mmmmmmmm N NNHH.o N ONNN.N mmmmmmm HH NNNo.o H NNNN.N umNaHewmoum NH NNNo.o NH ONNN.N msoHanem xcmm . HHOU Xfimm CMHUGE wDHM> «.Amm fl zv muwmm msoum HUNHmcoocon How mmsHm> Hmucmaduumcw How mxcmu can NcoNumamuHoo m mumfifioou ou H mumEEoou .mocmummwNG nMNvaIImHnmum mcflusoullmamEMm HchmNHOII.mH mqmda 150 Again, the main goal is to report and use (in the prediction process) the analyzed data rather than to ex- plain and interpret individual value correlations. To summarize routine stable as applied to the original sample, Spearman rho data revealed that instru- mental values conceived as an entity or system do not sig- nificantly distinguish conflict group pairs from noncon- flict group pairs. However, terminal values conceived as an entity or system do significantly distinguish conflict group pairs from nonconflict group pairs. Concerning indi- vidual values rather than value systems, the Pearson corre- lation statistic revealed that for terminal values, the conflict group pairs had a negative significant corre- lation for Pleasure (.10), a positive significant corre- lation for Salvation (.05), and a negative significant correlation for Wisdom (.05). For terminal values, the nonconflict group pairs had a positive significant corre- lation for National Security (.10), and a positive signifi- cant correlation for Wisdom (.01). For instrumental values, the conflict group pairs had a positive significant corre- lation for Broadminded (.10), a positive significant corre- lation for Cheerful (.05), and a positive significant correlation for Self-Controlled (.05). For instrumental values, the nonconflict group pairs had a positive signifi- cant correlation for Honest (.05), and a positive signifi- cant correlation for Cheerful (.05). However, the Wilcoxon test of statistical significance between two populations 151 (conflict and nonconflict groups) revealed that roommate pairs in the two groups only had statistically different rankings on two terminal values, namely, A Comfortable Life (.05), and Wisdom (.01). The Wilcoxon test revealed no statistically different roommate pair rankings among the instrumental values for the two groups (although the Ambitious value was close to .10 significance). How some of this information was used to formulate a "prediction process" is explained in the immediately following section of this chapter. Group Prediction of Unclassified Pairs The next goal of the research project was to utilize the information from routines oneway and stable about pairs in the original sample to predict pair-group membership for unclassified pairs in the cross-validation sample. The initial task was to form a "prediction pro- cess" from the original sample data. The prediction pro- cess was to be derived from statistically significant data, but was to be applied subjectively. That is, significant value correlations and median figures (significance in- directly implied from Wilcoxon test on the total frequency distribution for any specific value) were noted and the researcher was then to analyze the Rokeach Value Surveys of unclassified pairs of the cross-validation sample by observation, and accordingly to classify the pair as be- longing to either one group or to the other. Thus, no 152 statistical analysis of unclassified pairs occurred before these pairs were predicted by the author. Thus, the pre- diction process used was clinical and subjective rather than systematic and statistically reliable and valid. The prediction process was statistically formulated, but non- statistically applied. No precise prediction process could be determined. Recalling the results of routines oneway and stable as applied to the original sample, very few individual values proved to be significant. The few sig- nificant values and their corresponding quartile deviation figures, which usually indicated that conflict and non- conflict members did not differ greatly, made a scientific, systematic, fool-proof method of prediction an impossi- bility. The goal was to predict unclassified pairs by observation (thus, intuitively or subjectively), and thus correlation information was not obtained on the unclassi- fied pairs to aid in the prediction process.15 Thus, correlation information from the original sample was only indirectly and subjectively applied by the author who mentally conjectured a correlation figure after visually observing the differences in the individual value rankings for two roommates of any and all pairs in the unclassified sample. Thus, by utilizing most of the significant infor- mation contained from routines oneway and stable as applied 15After the prediction process was applied, such information was obtained. 153 to the original sample, the following prediction process was formulated by the author (see Table 16). The "steps" of the prediction process are listed roughly in their order of importance. Steps 1-11 are from pair information contained in Tables 8-15. Steps 12 and 13 are from information presented in routine oneway which re- vealed that only the value Logical at the .05 level and the value Forgiving at the .10 level were significant in identifying whether or not an individual belonged to the conflict group or to the nonconflict group. When the actual prediction process was being applied, steps 10-13 were not given much weight (influence) in determining 'whether or not the author predicted a pair as belonging to one group or the other. The prediction process more or less formulates a stereotype or profile for conflict and nonconflict pairs. For each group, the crucial values are listed; their corre- lation coefficients, levels of significance, quartile deviations, and median differences are also identified, and from the stable "frequencies" information the number of pairs (in each group and for each key value listed) with identical rankings and with differences of greater than ten is also noted. The key variable in the pre- diction process is the correlation coefficient. Notice that the steps (which are roughly in their order of im- portance) begin with the highest correlation coefficient and proceed to become increasingly smaller. The "median .Hma mo uso av msoum uoHHucoococ cH uoc anmnoum .NM mocmumwpr cmvaE wH .Hm no use ov msoum uofiaucoococ cN uoc .NA mocmumMMNp cmapwE uH .mEMm may uN omxcmu muNmm uowaucoococ m>Hm pan meow mnu Hmoflmoq pmxcmu mufimm uoHchoo 02 fi. .154 NH. NH.N NH HN NN.N NN.N NN N HN.N NH. NcH>HNNoN NH NN. NN.N NN HN NH.N NN.HH NN N NN.N NN. .mmmmmmm NH N-H NHm NH“ NaH mmao mmNu meo mmNu NN.N -wa NN.N HN.N Hmsuod Hmsuoc Hmsuod Hmsuod Hmsuo¢ Hmsuum NN.N N N NN.N NN.N N H NN.N acmsnmfimmsooom «0 mmcmm N HH HN.N N N NN.N NN.N N H NN.N NNHH mHnmuuoNsoo 4 NH NN.N H H NN.N NN.N NN.- N N NN.N NH.v wmmmmmmm N NN.N H N NN.N NN.N NN.- N N NN.N NH.v msoHuHQEN N NN.H N H NH.N NN.N NN.- N N NN.N NH. musmNmHN N HN.N H N NN.N NN.N NN.+ N N NN.H NH. NmNcHecmouN N NN.N N N NN.N NN.H NN.+ N N HN.N NN. coHum>HNN N NH.N H H NN.N NN.N NN.+ N H HN.N NN. NNHHouucooanmN N NH. NN.H H H NN.+ NH.N NN.N H H NN.N ammm Isomm Hmcoflumz m NN. NN.H N N NN.+ NN.N NN.N N H NN.N ummcom N HN. NN.H N N NN.+ NN.N NN.N NN.: N H NN.N NN. «ammmwm H . H . . . uoHHmcoocoz uoHHNcoo .mmmooum :oNuonoumllmaQEmm HchmHuOss.oH mamde 155 differences" variable proved also to be an important criterion in the prediction process in many cases. The "level of significance" and "quartile deviations" infor- mation was not very useful in actual predicting except to give the author an idea of the reliability of the re- spective values for the correlation coefficients and the median differences. The frequency information could be used only in special cases; sometimes this information assured the author that a certain pair had to belong to one group or to the other, often this information gave the author a high percentage guesstimate, but usually this information was useless.16 For examples, the author was "assured" that a pair belonged in the conflict group if the median difference on the value of Wisdom was greater than ten (indeed, as the asterisk on the table indicates, no nonconflict pairs had differences of greater than seven on this value). An example of a high percentage guessti- mate is that if a pair had a difference of greater than ten on the value Equality, six times out of seven occur- rences this pair belonged to the conflict group. The fre- quency information, however, was usually useless for generally pairs did not have identical values or values differing by greater than ten units. Although the table 16This statement and the ones to follow apply reliably only to the prediction of already analyzed origi- nal sample pairs. To be noted shortly, original sample pairs were "predicted" before the prediction process was later applied to the cross-validation sample. 156 appears to indicate a lot of occurrences of identical and greater-than—ten frequencies, this is an illusion since in number of pairs such frequencies were rare and more often than not one "odd" pair accounted for many of the identi- cal or greater-than—ten instances (that is, the same pair, for example, could have the same ranking or differences of greater than ten on more than one value). In addition, the frequency information was of limited use because the per- centage guesstimate was often no better or not much better than a chance guess. For example, if the pair had a difference of greater than ten on A Comfortable Life, five times out of nine this pair belonged to the conflict group. Also, although the values of Honest, Salvation, and Broadminded have significant correlations for one group or the other, it was difficult to use and apply this infor- mation in predicting due basically to small variances in the median differences for these three values for the two groups (note also the small quartile deviations). For example, although Honest has a +.50 correlation for the nonconflict group, the median difference for the noncon- flict group is 3.00 and the same figure is 3.87 for the conflict group. The quartile deviations for each group (conflict = 2.84 and nonconflict = 1.46) are small again illustrating the difficulty of using such information in the actual prediction process. The values of Pleasure, Ambitious, and Equality were somewhat helpful in the prediction process. The 157 levels of significance for the correlations (all of which are negative for the conflict group) are all around .10. The median differences for the values between the two groups vary by over 2.00 in each case but the quartile deviation figures are also large indicating wide fluctu- ation in median difference values for each value for each group. As indicated before, the values of A Comfortable Life and A Sense of Accomplishment were not in reality Very helpful in the prediction process, but since the median differences between the two groups for the two values were greater than 2.00, these values were accord- ingly admitted into the prediction process. The data presented in the prediction process (Table 16) on the values Logical and Forgiving come from routine oneway. Notice that the median and frequency figures reported are the "actual" medians and frequencies rather than the "differences" for these two values. This, accordingly, gives us a somewhat different interpretation of the quartile deviation figures which again are large and thus make the actual values less reliable as pre- dictive devices. Pragmatically, as their position on the table indicates, these values were not of much help in the actual prediction process. However, as the asterisk points out, in a few special cases, information about these two 158 values "assured" the author that a pair had to belong to one group or to the other.17 This then was the prediction process. The key prediction variable was the correlation coefficient. This variable, however, was only used indirectly. In predicting the pairs, the author did not have correlation coefficients reported for each of these key values for the pairs. If the researcher had had actual agreement coefficients for each value (and also stability coefficients for each pair), undoubtedly, the prediction process would have been easier, more accurate, and more reliable. The author thus did not utilize coefficient information directly but indirectly by looking at the median differences for key values in each pair and then "inferring" appropriate correlation coef- ficients for the differences. In the above manner, then, the author went through the various steps in the prediction process. Recalling the design of this research, twenty-five pairs of conflict roommates versus twenty-five pairs of nonconflict roommates have been analyzed. Remaining are twenty-five pairs of conflict roommates and twenty-five pairs of nonconflict roommates who have filled out and returned Value Surveys but whose Surveys as yet have not been analyzed at all. The goal is (after mixing these latter twenty-five pairs in each group together) to predict 17Again, this is true only of the prediction pro- cess as applied to the original sample. 159 to which group each pair belongs. Before doing this, how- ever, the author applied his prediction process to the initial twenty-five pairs in each group as a test of the prediction process.18 In doing so, the author initially classified twenty-eight pairs in the conflict group and twenty-two pairs in the nonconflict group, but the author also noted which pairs appeared to be borderline cases. By re-examining borderline cases the groups were equalized 19 Checking with twenty-five predicted pairs in each group. the number of correct and incorrect predictions revealed six incorrect pair predictions in each group for a total of twelve incorrect predictions in fifty cases (pairs). 18The pairs were mixed up and identified by an impartial third party in a manner in which the group identity of each pair was unknown to the author during the time he was applying the prediction process. 19Actually, the idea of equalizing the number of pairs in each group has questionable merit. If the number of pairs in each group is equal, then if the author incor- rectly identifies a pair (that is, he places it in the wrong classification), automatically another pair in the other group is also incorrectly identified. It is thus not by chance that six pairs were wrong in each group be- cause including a pair in a group automatically excludes this pair (and other pairs) from the other group (assuming that the total number of pairs in each group must be equal). Perhaps a higher correct percentage of predictions could have been obtained by not trying to equalize the number of pairs in each group. This is questionable, however, since in this case the 28-22 pair initial classification only involved reclassification of three pairs (decreasing twenty- eight and increasing twenty-two by three). Another factor to be considered is that the equalization of the number of pairs in each group was deliberately accomplished by the author because he smugly, and perhaps foolishly, recognized the possibility of predicting all pairs correctly. Ob- viously, you cannot predict all pairs correctly if you have unequal groups. 160 Thus, the prediction process applied to the original sample of conflict and nonconflict group pairs from which data determining the prediction process came revealed that cor- rect predictions were made in 76 per cent (thirty-eight out of fifty) of the cases. A note of interest, perhaps, is that out of eleven borderline cases, the author correctly classified eight and missed three. The other nine in- correct predictions were due to a faulty prediction pro- cess rather than due to the author's improper interpre- tation of the Value Surveys. That is to say, that nine cases seemed to defy the criteria, at least to some degree, established for each group in the prediction process. The author reanalyzed the twelve missed pairs and made a few minor adjustments in the prediction process none of which were substantial. Review of the results of the prediction process indicated that, indeed, the best criteria were the first four steps in the prediction process and that pre- diction decisions based on steps 5-13 often were incorrect. All in all, the author was quite pleased with the results and anticipated similar results upon application of the prediction process to the "unclassified" pairs of the cross-validation sample. When applying the prediction process to the cross- validation sample (the unclassified pairs--twenty-five pairs of conflict and twenty—five pairs of nonconflict mixed up and identified again by an impartial third party so that their proper group identity was unknown to the 161 author) initially twenty-seven pairs were classified in the conflict group and twenty-three pairs were classified in the nonconflict group. Eight pairs were borderline cases and these were re-examined and two pairs were switched to equalize the groups.20 To the author's amaze- ment, the results revealed that twenty-four of the pre- dicted pairs were classified in the wrong group and only twenty-six pairs were properly predicted. This indicates just chance probability and roughly a chance ability of the author to predict the proper group classification of a pair. Of the eight borderline cases, half were properly identi- fied, the other half being improperly identified. The author's initial reaction, besides shock, after viewing the results of the prediction process as applied to the cross-validation sample, was that the impartial third 20Actually in both the initial and the cross- validation samples virtually all cases were "borderline" in some respect (criterion) or another; that is, almost all cases (pairs) contained some contradictory information. However, the author reserved the borderline classification basically for those pairs which had contradictory criteria throughout and within the various stages of the prediction process. The author attempted to use this classification as little as possible for, in reality, there was no border- line group and thus the author forced himself to make a conflict-nonconflict decision whenever and wherever possi- ble. Borderline cases represented indecision and were only useful when trying to equalize the number of pairs in each predicted group. Only borderline conflict cases were re- examined to any great extent. In both the original and the cross-validation samples, the initial result of the pre- diction process revealed unequal groups with too many pairs in the conflict group. Thus, some additional nonconflict pairs were needed, or conversely, some conflict pairs had to be eliminated, and thus borderline conflict cases were re-examined and somewhat useful in attempting to balance the number of pairs in each predicted group. 162 party must have somehow mixed up her identification system. But checking and rechecking the pairs revealed that the pairs had properly been identified and that, indeed, the prediction process was a failure. Apparently, the pre- diction process was rather successful (76%) when applied to pairs which actually made up the data from which the prediction process was derived. But, the prediction pro- cess was a failure when applied to the unclassified pairs of the cross-validation sample. To summarize this chapter section, no systematic, statistically reliable or valid prediction process could be derived from routine oneway and/or routine stable in- formation from the original sample. Instead, a subjective, intuitive, clinical prediction process was formulated and applied indirectly (and by mere observation) to both the original and cross-validation samples. The clinical pre- diction process was 76 per cent successful when applied to pairs which actually made up the data from which the pre- diction process was derived. But, the prediction process failed when applied to the cross-validation sample pro- ducing only chance probability of proper pair-group classi- fication. Thus, no statistically reliable and valid pre- diction process could be ascertained either scientifically or subjectively. 163 Cross-Validation Sample--Routine Stable The author originally did not plan to perform any statistical analysis on the cross-validation sample (this is probably due to the fact that this researcher antici- pated that the prediction process would be successful). However, apparently some or all of the steps mentioned in the prediction process are not valid. Thus, as an endeavor to explain why the prediction process failed, the author decided to run additional tests on the cross-validation sample. Since the results of routine oneway were insignifi- cant in the original sample, it was decided that this routine would not be run on the cross-validation sample.21 Even if it were run, all it would do is support or refute steps 12 and 13 of the prediction process which as ex- plained earlier were relatively unimportant prediction criteria anyways. However, again two routine stables were run, one for the conflict group and one for the noncon- flict group of the cross-validation sample. The data out- put from these routines has again been reorganized in a manner designed to aid visual representation and reader understanding. The results of routine stable as applied to the cross-validation sample is the subject matter of this section of this chapter. It is necessary to compare and contrast the origi- nal sample of twenty-five pairs of conflict and twenty-five 21Time wise and economically also, the application of routine oneway to the cross-validation sample could not be justified. 164 pairs of nonconflict roommates with the cross-validation sample of twenty-five pairs of conflict and twenty-five pairs of nonconflict roommates. To begin, observe the agreement coefficients for each pair of roommates in the cross-validation sample (see Table 17). Table 17 must be compared and contrasted with Table 6. Again the negative correlations should be noticed for the cross-validation sample. Twelve negative correlations exist for the con- flict group versus nine negative correlations in the non- conflict group (in the original sample these respective figures were twelve and six). Corresponding to a similar tabulation for the original sample, Figure 8 summarizes the number of pairs in the two groups of the cross- validation sample which had significant positive corre- 1ations at various levels of significance: Required Significance Conflict Nonconflict Correlation Level Number of Pairs Number of Pairs .4000 .10 17(11 + 6) 24(15 + 9) .4683 .05 12( 8 + 4) 21(13 + 8) .5899 .01 7( 5 + 2) 10( 8 + 2) Note: Notice the recounting of pairs, that is, any pair counted in the .01 level of significance classifi- cation will also automatically belong to the .05 and .10 classifications also. Also, only positive correlations are countered. Figure 8. Cross-Validation Sample--Routine Stable-- number of pairs in the conflict and non- conflict groups which had meaningful corre- lations at various significance levels for the terminal and instrumental value scales (N = 25). 165 NNNm.o| NNNo.o NN NHNo.o NmmH.o mN mmNN.o ONNN.N NN MNmm.o MNmN.o NN OHNO.N: mNNN.o MN NmNm.o mmNN.o mN mNOH.o NNoN.o NN NHNN.ou NHNN.o| NN NNNo.o| NNmm.o HN MNmN.o NONN.O HN NNNN.o NmHN.o om NNMN.o NNNN.o ON mmNN.o| NNNH.o NH NNoN.o NNNN.o mH HNNN.o NmNN.o NH NNHH.o mmNN.o NH NNOH.o Nmmm.o NH NNoH.OI NNMN.o NH ONNo.o| Nmmm.o NH NNmN.o NNMN.o NH NNNm.o NHON.o NH NNmm.o| OHNO.N: NH mmNo.o mmmH.o NH ONNN.N NONH.o NH oooo.HI oooo.H NH NMNN.o| NNHN.N MH NMNo.o mNNN.o NH NNON.o| NmNm.o NH NNNN.o| ONNN.N HH NNNH.oI mNmN.o HH NNNN.o NmNN.o OH NNHN.N moom.o 0H NNNH.o| NmNo.OI m NNMN.o NNNN.o m HmNN.o HmHN.o N NMNH.o mmmm.o N mNNN.o NHNN.o N NNoo.o HHNN.o N NmNN.o ONNN.N N NMNo.o NHHN.o N mmmo.o NNNN.o m NNNN.o mmoo.o N NmmN.o NHNN.o N ONNN.N NNNN.o N HNNN.o HNmm.o m m0N0.ou HNNH.o m mNNN.o mNmm.o N omNo.oI ONNN.N N ONNN.N mmNN.o H ONNN.N: omNo.oI H mmDHM> mmDHw> Hmflfiflz m05am> mGDHM> Hmflgfiz HmucmfiduumcH HmcHEHwe HNmm HmucmEdHumcH HmcHEumB “Ham muonu uOHchoocoz msouo HUHHMGOU .HNN n zv mnHmm msoum uOHHmcoosoc cam “UHHmcoo How mucmNonwwoo msHm> HmucmEsHNmCH paw HMGHEHmu ”HHmm some How mucmHOHmmmoo quEmmHNMIImHQmum mcHusoulanmEmm GOHuMUNHm>ImmOHUII.NH mqmde 166 The parentheses again indicate the number of corre- lations in the terminal and instrumental scales respec- tively. Again, the number of significant correlations is greater within the nonconflict group rather than the con- flict group and within the terminal value scale rather than the instrumental value scale.22 Notice that the 2:1 ratio of significant correlations in the nonconflict group versus significant correlations in the conflict group which was established in the original sample is not supported by the data result of the cross-validation sample. The 2:1 significance ratio of terminal to instrumental values, however, again exists. The following table reorganizes and reports the re- sults of additional statistical tools applied to this same data (see Table 18). Table 18 of the cross-validation sample is analo- gous to Table 7 of the original sample. For the cross- validation sample, the median correlations of the non- conflict group are again larger than the correlations of the conflict group for both the terminal and instrumental value scales. Again, for each group, the median coef- ficients for instrumental values are not significant. The 22Notice the perfect correlation in both the posi- tive and negative directions for the terminal and instru- mental values respectively of pair number thirteen of the nonconflict group. Obviously, a pair of roommates deliber- ately ranked their values accordingly as a prank. The correlations from this pair then somewhat bias the reported results. 1(57 oov.o NmH.o mNH.o mN o o H N H m H H N N m H N H H moDHm> HmucmE usuumcH NNN.o ovv.o mmv.o mN H H H N m m H N N N H o o o o mmnHm> HmcHEuoe maouo uoHHucoosoz mHm.o MNH.o Nmo.o mN o o H H o N N N N v m N m o H mwaHm> . Hmucmfi usuumCH mmN.o NNm.o omn.o mN o o H N H N m m m N N o H o o mmsHm> HmcHEume msouw uoHchou .N.N cum: cchmz Hmuoe N.N N.N N.N N.N N.N N.N N.N N.N H.N N.N H.N: N.N- N.N- N.N- N.N- mHm>uoucH .HNN u 2N mmsoum uoHchoococ 6cm uoHchoo ecu anon CH mHHmm HON mosHm> HmucmEduumcH new HmcHEHmu new mucmHonwmoO ucwsmmumm mo :oHumH>op oucncmum New .cmme .cmemE .coHuanuume NocwsqmuwlumHamum mcHusouauonemN coHumvNHm>nmmouU|n.NH mqm<9 168 median correlation of terminal values for the nonconflict group has increased from just under .10 to the .05 level of significance. The median correlation for the terminal values of the conflict group is approaching the .10 level of significance. Again it is necessary to test the mean figures for significance. Applying the significance test which was explained during the similar discussion concerning the original sample, a t value of 1.53 is obtained for the terminal value scale and a t value of .32 for the instru- 23 mental value scale. Both values are too small to reject 23The formulas involved are: A‘— 2 2 _ ///1nl - 1)sl + (n2 - 1)s2 S— n - n - 2 1 2 17*? l 2 For the terminal values: _ 24(.262)2 + 24(.258)2 _ s — 48 - .26 _ .440 - .327 _ t ’ .26(.284) ' 1'53 For the instrumental values: /24(.400)2 + 24(.318)2 48 _ .156 - .123 _ t ‘ .36(.284) ‘ '32 169 the null hypothesis and we, accordingly, must accept the fact that there is no significant difference between the mean of the nonconflict group and the mean of the conflict group. Thus, for the cross-validation sample, in general and in contrast with the original sample, neither the pair correlation coefficients of the instrumental value scale (taken as an entity) nor the pair correlation coefficients of the terminal value scale (taken as an entity) differ sig- nificantly between conflict and nonconflict group pairs. This data help to explain then why the prediction process was a failure. So much for the agreement coefficients of each pair in the cross-validation sample. Now, again it is necessary to shift attention from "pair" agreement to "value" agree- ment, and correspondingly to change the focus of the analy- sis from the study of "scales" of values to the study of "individual values" within the scales. Again first the terminal values and later the instrumental values will be analyzed. Tables 19 and 20 present the frequencies, medians, and quartile deviations of differences between pairs of roommates respectively in the conflict group and in the nonconflict group of the cross-validation sample (see Tables 19 and 20). Tables 19 and 20 correspond to Tables 8 and 9 respectively of the original sample. Whereas in the original sample where median differ- ences of greater than 2.00 existed between the conflict and nonconflict groups for the terminal values of A Comfortable 1370 Nmmo.m ONNN.N mN Eovax MNoN.N oomN.N NN chmUcmHum mDHB MNNH.N omNH.N mN coHuHcmoomm HmHoom NNHm.N NNNN.N NN uommmmmamem mNNo.N OOON.N mN :oHum>Hmm NNNN.H mmNm.m mN musmmmHm HHmN.N oooo.N mN NuHusomm HmcoHumz NNHm.m NNNN.N mN m>OH musumx NNHN.N ONNN.N NN Ncofiumz umccH Nmmo.m omNN.N NN mmmmmmmmm Nmmm.N oooo.N NN EOGmmuh MNoN.N mmmm.m NN muHusomw NHHEmm mHmm.m oooo.N mN _waflmmmm comm.N NNNN.N NN musmmm mo UHuoz d mmmm.m NNNN.m NN momma um vHuoz 4 MNoN.N ONNN.N NN ucmeanHmamo :04 mo mmcmm d mmmm.N oooo.m NN meH mCHuHoxm ad NNmm.H oooo.m mN meH memuu0mEou d NH NH mH NH NH NH HH OH m m N N m N m N H o dd :38: H38. 93; mocmumuwHQ no ucsofid .HNN u zv mHHmm msoum uoHchoo HON mmsHm> HmcHEHmu HON mmocmumwwHU mo mQOHHMH>mc mHHuHmsv 6cm .mCMvaE .mmHocmsvmuwlsmHnmum mcHusouuanmEmm :oHumvHHm>nmmouolu.mH mqmae l7]. NNNN.H NNNN.N NN N N N N N N N N N N mmmmmm NNNN.N NNNN.N NN N N N N N N N N N H chmNcmHHN mane NNNN.N ONNN.N NN N N H N N H H N N H coHqumoomm HmHoom NMNm.N NNHm.N mN o o o o o H o o N H uommmmmeHmm mHNm.N oomN.H mN o o o o N o H H o o coHum>Hmm NONN.N ONNN.N NN o o o H o o o H o o musmmmHm NNNH.N mmmm.N NN o o o o H H o H o o muHusomm HmcoHumz HNmN.H NNNN.N mN o o o o o H o o N H m>OH muzumz NNHN.N oomN.m mN o o o o H H o H N N NGOEHM: HmccH NNNN.N NNHN.N NN N N N N N N N N N N mmmmmmmmm NNNN.H oooN.N mN o o o o o H H H o o EOUmmHm omNH.m ONNN.N NN o o o o o o o H N m NuHusomm.NHHEmm NNNN.N NNNN.N NN N N N N N N N N N N Nmmmmmmm NNHN.N ONNN.N mN o o H o N o o o H m husmmm mo UHHOK d NNNN.N NNNN.N NN N N N N N N N N N H mommm um NHuos 4 NNHN.N NNNN.N NN N N N N H N N H N H pamenNHHmsoo nod mo mmcmm 4 ooom.H NNNN.m mN o o o o o o o H H H mmHH mmwuHUxm ca cmNH.m mmmm.N mN o o H o H N o H o N mMHH mHnmuHOMEOU d NH NH mH NH NH NH HH OH m N .o.0 cchmz Hmuoe msHm> mocmumNmHo No ucs054 .HNN u zN mHHmm msoum uoHchoococ How mmsHm> HmcHEHmu How mmoqmumwva mo mcoHumH>m© mHHuumsv cam .mcmvaE .mmHosmsUmHmulmemum msHusounlmHQEmm :oHumoHHm>nmm0HUIl.oN mqmde 172 Life, A Sense of Accomplishment, Equality, National Secur— ity, Pleasure, and Wisdom; only two of these median differ- ences of greater than 2.00 between groups carried over into the cross-validation sample. Only the values of A Sense of Accomplishment and Wisdom continued to have median differences of 2.00 or more in the cross-validation sample. Besides these two values, median differences of 2.00 or more between groups in the cross-validation sample also appeared for the values of A World at Peace, Inner Harmony, Mature Love, and Self Respect. None of these latter values had group median differences of this magnitude in the original sample. One key reason the subjective prediction process failed is that it incorporated in the decision process "significance" to individual values which had a median difference of 2.00 or more between the conflict and noncon- flict group pairs. The Wilcoxon technique illustrated earlier that of the six terminal values with median differ- ences of 2.00 or more between conflict and nonconflict group pairs, only A Comfortable Life (.05) and Wisdom (.01) sig- nificantly distinguished between pairs in these two groups. In the cross-validation sample, the Wilcoxon technique and Figure 4 reveal that significant 2 scores exist only for Wisdom (z = 2.45) and Self Respect (z = 2.09), both values being significant at the .05 level. Obviously, subjective "significance" is no substitute for statistical signifi- cance. 173 Additional insight into the matter of individual terminal value agreement is provided in the following two tables (see Tables 21 and 22). Again, Tables 21 and 22 of the cross-validation sample need to be compared and con- trasted respectively with Tables 10 and 11 of the original sample. Again referring to Figure 6, it is revealing to compare the two samples. None of the significant correlations of the origi- nal conflict group are replicated in the cross-validation sample--and in fact, most of these differ dramatically. The strong negative Wisdom correlation has become positive as have the formerly negative and significant (or near significant) correlations for Pleasure anquuality. The original Salvation coefficient is also no longer signifi- cant. In the cross-validation sample appear three differ- ent significant values, namely, A Sense of Accomplishment (.05), Happiness (.10), and Self-Respect (.10); none of which were even close to significance in the original sample. For the nonconflict group, the original sample had significant correlation coefficients for the values of Wisdom (.01) and National Security (.10). The Wisdom value continues to be significant at the .01 level, but National Security now has virtually no correlation whatso- ever. In addition, the cross-validation sample identifies new significant values which did not appear in the original sample. The values of Mature Love (.05), An Exciting Life 174 .mUCQHGMMHQ UmOE On. #mmmfl EOHM UmvHGMH MHHHWMGE SUmmfi N NHNN.N NH NNNN.N somez NH HHNN.N- HH NNNN.N mHamncmHum mane N NNHN.N N NNNH.N coHuHcmoomm HmHoom .NH NNNN.N- NH NNNN.N HommmmmumHmm N NMNH.o m OOON.N COHum>Hmm N NNNH.N N NNNN.N .mmmmmmmm NH NNNN.N- H NNNN.N NuHusomN HmcoHHmz NH HHoo.ou NH NNNN.N m>0H musumz HH OHmo.o NH ONNN.N NGOEHmm umccH H mmmm.o N ONNN.N mmmchmmm NH NNNH.N NH NNNN.N .ammmmmm m NNHN.N N mmmm.m NMHHsomm NHHEmm N NNNH.N N NNNN.N _Nmmmmmmm N NNNH.N NH NNNN.N Nwsmmm mo NHuoz N NH NNNN.N- NH NNNN.N mommm um NHHoz N NH ONNN.N: NH oooo.N ucmEanHmEooo< mo mmcmm d NH NNNN.N N NNNN.N NNHH NcHHHoxm :4 N NNHN.N N NNNN.N mNHH mHanuoNaoo N HNNN .uuoo xcmm NNHNmz msHm> *.Amm 9 2V mHHmm muonm HUHHNGOU Mom mmsHm> HmcHEHmu How mxcmu Una .QOHumHmHHOU N mumBEoou 0» H mumEEoou .mocmnmmec amvafillmHnmum mcHusoullmHmEMm GoHumvHHm>lmm0HOIl.HN mHmNB 175 .mocmummme umoE on ummmH Eonm mecmu musmmmE somm« N NNNN.N N NNNN.N .mmmmmm MH NmNo.o NH ONNN.N mwzmcamHHm mums NH NHNN.N NH NNNN.N soHuHcNoomm HNHoom N NNHN.N N NNHN.N HommwmmumHmN H NHNN.N H NNNN.H :oHuN>HmN HH HNNN.N NH NNNN.N mmmmmmmm mH NNmo.o N mmmm.N NuHusomm HMCOHumz N NHNN.N N NNNN.N m>0H muaumz N HNNN.N N NNNN.N N:o&nmm umccH NH NNNH.N NH NNHN.N mmmchmwm NH NNNN.N N NNNN.N _mmmmmmm N NNNN.N N NNNN.N NuHHsomN NHHENN N NNNN.N NH NNNN.N .xmmHmmmm NH HNHN.N- NH NNNN.N .Nusmmm «0 NHuoz 4 N HNHN.N N NNNN.N momma um NHuoz N NH NNHN.N NH NNN.N HamssmHHHEoooN No mmcmm N N NNNN.N HH NNN.N NNHH quuHoxm cN NH NNNN.N NH NNNN.N NNHH mHnNHHoNsoo N xcmm .Huoo xcmm CMHUmE msHm> «.HNN n zN muHmm maoum NUHHmcoono: How mmsHm> HmcHEHmu How mxcmn can .mGOHuMHmHHoo N mumEEoou on H mumEEoou .mocmummmHU QMHNmEIImHnmum mcHusoullmHmEMm coHumcHHm>lmm0HO|l.NN mHm¢H 176 (.05), A World at Peace (.05), Equality (.10), Inner Harmony (.10), and Salvation (.01) all now have signifi- cant positive correlations. Again, the contrast to the original sample is striking. It is also revealing to compare and contrast the same information between the original and cross-validation samples for the instrumental values. Once again the analy- sis begins by looking at the frequencies, medians, and quartile deviations for the instrumental values of both the conflict and nonconflict groups (see Tables 23 and 24). Tables 23 and 24 of the cross-validation sample must be compared respectively to Tables 12 and 13 of the original sample. In the original sample, conflict versus nonconflict median differences were greater than 2.00 for only two values, namely, Ambitious and Self-Controlled. For the cross-validation sample, the Ambitious difference is now exactly 2.00 and the difference of greater than 2.00 no longer exists for the Self-Controlled value. Instead, new differences of greater than 2.00 now appear for the values of Cheerful, Forgiving, and Intellectual. Again the origi- nal sample versus the cross-validation sample differences are acute. In terms of the Wilcoxon test, no instrumental values on the original sample significantly distinguished between the conflict and nonconflict groups. For the cross-validation sample only the individual value of 1377 NNNN.m NNNN.N NN H o H @mHHouuCOUINHmm NNHN.H ONNN.N NN o N o meHmcommmm NNNN.N NNNN.N NN H N N wuHHoN NNON.H NNmN.N NN o H o acmvabo NNNN.N NNNN.N NN H H H mcH>0H NHNN.H ONNN.N mm H H H HMUHmoq NNNN.N ONNN.N NN H o N HmsuomHHmucH NNHN.N ONNN.N NN H N N ucmncmmmch NNHN.N NONN.N .NN o N H m>HumchmEH NNNN.N ONNN.N mm H N H ummcom NNNN.N NNNH.N NN N N H HstHmm NNHN.N mmmm.N NN N m m mcH>Hmuom NNHN.N NNNN.N NN N N N mmmmmmmmmm NHNN.H NNNN.N NN N o o cmmHU mmmN.N ONNH.N NN N N N stummnu NNHN.H OOON.N NN H H o mHQMQmu mmmo.m ONNN.N NN o H N UmvcHEUmoum ONNN.N oooo.N NN N N N msoHanfid NH NH NH NH MH NH HH oH m N N N N N m N H o .o.o :mHomz Hmuoe wsHm> mocmummea No u::0&< .ANN u zv muHmm msouo uOHHNcoo HON mmsHm> HmucmEsuumcH HON mmocmHmNNHv No mcoHumH>mU mHHuHmsw 6cm .mcmHUmE .mmHocmsvaNuumHnmum mCHusounandEmm coHquHHm>ummouOIl.NN mqmde 178 NNNN.N oooo.N NN H H N @mHHOHHCOUINHmm mmmm.m NNNH.N mm o m H meHmcommmm NNNN.N ONNN.N NN m H H .muHHom NHNo.m ONNN.N NN o H N uanpmno NNHN.N NNNN.N NN N H N mmmmmm NNNN.N NNNH.N NN H N H mmmmmmm ONNN.N NNNN.N NN m o H HmsuomHHmucH NNNN.N ONNN.N mm o N N ucmvcmmmvcH NNHN.N ONNN.N NN N o o m>HumchmEH NNNN.m mmmm.N NN H o H ummcom NNNN.N NNNN.N NN N H N HummHmm NNNN.N NNNN.N NN N H H mummmmmmm ONNN.N NNNH.N NN o N N msommmusoo NNNN.H NNNH.N NN H o o :mmHO NNNN.N ONNN.N NN N o o Hamummco NNNN.N ONNN.N NN H m o mHnmmmu NNNN.N mmmm.m NN H o m @mccHEUmoum NNNN.N oooo.N mm o N H msoHanE¢ NH NH NH NH NH NH HH 0H m N N N N N m N H o .o.o cmHomz Hmuoe msHm> mocmummeQ mo undead .HNN u zv muHmm msouw uoHHmcoococ How mmsHm> HmucmfiduumCH HON mmucmumeHv mo mCOHumH>mN mHHuHmsv mam .mcmHomE .mmHucmswmuwlrmHnmum mcHusounudeEmm coHumeHm>|mmouunu.NN mqmde 179 Forgiving has a z score (2.61) large enough to be signifi- cant, the 2 value being large enough for significance at the .01 level. Additional information about roommate-roommate correlations for each individual instrumental value for both the conflict and the nonconflict groups of the cross- validation sample is provided in the next two tables (see Tables 25 and 26). Tables 25 and 26 of the cross-validation sample are analogous to Tables 14 and 15 respectively of the original sample. In the original conflict sample, four values had significant or near significant correlation coefficients, viz., Self-Controlled (.05), Cheerful (.05), Broadminded (.10), and Ambitious (under .10). Again, when comparing these correlations with their respective correlations in the cross-validation sample, we see that they have not only changed but that they have changed by surprisingly large amounts. The formerly positive 0.4683 coefficient for Self-Controlled is now -0.3494; the formerly positive 0.4539 Cheerful coefficient is now -0.4063; the formerly positive 0.3602 Broadminded coefficient is now -0.3461; and the formerly negative -O.3226 Ambitious coefficient is now positive at 0.0555. Broadminded and Self-Controlled are now significant at the .10 level and Cheerful is sig- nificant at the .05 level. No new correlations besides those mentioned above appear as significant coefficients 180 .mocmummmHN umoE on ummmH Eoum Umxcmu musmmmE comm« NH NNNN.N- NH NNNN.N NNHHouucoouNHmN N NNNN.N N NNNN.N mHnHNcomNmm N NHNN.N N NNNN.N .mmmHmm NH NNNN.N- H NNNN.N mmmmmmmm N NNHH.N N NNNN.N mmmmmm N NHNH.N: NH NNNN.N Hmmmmmm N NNNo.oI NH ONNN.N HmsuomHHmuCH N NNNN.N N NNNN.N ucmvcmmmNcH HH HNNH.oI NH ONNN.N m>HumchmEH N NNNN.N- N NNNN.N .mmmmmm NH NNHN.N- HH NNNH.N _Hmmmmmm NH NNNH.N: NH mmmm.N mcH>Hmuom NH HNNN.N: NH NNNN.N msommmusoo NH HNNH.N- N NNNN.N NNMHN NH NNNN.N- NH NNNH.N Hmmmmmmw H NNNN.N N NNNN.N MHmmmmm NH HNNN.N- N NNNN.N umccHsumoum N NNNN.N NH NNNN.N NNOHHHHEN xcmm .HHOU xamm GMHUGE 05Hm> .HNN u ZN mHHmm msoum HOHHMGOU How mmsHm> HmucmesuumcH HON mxcmu can .GOHHMHmHHoo N mumEEoou op H mumEEoou .mosmummch cMHUmEIImHQMUm mchsoullmHmEMm GOHumeHm>ImmOHOII.NN mqmfia 181 .wocmumMMHU pmoE on ummmH scum wmxcmu madmmma nomm« HH NNNH.N- NH NNNN.N NNHHouuaoouNHwN NH NNNH.N- HH NNNH.N mHnHNcomwmm . NH NNNH.N- N NNNN.N mmmmmm N NNNN.N H NNNN.N ucvamno NH NNNN.N- NH NNNN.N .Nmmwmm N HNNH.N NH NNNH.N Hmmmmmm N HNNN.N N NNNN.N HmsuomHHmucH N HNNH.N NH NNNN.N NewcammNNcH NH NNNN.N- NH NNNN.N m>NuNcNNNsH NH NNNN.N- NH NNNN.N mmmmmm N NNNN.N N NNNN.N Hmmmmmm H NNNN.N N NNNN.N NcH>NNuom NH NNNN.N- NH NNNH.N msommmusoo N, HHNN.N N NNNH.N cmmHo NH NNNN.N- NH NNNN.N Hmmmmmmw N NNNN.N N NNNN.N mmmmmmm NH NNHN.N- N NNNN.N chcHewmonm N NNHN.N N NNNN.N NaoNuHNEN v—Gmm . .HHOU Mfimm GMHUOE GDHMNV .Hmm u zv mHHmm msoum uoNHmcooco: Mom mmDHm> Hmucmfisuumcfl How mxcmn cam .GOHuMHmHHoo N mumfifioou on H mumfifioou .wocmHmMMHU nMHUmETImHQmum mcHusoutlmHmfimm coflumvflHm>lmmouoll.mm mamma 182 in the cross-validation sample.24 Obviously, the original versus cross-validation sample differences are substantial. In the original nonconflict sample, two values had significant correlation coefficients, viz., Cheerful (.05), and Honest (.05). Again the comparisons are lucid. The formerly positive 0.5012 Cheerful coefficient is now -0.2545; the formerly negative -0.334l Helpful coefficient is now positive at 0.4948; and the formerly positive 0.5007 Honest coefficient is now negative at -0.2434. In addition to the now significant Helpful coefficient (.05), new sig- nificant correlation coefficients now exist in the cross- validation sample for the values of glean (.05), and ESE? giving (.05). Needless to say, again the results obtained from the two samples are inconsistent. Having compared the stable routine outputs for the original and cross-validation samples, it is not hard, and in fact, it is extremely easy to understand why the pre- diction process was a failure. It is easy to summarize and review many of the comparisons between the stable out- puts of the original and cross-validation samples by refer- ring again to the "prediction process." The following table (Table 27) should be collated with the formerly 24It might be noted in passing that the interpre- tation of most of these new correlations make more "common sense" than did many of the old correlation coefficients which were originally reported. This is true of many of the new cross-validation correlation coefficients in both the conflict and nonconflict groups and in both the termi- nal and instrumental value scales. 183 NN.N NN.N NN.- NNNENNHHmsooom No NNNNN N HH NN.N NN.N NHHH mHnmuuoNaoo N NH NN.N NN.+ NN.N NH.+ Nmmmmmmm N NN.N NN.N NN.+ msoHansd N NN.N NN.N NH.+ ucmmmmHm N NN.N NN.N NN.- nmnaHecmoum N NN.H NN.+ NN.N NH.+ coHum>HNN N NN.N NN.N NN.- NNHHouucowuwwa N NN.N NN.+ NN.N NuHusomm HmaoHumz N NN.N NN.- NN.N mmmmmm N NN.N NN.+ NN.N NN.+ .ammmmm H mocuumMMHo ucmHonwmou mocmummmwa ucmHonmmoo cmemz coHHMHmHHOU CMHpmz coHuMHmHHOU msHm> mmum uoHHNcoocoz uoHHNcoo .mmmooum coHuoHanmllmHmEMm GOHumpHHm>Imm0HUII.NN mqmfis 184 presented Table 16. Table 27 only provides information about correlation coefficients and median differences since these two variables were the only key variables applied during the prediction process. Notice in step 1 of the decision process that the formerly negative -.42 correlation for the conflict group is now positive +.22 in the cross-validation sample. The strong positive correlation still remains for Wisdom in the nonconflict group. Honest, which was second in importance as a pre- dictor, now, instead of having a +.50 correlation for the nonconflict group, has a negative -.24 figure! The median difference figure, contrary to the original sample, is also now larger for the nonconflict group than for the conflict group. National Security, which was considered to be the third best predictor, has lost its +.46 correlation in the nonconflict group and the cross-validation coefficient is now-+.04. Again, contrary to the original sample, the median difference figure is larger for the nonconflict group rather than for the conflict group in the cross- validation sample. Self Control, formerly viewed as the fourth best predictor, has reversed significant correlations for the conflict group from +.46 to -.35. Salvation again has a significant correlation in the positive direction--but it has switched from the original conflict group to the 185 cross-validation nonconflict group! The relationship of the magnitudes of the median differences for the conflict and nonconflict groups has also reversed itself from the original to the cross-validation sample. The correlation coefficient for Broadminded in the original and cross-validation sample groups is roughly the same . . . but the sign is different!! The Pleasure coefficient for the conflict group has changed from nega- tive significance to positive insignificance. Also for Pleasure, the median difference figure is now greater for the nonconflict group rather than the conflict (the reverse was the case in the original sample). The Ambitious corre- lation coefficient has changed for the conflict group from -.32 to +.05. The -.32 original conflictEguality coef- ficient has become +.18 and now a +.37 Eguality coefficient appears for the cross-validation nonconflict group where no previous significant such correlation existed in the origi- nal nonconflict group. A Comfortable Life now has a larger median difference figure for the nonconflict group rather than the conflict but the reciprocal relationship existed in the original sample. And lastly, a strong negative A Sense of Accomplishment coefficient of -.48 now appears in the cross-validation conflict group although no signifi- cant corresponding correlation existed in the original sample. Thus, apparently none of the steps in the decision process proved to be reliable except, perhaps, step number 186 one. These cross-validation statistics not only explain why the decision process was a failure, but they also make the outcome of roughly a 50-50 chance for a correct pre- diction seem, perhaps, high in view of the fact that many of the steps in the decision process seem to have reversed themselves as group predictors. To summarize the pre- diction process, it failed for two basic reasons: (1) it was a subjective process which by observation either only implied "significance" or attempted to use statistical significance indirectly, and (2) the data reported from routine stables for the two samples (original and cross- validation) were usually inconsistent (and thus unreliable). To summarize routine stable as applied to the cross- validation sample, in general the results were found to be quite different from the results of the same routine as applied to the original sample. For the cross-validation sample, routine stable output reported the following re- sults. Spearman rho data revealed that both terminal and instrumental value scales conceived as entities or systems do not significantly distinguish conflict group pairs from nonconflict group pairs. Concerning individual values rather than value systems, the Pearson correlation sta- tistic revealed that for terminal values, the conflict group pairs had a negative significant correlation for A Sense of Accomplishment (.05), a positive significant correlation for Happiness (.10), and a negative significant correlation for Self-Respect (.10). For terminal values, 187 the nonconflict group pairs had a positive significant correlation for An Exciting Life (.05), a positive sig- nificant correlation for A World at Peace (.05), a posi- tive significant correlation for Eguality (.10), a posi- tive significant correlation for Inner Harmony (.10), a positive significant correlation for Mature Love (.05), a positive significant correlation for Salvation (.01), and a positive significant correlation for Wisdom (.01). For instrumental values, the conflict group pairs had a negative significant correlation for Broadminded (.10), a negative significant correlation for Cheerful (.05), and a negative significant correlation for Self-Controlled (.10). For instrumental values, the nonconflict group pairs had a positive significant correlation for 91222 (.05), a positive significant correlation for Forgiving (.05), and a positive significant correlation for Helpful (.05). However, the Wilcoxon test of statistical significance be- tween two populations (conflict and nonconflict groups) revealed that roommate pairs in the two groups only had statistically different rankings on two terminal values, namely, Self-Respect (.05), and Wisdom (.05). The Wilcoxon test revealed only one statistically different roommate pair ranking among the individual instrumental values for the two groups, namely, the value of Forgiving at the .01 level. 188 Entire Sample--Routine Oneway The author previously mentioned how the application of routine stable to the cross-validation sample was not initially intended. Similarly, this researcher did not initially intend to apply routines oneway and stable to the entire sample. However, because routine data outputs were in general inconsistent between the original and cross- validation samples, the author chose to apply the two rou- tines to the entire sample in order to see which relation- ship persisted throughout the entire sample, and in order to make additional generalizations about the samples in this research study. The entire sample consists of the original sample plus the cross-validation sample. Thus, this and the next section of this chapter merely reports much of the same type of information that has previously been reported-- only now in a combined form. The reader should now be familiar with the type of statistical techniques applied to the data. Accordingly, in the remainder of this chapter, only "findings" from routine oneway and routine stable as applied to the entire sample will be reported. To con- serve time and space, a detailed account of the methodology and procedure has been avoided as has an item-by-item or value-by-value comparison and contrast of the entire sample with the original and cross-validation samples. The reader is encouraged to pursue such avenues for himself. The 189 author will provide a brief summary comparison among the three samples in the final chapter of this report. This section of this chapter reports the results of routine oneway as applied to the entire sample. The final section of this chapter reports the results of routine stable as applied to the entire sample. Table 28 reports the frequency distributions, medians, and quartile devi- ations for terminal and instrumental values for the entire sample (neglecting subsample and subgroup classifications so that N = 200). Table 29 reports the medians and ranks for terminal and instrumental values for these 200 indi- viduals (entire sample). Next, the same information is reported but now the 200 individuals are split in half forming the two groups under analysis in this study, namely the conflict and the nonconflict groups. Thus, Tables 30, 31, and 32 provide the same information about the entire sample as did Tables 28 and 29, but now the information has been reorganized and reclassified into the conflict and nonconflict subgroups. The author will summarize the preceding five tables as briefly as possible. Notice the similarity for the rankings of both scales of values among the entire, con- flict and nonconflict groups. Notice also the small differences in medians reported for the entire, conflict and nonconflict groups. No medians for any of the values in either scale had a difference of 2.00 or more (or even close to this amount) between the conflict and nonconflict :190 NNNN.N NNNN.NH NON ON NN NN NH NH NH NH N N N N N N N H O N H Hmuoe u msouo NMHHsowm HmcoHumz NNHN.N ONNN.N NON H N N N N N NH N NH NH HH N N NN NH HN NN NH Hmuoe n msouo m>0H musumz NNHN.N NNHH.N NON N N N N N N N NH NH NH N NH NH NH NH NH NH NH Hmuoa u muouw Ncosumm HmccH NNHN.N HNNH.N NON N N N N N N N HH N NH NH NH NH N HN N NH NN Haves n msouo mmmsHmmmm NNNN.N NNNN.N NON O H H N N N N N N OH NH NH NN NH HN NH NN NH Hmuoe u msouo Ecummuh ONNN.N OOON.N NON O N N NH N NH NH NH NH N NH NH ON NH N HH N N NNNN.N HNNN.N ONN N NH NH HH NH N N N N NH NH NH N NH NH NH NH H Hmuoe u macho NuHHmsmm NNNN.N NNNN.NH NON NH NH NN NH N NH NH N NH NH HH N N N N N N H Hmuoa u anouw Nwsmwm mo pHuoz d NNNN.N NNNN.N NON H N N N HH HH N NH N HH NH NH N OH NH NH HH NN Hmuoe u macaw momma um UHHOS m NNNN.N OOON.N NON N N N N N HH NH NH NH NH OH NH NH NH N NH OH NH ku09 u macho unmEanHmEoo ION wo mmcmm d NNNN.N NNNN.N NON N NH N HH NH NH NH N NH NH NH NH N HH N HH N N Hmuom u muouu meH NcHuHoxm ca NNNN.N NNNN.NH NON N NH NH NN NH NH HH NH NH OH N N N N N N N N HWuom u mfiwuo mmHH mHnmuuomEou 4 . . NH NH NH NH NH NH NH HH OH N N N N N N N N H anmHum> o 0 cmem2 Hmuoa usmpswmmo mmHocmnwmum .HNON n zv mdem> HmucmfiduumcH paw HmcHEHmu HON mcoHumH>mN mHHuumsw Nam .mGMHme .wsoHuanuume Nucmswmuwlummzmco mcHusouuumHmfimm wHHucmun.NN mqmda 191 HNNH.N NONN.N OON N N NH NH NH NH HH N HH OH NH NH NH HH NH NH OH N Hmuoe u maouu msowmmusoo NNNN.N NNNN.NH NON ON NN HN NH HN NH NH N N OH N OH N N N N N H Hmuoe u msouu smeo NNNN.N HNNN.NH OON N NH NH HH NH OH NH NH NH HH N NH N NH OH N N N Hmuoa u macaw Hsmuwmnu NNNH.N NNNH.N NON N H N N N NH HH ON NN NH NH N NH NH N N HH HH Hmuoa n maouu mHnmmmu NNNN.N NNNN.N NON N N N N N N N N N HH NH NH NH ON NH NN NH ON Hmuoe u maouw omchEUmoum NNNN.N NNNN.OH OON OH NH NH HH N NH N NH OH NH NH NH N NH N NH N N Hmuoa u msouu mSOHanad NNNN.N HHHH.N NON N N N N N N OH NH HH NH NH NH NH NH NH NH ON NH Hmuoa n msouo Eopme NNNH.N HHNN.N NON H H H N N N NH NH NH NH NH NH NH NH NN NH NH N Hmuomlm macho mwnmpcmHum «one NNNN.N NNNO.NH NON NN NH NN NN NH NN HH HH OH N N N N N N N N H Hmuoe u msouu :oHMHc Hwoowm HMHoom OOON.N OOON.N OON H H N N N HH N NH NH NH ON OH NH NH NH NH NH NH Hmuoa u maouu vommmmmumHow NNHN.N OOON.NH OON NN ON NH N OH N N N N N H H N H N N O NH Hmuos u macaw coHum>Hmm NNNN.H NHNN.NH NON HH HN NN ON ON NH NH NH HH N N N N N N N N N Hmuoa u macho wNSmmmHm NH NH NH NH NH NH NH HH OH N N N N N N N N H .a.o cmHNmz Hmuos umwnmwmww mmHocmskum v NmscHucoouu.NN NHNNN 2192 mmHocwsvmum I!" “at! 1.. ”It‘ll!!! .1! .F'l. pmscHuc00:1.NN OOON.N NNNN.N NON N N OH .NH NH OH N OH NH N NH NH NH NH OH NH NH N Hmuoe u msouo pmHHouucOU:NHoN NNNN.N NNNN.N NON O N N N N N NH NH NH OH HH OH NH N NH NN ON NH Hmuoe u msoum meHmcommom NNNN.N OOON.NH NON NH NH ON NN NH NH OH NH N OH HH N N N N N N NN Hmuoe u msouw NNHHON ,NNNN.N HNNN.NH NON NN NN NN NH NH NH N N N N N N N N N N N N Hmuoe u macho ucmHUmbo NNNN.N NNHN.N NON N N N N N N N N NH NH NH HH N N NH NH NH NH Hmuoe u msouo mcH>0H NNNH.N NHNH.NH NON N HH NH NH NH OH NH HH NH OH N NH N NH N N N N Hmuoe u Qsouu HMUHMOH NNNN.N ONNN.N OON N NH HH HH NH NH NH HH N NH NH N OH N HH HH NH N Hmuoe u msouo HmauomHHmucH HNNN.N OOON.N NON OH N N N N N NH N HH N NH OH NH N NH NH NH NN Hmuoe u macho ucmnsmmwvcH NNON.N NNHN.NH OON NH N NH NH NH N N NH NH NH N NH NH HH N N N N Hmuoe u maouu w>HuMCNmmEH NHHN.N ONNN.N NON N N N N H N N N N NH N N NH ON NN ON HH NN Hmuoe u dsoum ummco: NNNN.N HNNN.N ONN N N N HH N NH NH NH NH HH OH N HN HN N N OH N Hmuoe u dsouu HstHo: HNHH.N NNNN.N OON N N NH NH NH NH NH N N NH N NH NH N NH NH NH N Hmuoe u aaouo mcH>Hmuom NH NH NH NH NH NH NH HH OH N N N N N N N N H u . . anmHum> o o :mewz Hauoa ucmncmmmo mqmda. 193 TABLE 29.--Entire sample--routine oneway--medians and ranks for terminal and instrumental values (N = 200). Terminal Values Instrumental Values Total Total Med. Rank Med. Rank A Comfortable Ambitious 10.67 15 Life 12.23 13 Broadminded 5.83 2 An Exciting Life 9.97 12 Capable 9.14 8 A Sense of Ac- complishment 7.90 8 Cheerful 10.34 13 A World at Peace 7.97 9 Clean 13.98 17 A World of Courageous 9.59 10 Beautx 12.94 14 Forgiving 8.94 7 Egualitx 8.79 11 Helpful 9.41 9 Family Security 8.50 10 Honest 4.55 1 Freedom 5.69 1 - Imaginative 10.42 14 Happiness 6.15 3 Independent 7.30 4 Inner Harmonyp 7.12 6 Intellectual 9.63 11 Mature Love 5.88 2 Logical 10.13 12 National Secu- ritx 15.69 17 Loving 7.32 5 Pleasure 13.07 15 Obedient 16.24 18 Salvation 16.90 18 Polite 13.50 16 Self-Respect 7.40 7 Responsible 6.03 3 Social Recog- Self-Controlled 8.67 6 nition 14.03 16 Concordance True Friendship 6.92 4 Coefficients 0.19 Wisdom 7.11 5 Concordance Coefficients 0.24 4 9 1... mmHUcmsvmum NNNN.N NNNN.N OOH O N H N N N N N N HH N N N N NH N N NH uoHchoocoqusouo NNNN.N ONNN.N OOH O O N H N N N N N N N N NH N N N N NH uoHHmcouanouu mmmchmm: NNHN.N NNNN.N OOH N H N N N N N N N N N N NH N NH N NH N uoHchoocozumsouu ONNN.N ONNN.N OOH O N H N N N N N N N N N NH N N NH N HH uoHHucouumoouw Eooomum HNHN.N ONNN.N OOH O N N N N HH N N OH N N OH N N N N N N uoHHmcoocozumsouo NNHN.N NOOH.N OOH N H N N N N N N N N OH N HH N N N N H uomecouumsouu NUHusomm NHHEmm NNHN.N ONNN.N OOH N N N N N N N N N NH N N N N N N N O uoHHmcoocozumsouu NHNN.N ONNN.N OOH N N OH N NH N N N N N N N N N N N N H uoHHmcouumsouo NuHHmsmm NNNN.N NNNN.NH OOH NH N NH N N N N N N N N N N N O N N H uoHchoocoznmoouU _OOON.N OOON.NH OOH N OH NH OH N N HH N N N N N H N N N H O uUHchooumzouD wusmmm mo oHuoz < ”NNNH.N NNNN.N OOH O N N N N N N N N N N N N N N N N NH uoHHmcoocozumsouo NNNN.N NOOH.N OOH H N N N N N N N N N NH N N N N N N N uoHchouumsouo momma um oHuoz 4 NNNN.N ONNN.N OOH N N N N N N N N N N N NH N N N N N N uoHHucoocozumsouu NNNN.N ONNN.N OOH O N N N N N N N N N N N HH NH N N N N uoHchouumoouo unmeanHmEoo nod mo mmcwm 4 HNNN.N HHHN.N OOH N OH N N N N N N N N OH N N N N N N N uoHchoocozumsouu ONNN.N OOON.NH OOH N N N N N HH N N N N N N N N H N N N uoHHmcouumsouu meH NCHuHoxm um NNHN.N ONNN.NH OOH N HH N NH N N N N N N H O N N N N N N uoHchoocozuasouo NNNN.N OOON.HH OOH N N N NH N N N N N N N N N N N N N N uoHchouuQsouo MNHH mHnmquNEou m . .o cad m muo NH NH NH NH NH NH NH HH OH N N N N N N N N H mHanum> O .o S H a ucmocmmmo 83 u E .mmooum uoHchooco: ocm uoHHmcoo new mosHm> HmucmEOuumCH pom HmcHEumu MOM mcoHumH>mU oHHuumov one .mcmHooE .mcoHuanuumHo Nocmswwuwuu>m3mco mcHudoulumHQEMm muHucmul.ON mqmdh 195 HHHN.N OOON.N OOH O N N N N N N N N N N. NH N N N N NH N uoHHNcoocozumsouo NNNN.N NNNN.N OOH N N N O N N N N N HH N N N OH N N OH N uoHHmcooumsouo EopmHz NNHN.N NNNO.N OOH H H O N N O N HH N N N N N N NH N N H uoHHmcoocozumoouu NNNH.N OOON.N OOH O O H N N N N N N N N NH N N HH N HH N uoHHmcooumsouo mHnmncmHum mane NNNN.N NNNH.NH OOH NH N NH N HH NH N N N N H N H N H N H O uoHHmcoocozumsouo NNNN.N ONNN.NH OOH OH OH HH NH N N N N N N N N H H N H N H uoHHmcooumsouo coHuHcmmumm HmHoom NNHH.N OOON.N OOH O O N N N H N N N N NH N N N OH HH N HH uoHHucoocozumsouo ONNN.N NNNH.N OOH H H N N H OH N N HH N N N N N N N N N uoHHmcouumaouw vommmmmlemm HNNN.N NNNN.NH OOH HN N N N N N N N N N H H N O N N O HH uoHchoocoqusouo ONNN.N NNNN.NH OOH NN NH N N N N N O N H N N H H N N O N uoHHmcouumsouo coHum>Hmm NNNN.N NNNN.NH OOH N HH NH NH N N N HH N N N N N N H H N H uoHHmcoocozumoouu NNNN.N OOON.NH OOH N NH N N HH N OH N N N N N N N N H N N uoHHmcouumsouu musmmmHm NNHN.N OOON.NH OOH NH NH NH OH N N N N N H N H N H H O H O uoHHmcoocozumsouu NNHH.N NNNN.NH OOH NH ON NH N N N N N N N H N O H O O H H uUMHmcoonosouo NuHusomm HmcoHumz HNNN.N OOON.N OOH O N N H N N N N N N N N N NH N N HH N uoHHmcoocozumoouu NNNH.N NNHN.N OOH H N N N N N N N N OH N N N NH N NH HH N uoHHmcouumsouw m>oa wusumz NOOH.N NNNH.N OOH N N N N N N N N OH N N N N N N OH NH N uoHchoocozumsouo NNON.N NNNH.N OOH O N N N N N N HH N N N N N N N N N OH uoHHmcooumsouU mcoeum: umccH NH NH NH NH NH NH NH HH OH N N N N N N N N H mHnmaum> .Q.O GMHCTZ HMUOB , ucmflcwmwo mmHocmsvmum omDCHucoo||.ON mqmda 196 NNNN.N OOON.N OOH N H N O O N N N N N N H N OH N NH N NN uoHHucoocozumsouo NNNN.N NNNN.N OOH H H H N H H N N H N N N N OH NH N N NN uoHHmcooumsouo umwcom NHNN.N OOON.N OOH O N N N N N N N N N N N OH HH N N N N uoHHucoocozumsouu HNNN.N NNNN.N OOH N N N N N OH OH N N N N N HH NH N N N N uoHHucounmsouo HsumHom HNNN.N HHHN.N OOH O N N N N N N N N N H N N N N N N N uoHHmcoucozumoouo NNNN.N NNNN.N OOH N N N N N N N N N N N N N N N N N N uOHchooumnouu mcH>Hmuom NNNH.N OOON.N OOH N H N N N N N N .N N N N N N N N N N uoHchoocoznmzouo NNNH.N NNNN.N OOH O N N N N N N N N N N N N N N N N N uoHHucoonmsouu msommmusou NNNO.N HHHN.NH OOH OH NH N N N NH N N N N N N N N H H H O uoHchoocoznmnouu NNNN.N ONNN.NH OOH ON N NH N NH N N N N N N N O N N N H H uOHHmcoonmaoum cmoHu NOOH.N OOON.OH OOH N N N N OH N N NH N N N N N N N H N N uoHHucoocozumsouU NNNN. OOON.OH OOH H OH N N N N HH N OH N N HH N N N N N N uoHHmcouumsouu Hsmumwnu OOON.N NONN.N OOH N H N N N N N OH HH N N N N N N N N N uoHHmcoocozumsouu OOON.N NNNN.N OOH N O N N N N N OH HH N OH N N N N N N N uoHHmcoou souo mHnmmmo NNNN.N NONN.N OOH H H H H N N N N N N NH HH N N N NH N N uoHchoocozumsouo NNNN.N NONN.N OOH N N H N O H N N N N N N N NH HH OH N HH uoHHucooumsouo pmNCHEUmoum NNNO.N OOON.OH OOH N N N N N N N N N N N HH N N O N H N uoHHmcoocozumsouu ONNN.N OONN.HH OOH N NH N N N OH N N N N N N N N N N N N uoHchounmsouu msoHanE¢ NH NH NH NH NH NH NH HH OH N N N N N N N N H . o . o :33: H38. umwmmwmww mmHocmswmum vmscHucouul.ON mqmda 197 NNNN.N ONNN.N OOH N N N N N N N N N N N N N N N N N H uoHHmcoocozumdouo NNNN.N HNNN.N OOH N N N N N N N N N N N OH N N N OH N N uUMchouumsouo omHHouuCOUIMHmm ONNN.N OOON.N OOH O N H N N N N N NH N N N N N NH N NH N uoHchoocozumoouo NNNN.N OOON.N OOH H H H H N N N N N N N N OH N N NH NH N uoHHmcooumoouo mHnHmcommom NHNH.N OOON.NH OOH HH N NH N N OH N N N N N N H N N H O N uoHHmcoocozumoouo NHNN.N OOON.NH OOH N HH NH NH N N N N O N N N N H N N N O uoHHmcooumoouw muHHom NNNN.N NONN.NH OOH ON NN N N N N N N N N O O H N H O H H uoHchoocozumoouu ONNN.N NONN.NH OOH NN NH NH N N N N N N N N N H H N N H H HUHHucoonmsouu ucwHomno ONNN.N OOON.N OOH N N N N H N N N N N OH N N N N N HH N uoHHucoocoznmoouw NONN.N OOON.N OOH N H N H N N N N N N N N N N OH OH N HH HOHHucoonmzouo MGH>OH NNNN.N NOOH.N OOH N N N N N N N N N N N N N NH N N N N uoHHmcoocozumsouo NNNN.N ONNN.OH OOH N N N N N N N N NH N N N N N N N H N uoHchoonmoouo I HMUHNOA OOON.N .NNNN.N OOH N N N N N N N N N N N N N N N N N N uoHchoocozumoouo NNNN.N ONNN.N OOH N N N N N N N N N N N N N N N N N N uoHHmcouumoouw HmsuomHHmucH OOON.N NNNN.N OOH N N N N N N N N N H N N NH N N N N NH uoHHmcoocozumaouo NNHN.N NNHN.N OOH N N N N N N N N N N N N H N N N N NH uoHHMcooumoouw uswoammmncH NONN.N NNNN.NH OOH N N N N N H N N N N N N N N N N N O uoHHmcoocozumsouu NNNH.N OOON.OH OOH N N N N OH N H HH N N N N HH N N N N N uoHchouumsouu w>HuwchmaH .NH NH NH NH NH NH NH HH OH N N N N N N N N H . a . a 5:6: H38. umwnmwmwm mmHocosvmum p vmchMCOUII.om mqmdfi 198 NN.O NN.O mucmHonmmou mocmpuoosoo N OO.N N NN.N ammmmk N NN.N N NN.N MHnmvcmmum mnmm NH NH.NH NH NN.NH aoHuHcmoommHmmoom m NN.N m NH.N uommmmmumHmm NH NN.NH NH, NN.NH coHum>Hmm NH NN.NH NH NN.NH madmmmfim NH OO.NH NH NN.NH NuHuaomNNHmcoHumz N NN.N H HN.N m>oq muons: N NH.N N NH.N Ncoaumm HoccH N NN.N N NN.N mmwchmmm H NN.N N NN.N a Emommum HH NN.N N NH.N uHusomm HHEmm OH NN.N HH NN.N NuHHmsmm NH NN.NH NH NN.NH mwsmmm mo NHuoz N N NN.N N NH.N mommm um UHHoz m N NN.N OH NN.N ucwESmHHmEooom mo omcmm d NH HN.N NH ON.OH oqu NcHuHoxm :N NH NN.NH NH NN.HH mMHA mHnmuH0mEoo d AOOHuzO AOOHuzO xcmm QMHpmz xcmm GMHUmz mmaHm> uOHHmcoosoz uoHHmcoo .HOOH u zO mmsoum uONHmcooaoc can HOHHM Icoo How mxcmu Nam msmemErlmmSHm> HmcHEHmuIINm3wco mcHuaoulumHmEMm mHHucmnl.HN mqmda 199 NH.O NH.O mucmHonmmou mocmpuoocoo OH NN.N N NN.N NmHHonuGOUIMHmm N NN.N N NN.N mHnHmcommmm NH NN.NH NH NN.NH wuHHom NH NN.NH NH NN.NH ucmemno N NN.N N NN.N maH>OH N NH.N NH, NN.OH HMOHNOH NH NN.N N NN.N HmsuomHHmfiwm N NN.N N HN.N ucmpcmmmcsH NH NN.OH NH ON.OH m>HumcHNmEH H NN.N H NN.N ummcom N NN.N OH NN.N HsmmHmm N HN.N N NN.N GH>H Mmm N NN.N HH NN.N msommmusoo NH HN.NH NH NN.NH cmmHU NH ON.OH NH O0.0H Homummcu HH NN.N N NN.N memmmm N NN.N N NN.N NmNGHeomoum NH ON.OH NH NN.HH msoHanE< HOOHuzO HNOHuzO xcmm quomz xcmm GMHomz mmDHm> UUHHmcoosoz HUHHmcoo .AOOH u zO mmsoum uoHHmcoococ can uOHHmcoo How mxcmn 6cm mGMHmeIImmsHm> HmucmesuumsHI>m3mco msHusounlemfiwm mHHuchI.NN wands 200 individuals. However, recalling the Wilcoxon technique and Figure 4, it was found that three individual values from the two scales significantly distinguished between conflict and nonconflict individuals, the values being National Security (2 = 1.91, level of significance being .10), Self- Respect (z = 2.19, level of significance being .05), and Logical (z = 1.86, level of significance being .10). This Wilcoxon finding is surprising given the small discrepan- cies reported between medians for these three values for the two groups. Finally, observe the equality of the con- cordance coefficients for the conflict and the nonconflict individuals for both the terminal and instrumental value scales. Surprisingly, the groups are equally homogeneous. To conclude, in general, the facts reveal that the conflict and nonconflict individuals come from identical populations. For thirty-three of the thirty-six values, individuals in the conflict and nonconflict groups do not significantly rank terminal and instrumental values differ- ently. The idea of individuals in the conflict group hav- ing different "value systems" than individuals in the non- conflict group must accordingly be rejected (as evidenced by the overall and similar median and ranking data). The idea of individuals in the conflict group having different "individual values" than individuals in the nonconflict group must similarly be rejected--at least for thirty- three of the thirty-six values, the remaining three 201 significant values oddly enough not having very divergent median and rank data between the two groups. Entire Sample--Routine Stable Table 33 reports the agreement coefficients (Spear- man) for terminal and instrumental values for conflict group pairs; Table 34 reports the same data for the non— conflict group pairs. Table 35 reorganizes the information from Tables 33 and 34 and presents it in the form.of fre- quency distribution, median, mean, and standard deviation data. Following the similar type of analysis in the origi- nal and cross-validation samples, notice the number of negative correlations, again disregarding the level of significance, for the two groups in the entire sample. The conflict group has ten negative terminal correlations, and fourteen negative instrumental correlations for a total of twenty-four. The nonconflict group has three negative terminal correlations and twelve negative instrumental correlations for a total of fifteen. Referring to Figure 6 and realizing that N is still 18, Figure 9 can be derived. Notice again that significant terminal correlations outnumber significant instrumental correlations by almost a 2:1 ratio. Significant nonconflict correlations (ninety- four in total) again far outnumber significant conflict correlations (fifty-six in total). 202 NHNN.N NNNH.N Om NNNN.O NNNN.O NN NNNN.O NNNN.N NN NNNO.N NNHN.N NN NNNN.O NNNN.O av NONN.OI NNNO.N NN NHNN.N! NHNN.NI Nd NNNH.N HNNN.N NN NNNN.N NONN.O NN NNNO.N HNNN.N HN NNNN.O NNNN.O NN NNNO.N NNNN.O ON NNON.O NNNN.O NN HNNH.N: NNNO.N NH NNHH.N NNNN.O NN NONN.O HNNN.N NH NNNH.N: NNNN.O NN NNNN.O HHNH.O NH NNNN.O NNNN.O HN NNON.O HNNN.N NH NNNN.OI OHNO.NI ON NNNN.O NNNH.N: NH ONNN.N NNNH.N NN HHNN.N NNNO.N: NH ONNN.N: NNHN.N NN HNNH.N NNNN.O| NH NNON.OI NNNN.O NN NNNO.N NONN.O NH NNNH.NI NNNN.O NN NNNN.O NHNN.N HH NNHN.N NONN.O NN NNNN.N NNNO.N: OH NNNN.O NNNN.O NN NNNO.N: NHNN.N: N NNNH.N NNNN.O NN NNNN.O ONNN.N N NNNO.N HHNN.N NN NNNN.O NNON.O N NNNO.N NHHN.O HN NNHN.N NNHN.N N NNNN.O NNNO.N ON NNNN.O NNNN.O N ONNN.N NNNN.O NN NHHN.O: NONN.O: N NONN.O: HNNH.N NN NNNH.N NNNH.N: N NNNO.N: NNNO.N NN NNHN.N: ONNN.N N NNNO.N: NNNO.N: NN NNNH.N NNNN.O H mwuHm> mmsHm> Hmnfisz mmsHm> mmon> Hmnadz HmucmfisuumcH HmcHEHmB HHmm HmucmEduumcH HmcHEHmB uHmm moouo uoHHmcoo .AON u zv mHHmm msoum uOHHmaoo How mucmHonmmoo mSHm> Hmucofiduumcw can HmcHEumu ”HHMQ comm HON mucmHonmmoo unmsmmnmsllenmum wcHusonlumHmEMm mHHucmuu.NN mqmde 203 NNNN.NI NNNO.N ON NNNO.N NNNN.O NN NNNN.O ONNN.N NN NNNN.O NONN.O NN OHNO.OI NNNN.O NN NNNO.N NNON.O NN NNNH.N NNON.O NN HNNN.NI ONNN.N NN NNNO.NI ONNN.N NN NNHN.N NNNN.O HN NNNN.O NNHN.N mN NONN.O NHNN.N ON NNNN.OI NNNH.N NN NNNN.O NNNN.O NH HNNN.N NNNN.O NN NNHH.N NNNN.N NH NNNH.N NNmm.O NN NONN.O NNHN.N NH ONNN.N: NNNN.N HN NNNN.O NNNN.N NH NNNN.O NHNN.N ON NNNH.N NNNN.O NH NNNO.N NNNH.N NN ONNN.N: NNNN.O NH OOON.HI OOON.H NN NNNN.O NNNN.O NH NNNO.N NNNN.O NN NNNN.O NNNH.N: NH NNNN.OI ONNN.N NN NNNN.O NNNN.O HH NNNN.O NNNN.O NN NNHN.N NHHN.O OH NNNH.N: NNNO.N! NN NNNN.O NNHN.N N HNNN.N HNHN.N NN NNNN.O NONN.O N NNNN.O NHNN.N NN NHHN.OI NNNO.N N NNNN.O NNNO.N Hm NNNN.O HNNN.N N NNNO.N NNNN.O ON NNNN.O NNNN.OI N NNNN.O NHNN.N NN NNNN.O NNNO.N N HNNN.N HNNN.N NN NNNO.N! NNNN.O N NNNN.N NNNN.O NN NNNN.O NHNN.N N NNNO.N NNNN.O NN NHNN.N HNNN.N H mmsHm> mmsHm> umnEdz mmsHm> mmsHm> umnfidz HmusmfiduumcH HmcHEHwB HHmm HmusmadnumcH HmcHEHmB HHmm moouo uOHHmcoocoz .23 u E mHHmm moonm “OHHmcoococ How musmHOHmmooo msHm> HmucwfisupmcH can HmcHEHmu “HHMQ sumo How musmHonmwoo ucmEmmHNMIlmHnmum mcHusoullmHmEMm oHHusMII.NN wands 2()4 HmN.O OO~.O NN~.O ON O O H N m O O n N O m H N H H mmsHm> HmucmE :suumcH Nom.O NON.O ONN.O ON H H m O OH m n O N N H H O H O mmsHm> chHEume msouu uoHHucoocoz Nmm.o NOH.O NNH.O Om O O H N O m m m w m N N N H H m@212, Hmucme IsuumcH Hw~.O OON.O Nnm.O Om O O H N N O O O N N N m N O O mwsHm> HmcHEume msouo uoHchoo cam? U) cmflnm: Hmuoe m.o m.o ”.0 w.o m.o v.0 m.o ~.o H.o 0.0 H.o- «.0- m.o- N.u- m.o- mHm>uwucH mucwHonwmoo .Aom u zv museum uoHHmcoococ Ucm uoHchoo may cuon CH muHmm new mmaHm> HmucmfiduumcH new HmcHEuwu uo ucmemuom mo coHumH>mn vumvcmum van ~came .cchwE .coHuanuumHv aucmswmuwlumHnmum mnHusoullmHmEMm muHucmul.mN mHmHNN NNNH.N NNNN.N ON O O O N O N N N H N H H N N musmmmHm NNNN.N OOON.N ON O H H O O O H H N H N N N N HuHusooN HmcoHumz NHNO.N NNNN.N ON O O H O O H N N N N N O N N m>oH ensue: NNNN.N OOON.N ON O O O O O H O N N N N N N N Hcosumm umccH OOOO.N NNNH.N ON O O O O O O H N N N N N OH N mmmmmmmmm HNNN.N OOON.N ON O O O O H H H H N N H N N N soommum NNNN.N OOOO.N ON O O O O H H O N N N N N N N HuHusomN.HHHsmN NNNN.N OONN.N ON O H N N N H H N N N N N N N ammmmmmm ONNN.N OOON.N ON O O O H H H H H N N H N N N Nmsmmm No nHuo: N ONNN.N OOOH.N ON O O N O N N N N O N N N N H momma um NHuoz N NNON.N NNNN.N ON O O O O H H N N N N N N N N unmenNNHNEoo no¢ mo mmcmm < NNNN.N NNNH.N ON O H H H H N H O O N H N N N NNHH mcHuHoxm c< NONN.N OONN.N ON O O H N H N H N N N H N N N NNHH mHnmuuoNeou 4 NH NH NH NH NH NH HH OH O N N N H O .0.0 cmHomz Hmuoa msHm> mocmumwwHo mo uCDOE¢ .HON u 2O mHHmm mooum uoHHucoo How mmsHm> HmcHEumu How mmocwumwpr mo mcoHumH>mp mHHuHmsv ocm .mcmHomE .mmHocwsquwllmHnmum mcHusounlmHmem mHHucmnu.ON mqmde 208 NNHN.H OOOO.N ON O O O O O O O O O O N N N OH ammmmm NNNN.N OOON.N ON O O O O O O O N N H N N N N chmccmHHN mane ONHN.N NNNO.N ON O O H O O H H N O N N N N N mmmmwm smoomm HNHoom NHNO.N NNHN.N ON O O O H H H H H N N N H N NH nommmmm-NHmN NNNN.N NNNH.N ON H O H O N H N H H O N H N N coHum>HNN NNHN.N NNNN.N ON O H O H O O O N H O N N N NH musmmmHm NNHN.H NNHN.N ON O O O O H H H H O H N H N NH NNHusomN HNcoHuNz ONHH.N OOOH.N ON O O O O O N O O N H N N N N m>oH ousumz NNHN.N OOON.N ON O H O O H H H H N N N N N O NcoEHNN umccH OOON.N NNNN.N ON O O O O O N H O N N N N N N mmmmmmmmm NNNN.H OOON.N ON O O O O O H N N H N N N OH N eonmmum OONN.N OOON.N ON O O O O O O O N N N N N OH N .HNHusomN.HHHsmN NNNN.N NNNN.N ON O O O O O N H N N O N N N N Nmmmmmmm NNNO.N OOON.N ON O O H O N O N O N N N N N N HmmNmm No NHuoz N ONNN.N NNNH.N ON O O O H O N H N N N N N N N momma um NHuoz N NONN.N OOON.N ON O H O O H O N N N N N N N N NameanHmaoo noc mo mmcmm d ONNN.N OOOH.N ON O O O O O N O H N N N N N N NNHH mcHuHoxm cN NOHO.N OONN.N ON O H H O H N N H H N N N N N NNHH mHnmuuoNEou 4 NH NH NH NH NH NH HH OH N N N N N H .O.o cNHNmz Hmuoe mocmHmMMHo mo HQSOEN .HON n zv mHHmm msoum uoHchoococ MOM mode> HmcHEHmu MOM mmocwumwwHo mo mcoHumH>mp oHHuHmsv paw .mcmHomE .mmHocmsvmuwlumemum mcHusouuanmEmm mHHuCMII.NN mqmde 209 .mocmHmMMHp umoe on ummmH Eoum Umxcmu musmmme nomm .1 NH NNNH.O- NH NNNN.N .mmmmmm HH NHNO.OI HH OOON.N mHanQMHHm mane N NNNN.O N NNNN.N GOHuHcmoomm HMHUON NH NNNH.O: NH OOOO.N HommmmmumHmm H NNNN.O H OOOH.N coHum>Hmm NH ONN0.0- N NNNN.N _mmmmmmmm NH NNNH.O: N OOON.N NwHusomm HmcoHumz N NNN0.0 N NNNN.N m>oH wasps: N ONHH.O NH OOON.N defiumm HmccH N NNNN.O N NNNH.N mmmcwmmum N NNNH.O N OOON.N Eovomnm N HNNO.O N OOOO.N NuHusomN.wHHsNN NH NNN0.0- NH OONN.N Nmmmmmmm N NNNH.O NH OOON.N Nwsmmm mo NHuoz 4 OH NNN0.0| NH OOOH.N mommm um oHuoz 4 NH NNNN.O| NH NNNN.N HamsanHmsoooN mo mmcmm N NH NNNH.O- N NNNH.N NNHH mcHHHoxm :4 N ONO0.0| OH OONN.N NNHH mHanuomsoo N xcmm .Hhou xcmm GMHUmZ wsHm> «.HON n zO mHHmm muonm HOHHmcoo How mmsHm> HmcHEHmu Ham mxcmu pcm.:oHHMHwHHoo N mumEEoou OH H mumsfioon .mocmummmHU QNHomEIImHQMUm mchdoullmHQEMm mHHuchI.NN wands 210 .mocmumNMHp umOE ou ummmH Eoum wmxcmu musmmmfi zoom .1 H NNON.O N OOOO.N _mmmmmm NH NNN0.0 NH OOON.N .mwnmcsmHum mane NH NNN0.0- NH NNNO.N coHuHcmoomm HNHoom NH NNN0.0 N NNHN.N HommmumanmN N NNNN.O H NNNH.N quum>Hmm NH NNN0.0 N NNNN.N mmmmmmmm N NNNN.O N NNHN.N NMHuaomm HmsoHumz N NNNN.O N OOOH.N m>oH musumz N NNON.O N OOON.N Naoeumm umacH N NNHN.O HH NNNO.N mchHmmum NH NNN0.0 N OOON.N _mmmmmmm N OHHN.O NH OOON.N NwHusomN NHHENN N NNNN.O NH NNNN.N .Nmmmmmmm NH NNN0.0 NH OOON.N Nuammm No NHuoz 4 OH NONH.O NH NNNH.N mommm um NHuoz N NH NNN0.0 N OOON.N ucmsanHmeooUN No mmcmm a N NNNN.O NH OOOH.N wMHH mcHuHoxm ca HH NHNH.O OH OONN.N NNHH mHnmuuomsoo N xcmm .HHOU xcmm GMHUwz msHm> «.HON u zv mHHmm mooum HUHHmcoocos How mmsHm> HmcHEHmu How mxcmu osm.soHumHmuHoo N mumEEoou on H wumfifioou .moamummep GMHUmEIImHnmum mcHudouuanmEmm mHHucmul.mN mqmda 211 the .05 level, Self-Respect is significant at the .10 level, and Wisdom is significant at the .01 level. Again, the rankings from the median and correlation data differ substantially. Correlation data reveals ten negative terminal correlations for the conflict group and only one negative terminal correlation for the nonconflict group (again, disregarding the levels of significance). Realizying that N is now 50 and by reference to Figure 6, it is observed that for the conflict group, the positive correlations for Happiness, Salvation, and Social Recogé nition are all significant at the .05 level. For the non- conflict group, positive significant correlations exist for An Exciting Life (.05), Equality (.10), Mature Love (.01), National Security (.10), Salvation (.01), and Wisdom (.01). Tables 40, 41, 42, and 43 report identical infor- mation as mentioned above except now the analysis concerns instrumental rather than terminal values. Four instru- mental values have discrepancies of around 2.00 for the reported median difference figures between the two groups. These values are Ambitious, Forgiving, Intellectual, and Self-Controlled. When the Wilcoxon technique was applied to these four individual values, the following respective z scores were obtained: 1.80, 2.19, 0.20, and 0.82. By reference to either Figure 4 or Appendix D-2, the signifi- cance of the z scores can be determined. Intellectual and 212 mnmm.~ ooom.m mmom.m mmmm.m mvah.m thH.m vmoo.v mmmm.~ mmah.m smoo.m mmvm.m ooom.~ ~vmm.m Nvmm.m mamh.~ voao.~ mhmH.N mmmv.m oooo.v mmmm.v omvm.m hmmh.~ ooom.v ammm.m mmmh.w hmoa.m oomn.m omhm.m mvam.m mmmm.m mmmm.v mm~m.v oomh.v mmvm.m oomh.m ooom.w om om om om om om om om om om om om om om om om om om cmaaouucooumamm wanfimCQNWmm mmmmmm mmmmmmmm mmmmmm mmmmmmm HMSfiUwHH MUCH ucmpcmmwvcH m>fiumcmmmEH mmmmmm mmmmmmm mmmwmmmmm mmmmmmmmmw cacao Hauummno manmmmo consaspmoum msofluflnea .o.o caucus Hopes mocwummwflo wo undead moam> zv muflmd msouw nowamcoo uOm mosam> HmucmEduumcH HOW mmucmummmwp mo mcofiumfl>on mafluumsv cam .mcmwpme .mwflocmsvmumnnmanmum mcwusoullmHmEMm mufiucmuu.O¢ wands .213 oooo.v mmvm.m mom0.m mmmm.~ omhm.m hamh.m monu.m mm0>.~ ooo~.m mmhv.m omhm.m nmmv.~ Nmmo.m mmvm.a mamm.m ommH.m momm.m mmom.m ooom.m ooom.m mvam.m oom>.N hwwa.m mmmm.v ooom.v ooo>.m mmmm.v ooom.m ommm.m Humm.m ooom.m HHHo.m oomm.v oooo.¢ ooom.m Humm.v om om om om om om 0m 0m om om om om om om om om om om UmHHOMUCOUnMHmm manfimcommmm muwaom mmmmmmmm mmmmmm .wmmwmmm HmsuomaamucH ucmpcmmoccH m>wumcwmmEH ummcom mmmmmmm mmmmmmmmm Mflmmmmmmmm sumac stummno manmmmo bmpcfiEQMOHm msofiuanea .o.o cmflcmz Hmuoa m m m mocmuwwwflo mo uCSOE< msam> 2v madam dsoum uowaucooco: new mmsHm> Hmucmfiduumca MOM mmocmummmao mo mcofiuma>m© wafluumsv cam .mcmflpme .mmaocmsvmumnumanmum mowusoulanmEdm wuwucmul.av wands 214 o ammo.o m oooo.¢ cmaaouucou-mamm m vmmo.o m mmmm.q manflmcommom m mmmfl.o N mmvm.m muaaom «a mmaa.ou H hmwn.m mmmmmmmm m qamo.o a ooom.¢ .mmmwmm ma memo.o- 4H Hamm.m mmmmmmm a maao.o- ma -~n.m HmsuomaamucH v vvao.o NH homa.m ucmccmmmocH ma huma.ou ma oomn.m mwmmmmmmmmm a maso.o m ompm.m mmmmmmm ma vmo~.o- ma meam.m mmmmmmm ma ~muo.o- ma mmmm.m maw>flmwom Ha mmmo.o- Ha ommm.v msommmusoo 5H homa.o- h ooom.q .mmmmw ca mmmo.o- oa oom~.¢ mmmmmmmw H vvma.o m mmvm.m _mmmmmmm m anoo.o- a oomh.m nmccflevmoum ma mmma.o- 5H ooom.m maowuflnam xcmm .HHOU xamm cmwwmz madm> «.Aom u zv muwmm msoum goflamcoo Mom mmsHm> Hmucwesuumsfl MOM mxcmu pam.cowumamuuoo m mumaaoon on H mumEEoon .moqmummmav sodomEIannmum mcHuSOHIImHmEMm mHAuCMII.N¢ wands .GUCGHOMMHU “mos Cu. “mmmfl EOHM ”@MGMH mHSmMGE SUMW .4. 215 mH moo~.ou mH ooom.m cmHHouucooumHmm OH vmqo.o n ooon.m mHnHmcommmm «H ovmo.ou m qum.m mmmmmm NH mmmo.ou H oomn.~ mmmmmmmm mH mmmH.o- mH "SSH.m mmmmmm m mmmH.o HH mmmm.e Hmmmmmm HH nHmo.o| MH ooom.¢ HmsuomHHmucH a ammo.o pH ooo>.m unmocmmmcaH SH vva.ou qH mmmm.¢ m>HumchnsH m shao.o m ooom.m mmmmmm m Humo.o m mmmm.m Hmmmmmm H -mm.o v Humm.m mcH>HmHom mH memo.ou mH ooom.m muommuusoo v mNHH.o m HHHm.m mmmmw m mmoH.o OH oom~.v Hmmmmmmw m mSOH.o m oooo.¢ mmmmmmw mH hmho.o1 m ooom.~ Umcheomoum m mahH.o NH Hamm.v msoHanea xcmm .Huou xamm GMHcmz msHm> «.Aom u zv mHHmm msoum uoHHmcooco: Mom mmsHm> HmucmfiduumcH Mom mxcmu vcm.coHuMHmHHoo m wumEEoou on H mumfifioom .wosmumMMHo cmwmeIImHnmum mcHusouulmHmEMm mHHucmuu.mv mqm<9 216 Self-Controlled are insignificant, Ambitious is signifi- cant at the .10 level, and Forgiving is significant at the .05 level. Once again, the rankings from the median and corre- lation data differ substantially. Correlation data reveal eleven negative instrumental correlations for the conflict group and eight negative instrumental correlations for the nonconflict group (again, disregarding the levels of sig- nificance). Referring again to Figure 6, it is observed that for the conflict group, none of the instrumental values have significant correlations in either a positive or nega- tive direction. For the nonconflict group, only the indi- vidual value of Forgiving is significant being positively correlated and at the .10 level. To summarize routine stable as applied to the en- tire sample, Spearman rho information revealed that instru- mental values conceived as an entity or system do not sig- nificantly distinguish conflict group pairs from noncon- flict group pairs. However, terminal values conceived as an entity or system do significantly distinguish conflict group pairs from nonconflict group pairs. Concerning indi- vidual values rather than value systems, the Pearson corre- lation statistic revealed that for terminal values, the conflict group pairs had a positive significant corre- lation for Happiness (.05), a positive significant corre- lation for Salvation (.05), and a positive significant correlation for Social Recognition (.05). For terminal 217 values, the nonconflict group pairs had a positive signifi- cant correlation for An Exciting Life (.05), a positive significant correlation for Eguality (.10), a positive significant correlation for Mature Love (.01), a positive significant correlation for National Securipy (.10), a positive significant correlation for Salvation (.01), and a positive significant correlation for Wisdom (.01). For instrumental values, the conflict group pairs had no significant correlations (negative or positive) at any of the three levels of significance used in this study. For instrumental values, the nonconflict group pairs had only a positive significant correlation for Forgiving (.10). However and in addition, the Wilcoxon test of statistical significance between two pepulations (conflict and non- conflict groups) revealed that roommate pairs in the two groups only had statistically different rankings on three terminal values, namely, A Sense of Accomplishment (.05), Self-Respect (.10), and Wisdom (.01). The Wilcoxon test revealed only two statistically different roommate pair rankings among the individual instrumental values for the two groups, namely, the the values Ambitious (.10), and ,Forgiving (.05). CHAPTER V SUMMARY, CONCLUSIONS, AND IMPLICATIONS Introduction The main purposes of this chapter are twofold: first, to briefly summarize the procedures used in and the results obtained from this study; and, secondly, to comment on the limitations, contributions, and implications of this report. The final section of this report suggests the necessity of additional future research. Procedure The main purpose of this study was to determine whether or not interpersonal conflict could be attributed to differing individual values and value systems. To in- vestigate this idea, various initial research proposals gave way to a study of conflict and nonconflict male- college-roommate situations. Some twelve hundred students were screen-interviewed to obtain a final sample of 200 students (100 student pairs). Each student (and each pair) in the final sample also filled out and returned Form B of the Rokeach Value Survey, met various "entry conditions," 218 219 and accordingly, qualified for either the "conflict" or "nonconflict" group. The final 200 students formed 100 pairs, 50 pairs of conflict students and 50 pairs or nonconflict students. The fifty pairs in each group were equally sub-divided into an original sample and a cross-validation sample. Existing computerized statistical operations de- signed specifically for use with the Rokeach Value Survey instrument were applied to the original sample in the form of "routine oneway" and "routine stable." Using the results from these routines, a subjective "prediction process" was theorized and indirectly applied to first the old original sample and later to the new cross-validation sample. After the results of the "prediction process" were known, routine stable techniques were also run on the cross-validation sample in order to explain the results of the prediction process. To test certain inconsistent relationships and to add generality to the findings, later routines oneway and stable were also applied to the entire sample. Results It is necessary now to View and review the major findings of this research project. The goal of this section of this chapter is to place the research findings in their proper perspective so that broad generalizations can be made and significant results can be emphasized. The re- sults of this study and the "findings" from the previous 220 chapter are summarized in Table 44. Table 44 summarizes the results of routine oneway and routine stable as applied to the original sample, the cross-validation sample, and the entire sample. Table 44 also classifies routine stable findings as they pertain to either the analysis of "value systems" or "individual values." The results of the Spear- man, Pearson, and Wilcoxon statistical techniques are re- ported. Whether or not the Pearson correlation coefficients are positive or negative for the conflict and nonconflict group pairs is noted, as is the level of significance for all individual terminal and instrumental values reported. Because the discussion which follows is intended to draw broad, general conclusions, only direct reference is made by the author to the "summary of findings" for the entire sample. The reader is encouraged to investigate for himself the "summary of findings" for the original sample and the cross-validation sample (as indicated in Table 44). To place the proper emphasis where it should be, the discussion which follows also will deal largely with only the findings from routine stable, not with the find- ings from routine oneway (although both are summarized in Table 44). To date, routine oneway has almost been given equal time (and space) with routine stable. This has not been because the information provided by routine oneway is equally as important as the information from routine stable. It has already been shown that routine oneway provides a very limited amount of useful information which 2221 H H Hmo.V mcH>Hmuom~ _ # 10H.V msoHane<_ 10H.V cH>H uom+ mcoz i 10H.V HmonOH_ HoucmsonumcH _ HousmanuumcH _ oz I HmucmssuumcH Hmuc065uumcH V _ V AHo.~ EoomHz+ _ qua AHo. V coHum>Hmm+ V w W .25. V huHusoww HmcoHumoi _ .Nn. socmH3_ .Ho. V m>oH musumz+ .mo.V coHuHcmoomm HmHoom+ a a 10H.V How mmm LHH m. loH.V Hmmmmmmm+ Hmo.V coHum>Hmm+ _ imo. V Hume _ _ imo.V poommmmumHom -anHmEoooH mo omcmm 4. .mo.V muHH mcHuHoxm c<+ Hmo.V mmmchmm=+ _ ACH.V auHusumm anoHumz HmcHEHmB . HmcHEume _HHo.V wow u HmcHEume HMCHEMoB . H Ame.“ HammHm=+ .oH.V anHouucoo-MHm u _ imo.V :H>H uom+ Hmo.V,Hsmummn . _ a iHo. V mmmwmmmmm Hmo.V :mmHo+ 10H.V nwucHsnmoum- _ m HmuCOEHHHquH . HmucwghumCH _ 02 I HmufimfidhumGH M“ HHo.V aqumHm+ _ a.» i HHo.V quHmHHmw+ . “.4 . 30. V o>oq musumz+ .NH HOH. V NcOEum: umccH+ V a W . HoHuV HHHHmnmm+ HOH.V mmmmmmmnummm- V a Hmo.V soomH3. imo.V woman an cHuox 4+ HOH. V mmmcH _ m. V Ame. V ucmsanHmf _ u imo.V uommmmmumHmmi imo.V mHHH mcHuHoxm c<+ -soooa mo mmcmm a- _ umHHmma uoz HMGHEHQP _ HMCflEHQB _ OZ l HMCHEHOH >m3w¢0 Qfifluflom “ _ . Hmo.V cmHHouucoouuHom+ ioH.V mcH>Hmuom V imo.V Hsuumoso+ imo.V Hsmuommo+ " .mc.V Huonoq o mcoz .mo.V ummco=+ HOH.V omocHevmoum+ q.1 HmucmesuumcH _ HousmanuumcH _ 02 n HoucwesuumcH HmucwEDHumcH Mk... V Amo; EOmumH I — NM . . .IIIII. . i a e :p V 583: :0 V 2833+ 3o V 83??? I imo.V mWHH mHnmuuoHEoo H.AOH.V HwHuaomm HmcoHumz+ HoH.V ounmmmHau . HMCHEumB HmcHEumB _Amo.V wow I HmcHEHmB HmcHEume r V _ uoHHHcoocoz uoHHucoo _ _ _ mcoHuMHmuuou coxooHHz mcoHumHmuuou camummm _ cmEummmm coxooHHz V _ zwemumam msHm>c .mmsHm> HmsuH>HccHg _ imHuzuH>HocHV >m3wco mcHusom AmHHmmV mHnmum mcHusom H .mmchsHm no >HMEEsmul.vv wands 222 is relevant to the analysis of the stated objectives of this study (this is why no routine oneway was applied to the cross-validation sample). Routine oneway analyses' individuals in the conflict and nonconflict groups; all of the tables presented in Chapter IV reporting routine one- way findings involve actual frequencies of value rankings. But the major focus and concern of this study are roommate pairs and data reporting the amount of difference in value rankings between roommate l and roommate 2 of any and all pairs (conflict and nonconflict). Thus, in order to place the emphasis where it properly belongs, the discussion which follows disregards the findings from routine oneway and instead reports only the results from routine stable. Again the reader is encouraged to investigate for himself the "summary of findings" for routine oneway as applied to the original sample and the entire sample (as indicated in Table 44). Thus the discussion has been limited to the entire sample and to only the findings from routine stable. The general results and findings will be reviewed by recalling the stated objectives of this research project: The first objective of this study was: 1. To determine whether conflict (or its absence) between college roommates can be attributed to differing "value systems" by: a. Identifying interpersonal conflict and non- conflict situations between college roommates. 223 b. Studying, comparing, and contrasting the "value systems" of college roommates in conflict and nonconflict situations. "Value systems" were analyzed basically by routine stable in the form of pair agreement correlation coef- ficients (Spearman) between roommate l and roommate 2 of each pair in each group for both the terminal and instru- mental value scales (considered separately and as entities). In the entire sample (as summarized in Table 44), it was found that conflict and nonconflict roommate pairs col- lectively and significantly had diverse and different rank- ings for the terminal value scale but not for the instru- mental value scale. The terminal value scale (conceived as an entity) distinguished between conflict and noncon- flict pairs at the .01 level of significance. The null hy- pothesis in question is whether the conflict and the non- conflict pairs came from the same or identical population. This hypothesis can be rejected at the .01 level of signifi- cance. Thus, the author has fulfilled his main objective and supported his main contention (hypothesis) . . . con- flict and nonconflict roommate pairs do have significantly different "value systems" (terminal but not instrumental). The second objective of this study was: 2. To determine whether or not some "individual values" are more important than are others as possible determinants of conflict and non— conflict situations. 224 This was basically shown by routine stable in the form of Pearson correlation coefficients for both conflict and nonconflict pairs for both terminal and instrumental values. In the entire sample for conflict pairs, three significant individual terminal values were determined (see Table 44). For nonconflict pairs, six significant indi- vidual terminal values and one significant individual in- strumental value were discovered. In addition, the Wilcoxon technique revealed that significant differences for the ranking of the individual values between conflict and non- conflict group pairs occurred in five out of a possible thirty-six instances (again, see Table 44). Thus, some individual values were ranked differently by conflict and nonconflict pairs (however, these same individual values were not always consistent between and among subsamples as evidenced if original, cross-validation, and entire sample results from Table 44 are compared). Again, the null hy- pothesis which assumes that conflict and nonconflict pairs come from the same population was significantly rejected (at least for and in regards to the five individual values surviving the Wilcoxon test). Objective number two, simi- lar to objective number one, appears to have been fulfilled and the associated contention seems to have been supported. Objective number two could be more affirmatively fulfilled and supported if the word "determinants" were changed to "characteristics." Although much of the reported routine -stable data is "significant," it is not necessarily 225 "determinant." Remember, the failure of the prediction process as noted in the preceding chapter. Individual values may be characteristic of and significantly corre- lated within some samples--but not necessarily all sub— samples (again, compare the summary of findings for the original, cross~validation, and entire samples as indi- cated in Table 44). Across many samples (and subsamples), some individual values are consistently noteworthy (for examples, the individual values of Wisdom, Forgiving, et al, in this study). Thus, in general, the data reveal that some "individual values" are more important than are others as "characteristics" (but not necessarily "determinants“ or "predictors") of conflict and nonconflict situations. As explained in the beginning of this report, the design of this study was to provide information which could be used to fulfill the two main "objectives" of this study. Two other minor objectives were, however, initially stated at the onset. The third objective of this study was: 3. Possibly, to express and argue for the impor- tance of "individual values" and "value systems" as the primary determinants of individual be- havior. Having viewed the results of this research project, it would be unwise for the author to argue strongly for the above contention. The misuse of the term "determinants" has already been noted. This research also indicates that 226 the use of the term "primary" probably is not warranted. In the waning moments of this research project, the author now envisions a parallel between value theory and the personality-traits concept. For example, although, in general, successful executives may have certain person- ality traits in common (e.g., the need for achievement), each successful executive has a unique set of traits any one of which will not assure nor prevent his successful- ness. Similarly, although, in general, conflict and non- conflict roommates may have certain values in common, each roommate has a unique set of values any one of which will not assure nor prevent his compatibility. The fourth objective of this studywas: 4. To examine the interrelatedness of individual values, value systems, individual behavior, and interpersonal conflict. Assuming and narrowing our definition of individual behavior to be only conflict and nonconflict roommate situ- ations, this objective has been achieved throughout the entire report which contains over forty tables examining these relationships. The exact nature of the relationships, however, has not yet been (and probably never can be) pre- cisely determined. Limitations of the Study This study did establish some significant relation— ships among interpersonal conflict (or its absence), 227 individual values, and value systems. However, these relationships were not always consistent among subsamples (original, cross-validation, and entire samples). To the extent that subsample results are not always consistent, the significance and implications of these research find- ings are somewhat accordingly limited. Conceivably, certain factors may have influenced the results of this study. For examples, the researcher cannot recall precisely how the original and the cross- validation sample pairs were determined. The process was not randomized, however, and it is believed that the majority of the original sample are from roommates who quickly filled out and returned their Surveys as compared with the cross-validation sample, which undoubtedly con- tains more pairs who had to be coaxed into filling out and returning their Surveys. It is sometimes also believed that the atmospheres of and the personalities of students who live within certain dormitory complexes vary con- siderably. If this is true, again the research may be biased since more students from one dormitory complex com- prise one sample or the other. An almost infinite number of other such variables may have also influenced the re- search results to some degree or another. The author pre- dicts that all such variables individually and collectively have not significantly affected or altered the findings of this research study. 228 Perhaps the limitations of the study are best summed up by recognizing that this study did not attempt to explain conflict via a systems or situational approach. Any attempt to explain (and especially to predict) behavior (such as conflict) or any other phenomenon in terms of only one variable (e.g., individual values or value systems), is, of course, an oversimplification. Usually, a systems or situational approach which incorporates the idea of a number of complex, dynamic, viable, and interrelated vari- ables would be a more realistic method by which to explain conflict behavior. However, some variables in a systems or situational approach are relatively more important than are other variables; and it was this author's initial con- tention (which was later supported) that individual values and value systems could be considered to be important situ- . ational factors affecting interpersonal conflict behavior. Contributions of the Study In general this study supported the contention that conflict can be explained in terms of differing indi- vidual values and/or value systems. It was observed that in some cases (the original sample) information on values not only provided insight into but in addition provided a prediction technique for determining conflict situations. In other situations (samples), often different values seemed to be consequences, characteristics, and/or deter- minants of behavior. Due to an idea generated from this 229 study, hopefully, later research might be able to identify crucial values as explanations, characteristics, or per- haps even determinants of other different types of situ- ations. Conceivably, as Rokeach believes, additional in- sight into all types of human behavior can be obtained via the analysis of individual values and value systems. In the specific case of conflict behavior, if crucial values can significantly and consistently be identified in vary- ing conflict situations, mankind can learn to avoid, con- trol, manipulate and/or eliminate conflict situations. After and if inter-individual behavior is better explained via individual values and value systems, insight might possibly also be forthcoming in the areas of indi- vidual-group, inter-group, group-organization, individual- organization, or inter-organization behavior. As an example of how value information might be useful on, for example, an individual-organization level, it is conceiv- able to imagine how an instrument such as the Rokeach Value Survey might be used as a selection and/or placement technique. One might hypothesize that every organization has its unique climate or value profile, and an organi- zation theoretically could determine how any one individual might "fit" into its structure by comparing the individual's value profile with the company's profile. Other possible uses of value information at the individual, group, and organizational levels can also be conjectured. In terms of the specific phenomenon of conflict, a vivid imagination 230 could envision explaining or maybe even controlling any- thing and everything from family squabbles to international wars via the proper use of value information. Unfortunately, researchers are noted for over- exaggeration of the contributions, importance, and appli- cations of their research findings. Realistically, this study may not lead to universal and everlasting peace both at home and abroad; however, it does provide a pedagogical approach to and hopefully a better understanding of con- flict, values, and interpersonal behavior. Implications and Future Research Many of the implications and possibilities for future research associated with this study have already been discussed in the previous section of this chapter. It is appropriate to now briefly expand upon the tOpic of possible and desirable future research. The possibility of using value information in studying family, group, organi- zational, national, and international situations has pre- viously been conjectured. In order to properly understand the implications of this study, much additional future research is needed in the areas of values, conflict, and behavior, and much of this research is (will be) only indirectly or remotely associated with this research project. Questions such as the following need to be asked and answered: "Do values determine behavior?"; "What determines values?"; "Do 231 1 "How and why do values change?"; "Can values change?" behavior determine values (rather than vice versa)?"; "What is the relationship among values, value systems, attitudes, and beliefs?"; "What really is an 'individual value' or 'value system'?"; "How many values does one possess?"; "Can values be measured?"; "What is the best technique to measure values?";2 "What is conflict?"; "What determines conflict?"; "Can conflict be measured?"; "What is the best technique to measure conflict?"; "Can conflict be measured and explained via values?";3 "Should, and can, conflict be controlled?"; "What actually constitutes lConceivably, this may have been a factor in the somewhat inconsistent original versus cross-validation sample results. It should be remembered that originally routine stable was designed to measure changes in values from time 1 to time 2 rather than from roommate l to room- mate 2. 2It was noted in routine stable how different tech- niques for measuring the agreement on each value produced two diverse and inconsistent value rankings (the two mea- sures being median difference, and the Pearson product- moment correlation coefficient). Undoubtedly, more reli- able and consistent results could have been obtained if better measuring devices were available. Perhaps part of the problem also is inherent in the nonparametric nature of many of the statistical tools used. 3A better question may be whether conflict can be "significantly" and "consistently" measured and explained via values. It was noted that for the various samples high correlation coefficients were obtained in but often inconsistent among the three samples. This question could be answered more thoroughly if more stringently controlled human samples could be designed (remembering the possible bias in this study due to nonrandomized samples and other conceivable factors). 232 behavior?"; "What motivates behavior?"; "What determines individual, group, and organizational behavior?"; "Can behavior be controlled?"; "Why does human behavior change?"; "What is the best technique to measure behavior?"; . . . these and related questions need additional research before the contributions and implications of the current study can be properly evaluated. BIBLIOGRAPHY BIBLIOGRAPHY Books Argyris, Chris. Personality and Organization. New York: Harper and Brothers, 1957. . Understanding Organization Behavior. Homewood, Ill.: Dorsey Press, 1960. . Integrating the Individual and the Organization. New York: Wiley and Sons, 1964. Atkinson, John W. An Introduction to Motivation. Prince- ton, N.J.: D. Van Nostrand, 1964. Cofer, C. N., and Appley, M. H. Motivation: Theory and Research. New York: Wiley and Sons, 1964. Fraser, Donald. Nonparametric Method§_in Statistics. New York: Wiley and Sons, 1957. Hilliard, A. L. The Forms of Value. New York: Columbia University Press, 1950. Homans, George C. The Human Group. New YOrk: Harcourt, Brace and Company, 1950. Kaplan, A. The Conduct of Inquiry. San Francisco: Chandler, 1964. Kelly, George A. A Theory of Personality. New York: W. W. Norton, 1963. Kendall, M. G. Rank Correlation Methods. London: Griffin Publishing, 1948. Kraft, Charles. A Nonparametric Introduction to Statistics. New York: Macmillian, 1968. 233 234 Madsen, K. B. Theories of Motivation. Copenhagen: Munksgaard Press, 1959. Maslow, Abraham. Motivation and Personality. New York: Harper and Brothers, 1954. Mendenhall, William. Introduction to Statistics. Belmont, Ca1if.: Wadsworth, 1963. McFarland, Dalton E. Conflict and Cooperation in Personnel Administration. New York: American Foundation for Management Research, American Management Association, 1962. McNemar, Q. Psychological Statistics. New York: Wiley and Sons, 1962. Miller, George; Galanter, Eugene; and Priebram, Karl H. Plans and the Structure of Behavior. New York: Holt, 1960. Newcomb, T. M. The Acquaintance Process. New York: Holt, Rinehart and Winston, 1961. Osgood, C. E.; Suci, G. J.; and Tannenbaum, P. H. The Measurement of Meaning. Urbana, Ill.: University of Illinois Press, 1957. Rokeach, Milton. The Open and Closed Mind. New York: Basic Books, 1960. . Beliefs, Attitudes and Values. San Francisco: Jossey-Bass, 1968. Schramm, Wilbur. The Science of Human Communication. New York: Basic Books, 1963. Siegel, Sidney. Nonparametric Statistics for the Be- havioral Sciences. New York: McGraw-Hill, 1956. Tate, Merle W. Nonparametric and Short Cut Statistics in the Social, Biological and Medical Sciences. Danville, Ill.: Interstate Printers and Publishers, 1957. Vroom, Victor. Work and Motivation. New York: Wiley and Sons, 1964. Walsh, John Edward. Nonparametric Statistics.‘ Princeton, N.J.: D. Van Nostrand, 1962. 235 Whyte, William Foote. Man and Organization. New York: Wiley and Sons, 1964. Articles in Books Allport, Gordon. "Attitudes." A Handbook of Social Psy- chology. Edited by Carl Murchison. Worchester, Mass.: Clark University Press, 1935. Kluckhohn, Clyde K. M. "Values and Value Orientations in the Theory of Action." Toward a General Theory of Action. Edited by T. Parsons and E. A. Shils. Cambridge: Harvard University Press, 1951. Madsen, K. B. "Theories of Motivation: An Overview and Synthesis." Human Motivation: A Symposium. Edited by Marshall R. Jones. Lincoln: University of Nebraska Press, 1965. Smith, M. Brewster. "Personal Values in the Study of Lives." The Study of Lives. Edited by R. K. White. New York: Atherton Press, 1963. Williams, Robin M. "Values." International Encyclopedia of Social Sciences. New York: Macmillian, 1967. Periodicals Allport, Gordon, and Odbent, K. S. "Trait-Names: A Psycho-Lexical Study." Psychological Monographs, Vol. 47 (1936), Whole number 211. Anderson, N. H. "Likeableness Ratings of 555 Personality- Trait Words." Journal of Personality and Social Psychology, IX (1968), 272-79. Friedman, M. "A Comparison of Alternative Tests of Sig- nificance for the Problem of M Rankings." Annals of Mathematical Statistics, XI (1940), 86-92. Hotelling, H., and Pabst, Margaret R. "Rank Correlation and Tests of Significance Involving No Assumption of Normality." Annals of Mathematical Statistics, VII (1936), 29-43. Lovejoy, A. 0. "Terminal and Adjectual Values." Journal of Philosophy, XLVII (1950), 593-608. 236 Maslow, Abraham. "A Theory of Human Motivation." Psycho- logical Review, L (September, 1943), 370-96; 514-58. McKeon, Richard. "Communication, Truth and Society." Ethics, LXVII (October, 1956), 89-99. Olds, E. G. "The 5% Significance Level for Sums of Squares of Rank Differences and a Correlation." Annals of Mathematical Statistics, XX (1949), 117-18. Penner, L.; Homant, R.; and Rokeach, M. "Comparison of Rank-Order and Paired Comparison Methods for Measur- ing Value Systems." Perceptual and Motor Skills, XXVII (1968), 417-18. Rokeach, Milton. "The Role of Values in Public Opinion Research." Public-Opinionguarterly, XXXII (1968- 69), 547-59. Toussaint, Maynard. "Line-Staff Conflicts--Its Causes and Cure." Personnel, XXXIX (May-June, 1962), 8-20. Wilcoxon, F. "Individual Comparisons by Ranking Methods." Biometrics Bulletin, I (1945), 80-83. Willerman, B. "The Adaptation and Use of Kendall's Coef- ficient of Concordance (W) to Sociometric-Type Ranking." Peychological Bulletin, LII (1955), 132-33. Theses and Dissertations Beech, R. P. "Value Systems, Attitudes, and Interpersonal Attraction." Unpublished Ph.D. dissertation, Michigan State University, 1966. Hollen, Charles C. "The Stability of Values and Value Systems." Unpublished M.A. thesis, Michigan State University, 1967. Homant, R. "The Meaning and Ranking of Values." Unpub- lished M.A. thesis, Michigan State University, 1967. Other Publications and References DATRIX (Direct Access to Reference Information: A Xerox Service), University Microfilms, Ann Arbor, Michigan. 237 Boros, Oscar Krisen. Mental Measurement Yearbook. 6th ed. Highland Park, N.J.: Gryphon Press, 1965. Hollen, Charles C. "Program Description and User's Manual for VALUTEST." Michigan State University, Depart- ment of Psychology, January, 1969. Rokeach, Milton. "A Theory of Organization and Change Within Value and Attitude Systems." Presidential address to the Society for the Psychological Study of Social Issues, The American Psychological Associ- ation, September, 1967. Michigan State University, Department of Psychology, reprint. . "The Measurement of Values and Value Systems." Pre-publication copy. Michigan State University, Department of Psychology, reprint, Winter, 1970. APPENDICES APPENDIX A INTERVIEW-QUESTIONNAIRE APPENDIX A INTERVIEW-QUESTIONNAIRE I. Personal Introduction Noncooperation Study of roommates-approval Cooperation, non- Why roommates do or do not qualified get along Cooperation, qualified, Few minutes and partici- incomplete data pation Follow-up Confidence Cooperation, qualified, complete data NAME DATE ROOM STUDENT NUMBER CLASS STANDING PHONE NUMBER II. 1. Have you had any roommates at MSU with whom you have gotten along very well, that is, with whom you have experienced little or no incompatibility? (If more than one person is identified, obtain data for all of the following questions for the most compatible and/or most recent roommate). 2. What is the name of this former or present roommate? 3. Were you assigned this roommate or did you choose him yourself? 4. How long did you room together? 5. Besides this individual, how many other roommates did you have at this time? 238 10. 11. 12. III. 239 Does this former or present roommate still live in the East Lansing area? If yes, current address is Why, in your estimation, did you two get along so well? What qualities did your roommate possess that you liked best? Did you ever have arguments or disagreements, if so, how frequent and over what? Why did you two separate? Who initiated the separation? Would you say that the primary reason that you two (have or had) roomed together so long is because you two (get or got) along very well together? Have you had any roommates at MSU with whom you did not get along very well, that is, with whom you have experienced a high degree of incompatibility (if more than one person is identified, obtain data for all of the following questions for the most incompatible and/or most recent roommate)? What is the name of this former or present roommate? Were you assigned this roommate or did you choose him yourself? How long did you room together? Besides this individual, how many other roommates did you have at this time? Does this former roommate still live in the East Lansing area? If yes, current address is Why, in your estimation, did you two not get along very well? 240 8. What qualities did your roommate possess that you disliked most? 9. Did you ever have arguments or disagreements, if so, how frequent and over what? 10. Why did you two finally separate? 11. Who initiated the separation? 12. Would you say that the primary reason that you are no longer rooming with him is because you two could not get along very well together? IV. This concludes the preliminary questions that I have to ask of you. In my final report, I plan to study a limited number of cases in depth. These in-depth studies will be selected randomly in the not too distant future. If your case is one of those selected, I will be in contact with you later and will ask you to fill out an additional form. If your case is not selected for further Study, this con- cludes the information which I will ask of you. Thank you for your time, patience, and cooperation. APPENDIX B ROKEACH VALUE SURVEY, FORM E APPENDIX B VALUE SURVEY ' Form E Name Student # Room Below is a list of eighteen values arranged in alphabetical order. Your task is to arrange them in order of their im- portance to YOU, as guiding principles in YOUR life. Study the list carefully, Then place a 1 next to the value which is most important for yep, place a 2 next to the value which is second most important to you, etc. The value which is least important, relative to the others, should be ranked 18. Work slowly and think carefully. If you change your mind, feel free to change your answers. The end result should truly show how you really feel. A Comfortable Life (a prosperous life) An Exciting Life (a stimulating, active life) A Sense of Accomplishment (lasting contribution) A World at Peace (free of war and conflict) A World of Beauty (beauty of nature and the arts) Eqpality (brotherhood, equal opportunity for all) Family Security (taking care of loved ones) Freedom (independence, free choice) Happiness (contentedness) Inner Harmopy (freedom from inner conflict) 241 242 Mature Love (sexual and spiritual intimacy) National Security (protection from attack) Pleasure (an enjoyable, leisurely life) Salvation (saved, eternal life) Self-Respect (self-esteem) Social Recggnition (respect, admiration) True Friendship (close companionship) Wisdom (a mature understanding of life) Below is a list of another eighteen values. Rank these in order of importance in the same way you ranked the first list on the preceding page. Ambitious (hard-working, aspiring) Broadminded (open-minded) Capable (competent, effective) Cheerful (lighthearted, joyful) Clean (neat, tidy) Courageous (standing up for your beliefs) Forgiving (willing to pardon others) Helpful (working for the welfare of others) Honest (sincere, truthful) Imaginative (daring, creative) Independent (self-reliant, self-sufficient) Intellectual (intelligent, reflective) Logical (consistent, rational) Loving (affectionate, tender) Obedient (dutiful, respectful) 243 Polite (courteous, well-mannered) Responsible (dependable, reliable) Self-Controlled (restrained, self-disciplined) C) 1967 by Milton Rokeach APPENDIX C PROGRAM DESCRIPTION AND USER'S MANUAL FOR: VALUTEST APPENDIX C PROGRAM DESCRIPTION AND USER'S MANUAL FOR: VALUTEST A tape overlay system containing a series of routines written expressly for use with data from Rokeach's Value Survey. by Charles C. Hollen Department of Psychology Michigan State University January 17, 1969 Disclaimer Although this program has been tested by its con- tributor, no warranty, express or implied, is made by the contributor or Michigan State University as to the accuracy and functioning of the program and its related materials. 244 245 CONTENTS INTRODUCTION . . . . . . . . . . . Routines in this system . . . . . . Organization of this report. . . . . STATISTICAL TESTS FOR USE WITH ROKEACH'S VALUE SCALES . . . . . . . . . . . . . Group comparisons . . . . . . . . Concordance coefficients--an index of intra- group agreement . . . . . . . . Two-way group comparisons and the interaction question 0 O O O O O O O I 0 Value change. . . . . . . . . . Value stability. . . . . . . . . SUGGESTIONS FOR EFFICIENT USE OF VALUTEST ROUTINES Checking data for errors. . . . . . ROUTINE ONEWAY I O O I O O O O O O Routine description . . . . . . . Sample output . . . . . . . . . ROUTI NE TWOWAY O O O O O O O O O C Routine description . . . . . . . Sample output . . . . . . . . . ROUTINE CHEKSUMS. . . . . . . . . . Routine description . . . . . . . Sample output . . . . . . . . . ROUTINE CHANGE . . . . . . . . . . Routine description . . . . . . . Sample output . . . . . . . . . ROUTINE STABLE . . . . . . . . . . Routine description . . . . . . . Sample output . . . . . . . . . HOW TO USE THE VALUTEST SYSTEM . . . . . Control cards descriptions . . . . . Example sets of control cards for each routine (JUN 11 l3 16 18 19 19 27 26 26 35 34 34 41 4O 4O 45 46 46 50 55 55 65 246 Page 1 INTRODUCTION PROGRAM VALUTEST is a tape overlay system contain- ing a series of routines written expressly for use with data from Rokeach's Value Survey (M. Rokeach, Beliefs, Attitudes, and Values, Jossey-Bass, 1968). All the rou- tines are fully generalized, and may be used with any data from ranking scales, in which respondents are asked to rank k objects from 1 to k. Whenever possible, routines contained in this series have been set up so as to produce final summary tables of the results, in a form most readily understandable to the user. This consumes slightly more computer time, but it frees the user from the time-consuming and often confusing process of constructing tables by copying numbers from com- puter output. It is hoped that the slight increase in com- puter time needed will be more than compensated by the saving in human time and the reduction of human errors in transcription. Most of the statistical tests used in this series of routines are based on formulas taken from S. Siegel, Non-Parametric Statistics, McGraw-Hill, 1956. The remainder are from formulas found in Q. McNemar, Psychological Sta- tistics, Wiley, 1962. Users who wish more detailed pre- sentations of the statistical tests than are provided here may refer to these two books. 247 Page 2 This program contains several subroutines written by John Morris, and used with his permission. The bulk of the programming is the work of the present author, to whom all inquiries concerning the program or this report should be directed. Routines in this eystem The program includes five separate "routines," each one designed to suit a particular purpose: ROUTINE ONEWAY provides comparisons of the median value rankings of several groups, as defined by an inde- pendent variable. It is analogous to a one-way analysis of variance. ROUTINE TWOWAY compares the value rankings of groups simultaneously defined by two independent variables (e.g., white males, black males, white females, black fe- males). It provides a test of the effect of each inde- pendent variable and a test of the significance of their interaction, for each of the dependent variables (values). It is analogous to a two-way analysis of variance. ROUTINE CHEKSUMS provides a simple method of double-checking the data for respondent, coding, and key- punching errors by adding together the sum of each re- spondent's value rankings and printing out his data for error correction if this sum is not correct. ROUTINE CHANGE is designed for use with subjects who have been tested twice with an intervening influence 248 Page 3 attempt or change-producing event. It provides comparisons between groups, as defined by independent variables, in re- gard to the amount of value change from pre-test to post- test in each group. ROUTINE STABLE is designed for use with subjects who have been tested twice but with no intervening in- fluence attempt, e.g., for data in which the experimenter's primary interest is in the degree of value stability from Time 1 to Time 2. It provides results pertaining to both the degree of stability of each subject, and degree of stability of each value. Organization of this report Beginning on the next page is a general intro- duction to the routines in PROGRAM VALUTEST. It is in- tended especially for the user whose statistical background is slight, or who is not sure which type of statistical test is appropriate for his particular research design or research interest. Following the general introduction is a section of suggestions on how to use the program most efficiently. It is hoped that everyone who plans to use the program will peruse this section carefully. Next is a section of routine descriptions, which contains a specific, detailed explanation of each routine 249 Page 4 along with sample output to show what the results from each routine will look like. Then following the routine descriptions is a section on control cards, which describes in detail how to use the program. VALUTEST is a tape overlay system, which means that the user may gain access to it simply by "call- ing" it from a tape which is on file at the MSU Computer Center. Thus, he need not obtain a "program deck" for the routine he wishes to use. For those who may wish to modify this program to suit their own purposes, a complete FORTRAN IV source deck listing is available, along with a diagram of the deck organization. Copies may be obtained through the present author or through Milton Rokeach at Michigan State Uni- versity. STATISTICAL TESTS FOR USE WITH ROKEACH'S VALUE SCALES (OR OTHER RANKED DATA) We have received numerous inquiries concerning the problems of selecting appropriate methods of analysis for the value rankings. PROGRAM VALUTEST offers a fairly com- prehensive series of analysis routines, each one addressed to a particular type of experimental question. The follow- ing is intended as a general introduction to these various routines, and as a guide to the user who is not sure which routine may suit his particular purpose. 250 Page 5 The statistical techniques utilized in PROGRAM VALUTEST are by no means the only ones available for these types of questions. Some users may find other alternatives preferable. However, the techniques described here are those which we have used extensively in our research pro- gram and which we have found through our experience to be the most useful and efficient. Group comparisons The simplest type of eXperimental question which one might ask about values is: What are the differences (and similarities) between the values of two or more groups, such as alcoholics and non-alcoholics, Republicans, Democrats, and Independents, or policemen and criminals? One way of answering this question is to perform tests of the significance of the difference between the groups' value rankings, for each one of the values individually. Thus the question is rephrased: Do policemen rank A Comfortable Life significantly higher (or lower) than criminals? . . . and so on, for each of the values. ROU- TINE ONEWAY provides an answer to this question. As shown in the sample output on pages 29-30, a group median is computed for each of the groups, for each of the values. In the example Sex is the independent (cate— gory) variable and each of the values in turn is a dependent variable. For the 69 males in this sample, the median of 251 Page 6 their rankings of A Comfortable Life is 13.36, while for the 91 females, the group median is 14.06. The rank num- bers shown beside each set of medians are simply an order- ing of the medians from smallest to largest. Thus in this example Freedom is the most important value for both groups, while Salvation is the least important. The medians are derived directly from the grggp frequeney distributions shown on page 28. These frequency distributions show how many of the males (and females) ranked each value number 1, number 2, and so on up to 18. Thus one male and two females ranked A Comfortable Life number 1 (most important) while five males and eight fe- males ranked it 18th, (least important). On page 29, we find printed to the right of each set of group medians a Chi Square value and its probability of occurrence. Note for example that the males and females in this particular sample differ significantly on A World of Beauty. The Chi Square value is 11.426 and its proba- bility of occurrence (under a two-tailed test) is 0.001. We may conclude that A World of Beauty is significantly more important to females (group median 11.64) than to males (group median 14.90). The Chi Square values shown are derived from the Median Test for K Independent Groups (Siegel). Page 27 shows how these are computed. A grand median is computed 252 Page 7 for the entire sample--in the example, the grand median of both groups for A Comfortable Life is 14.00. Each group is then divided into those above and below the grand median, as shown in the table. The Chi Square test performed on this frequency table gives us the test of the significance of the difference between the groups. In ROUTINE ONEWAY the user is offered a choice of significance tests. He may specify either the Median Test or the Kruskal-Wallis H Test, whichever he prefers. The Kruskal-Wallis test, although claimed by its authors to be primarily a test of differences in central tendency, is sensitive to differences in the "shape"of the distri- butions. This may cause it to yield misleading results. Our experience has shown that unless the experimenter has hypotheses concerning differences in group variances, the Median Test is the more dependable of the two, though more conservative. Corcordance coefficients--an index of intragroup agreement The reader may have noticed at the bottom of the sample output on page 30 a "concordance coefficient" listed for each of the groups. Both ROUTINE ONEWAY and ROUTINE TWOWAY (see below) provide this additional sta- tistic. These coefficients provide one method of answer- ing the question of group homogeneity, which is often of interest. We may wish to know to what extent males agree 253 Page among themselves in their ranking of all 18 terminal (or all 18 instrumental) values, and whether they are more or less homogeneous a group than are the females. The concordance coefficient is an index of the degree of correlation between the "value profiles" of all group members. It is mathematically equivalent to the average of the Spearman rank-order (rho) correlations be- tween all possible pairs of group members. In general we have found, as one might expect, that subgroups defined by several category variables, such as middle-class white Protestant females, are more homogeneous than the super- groups from which they were taken. Two-way group comparisons and the interaction question Taking an actual example from one of our earlier studies, we have found through the use of ROUTINE ONEWAY that Protestants rank Salvation significantly "higher" (more important) than Catholics. Before drawing the con- clusion that Protestants as a group are more concerned with Salvation than Catholics, it is important that we discern whether this difference is a genuine one or whether it is spurious, i.e., due to the influence of a "contaminating variable." The difference may be due to religious denomi- nation, but alternatively it may be due to differences in religious importance. Supposing we suspect that many of 254 Page 9 those in our sample who identified themselves as Catholics are only nominal Catholics, not devout church goers. The presence of a large percentage of such individuals could be "depressing" the group median for Catholics. From one-way comparisons on church attendance we already know that the "devout"--those who attend church every week, regardless of denomination--rank Salvation extremely highly compared to those "nominal" religionists who attend church less often. Thus, taking church attendance as a second category variable, we would like to compare the value rankings of devout Catholics, devout Protestants, nominal Catholics, and nominal Protestants. Results from such an analysis would allow us to examine the extent to which the ranking of Salvation (and each of the other values) is influenced by religion and church attendance, and the extent to which these two variables "interact," i.e., the extent to which one may have a differential effect depending on the other. ROUTINE TWOWAY accomplishes this purpose. The final summary tables produced by this routine (see example on page 38) are very similar to those from ROUTINE ONEWAY. In the example given, the two independent variables are Religion and Importance. Thus we find sets of group medians for four groups. (The number of groups need not be limited to four. The routine will handle up to 12 groups). 255 Page 10 Note that on the right hand side of the table on page 38 are listed four Chi Square values and their associ- ated probabilities of occurrence. These are derived from statistical tests based on formulas provided by K. Wilson. The example output on page 35 shows the frequency tables on which these tests are computed. Again, as in the case of the "one-way" median test, a grand median is computed, and the groups' scores are divided into those above and below it. As in the case of the "parametric" two-way analysis of variance, a Chi Square value is com- puted for the "Religion effect," for the "Importance effect," and for the "Interaction effect." In addition, a "Total chi-square" is also computed. These statistics allow us to determine whether Religion significantly influences each of the dependent variables (values) when Importance is "controlled" and vice versa, as well as the extent to which these two variables "interact" in influencing each of the dependent variables. We have found that results from this test are vastly easier to understand and interpret when the number of groups is held to a minimum. We suggest that whenever possible, independent variables should be collapsed into two, or at most three categories. 256 Page 11 Value change Results from numerous studies conducted in our own research program as well as by other experimenters show that the value ranking scales may be used effectively as a pre-test to post-test measure of value change. Typically, a group of subjects are tested before and after some in- fluence attempt or life-change which is expected to affect one or more of their values. The ranking scales then pro- vide a measure of change not only in the "target value" but also in all the other values as systematic "re-adjust- ments" reverberate throughout the entire value system. ROUTINE CHANGE is designed to analyze data from pre-to-post change studies. (ROUTINE STABLE, described below, is also intended for test-retest data, but focusing on a different type of research interest). ROUTINE CHANGE is addressed to two questions: (1) Did a group of subjects change significantly (on each of the values considered indi-‘ vidually) from Time 1 to Time 2, i.e., does the group's rankings at Time 2 differ significantly from its own rank- ings at Time 1? (2) Did a particular group of subjects (Experimental group, Males, etc.) change significantly more than some other group (Control Group, Females, etc.)? For both of these types of analyses, ROUTINE CHANGE uses means and "parametric" significance tests rather than the medians and "non-parametric" tests used in the other routines. This is chiefly because we are now considering 257 Page 12 distributions of change scores rather than simply initial ranking scores, and we feel that these distributions of scores are better suited to the "parametric" tests. For example, they are more apt to be distributed normally. The example output on page 45 shows that ROUTINE CHANGE provides a separate table of results for each one of the values. The form of these results is reasonably self-evident. For each of the groups (the number of groups may range as high as 8) a mean and standard deviation are computed for the pre-test scores, the post-test scores, and the change scores. The significance of the change from pre-test to post-test is ascertained by computing a t-test for corre- lated measures and its associated probability of occurrence (under a two-tailed test). It may be seen in the example that the first group changed significantly (p less than .05) in a negative direction. The group changed 2.1103 units "downward" on A Comfortable Life, while neither of the other two groups changed significantly. Beneath this set of results are shown the results of tests of the differences between the groups' pre-test, post-test, and change means. In the example, none of these differences are significant, although the three groups differ "almost significantly" (p = .0587) in their pre-test rankings of A Comfortable Life. 258 Page 13 Our aim in providing this profusion of significance tests is to include tests of all the possible differences that the experimenter might conceivably be interested in, and let each experimenter use whichever ones he wishes. This routine, like the others in this series, has been "optimized" so that except in cases of extremely large samples, the amount of computer time needed to calculate all these various tests is trivial. Value stability The value ranking scales have been used effectively as a means of measuring value system stability over time. The stability question is similar to the change question in that both involve data obtained by testing the same group of people twice. However, studies of value stability normally differ from studies of value change in that (1) the experimenter's focus of interest is on the extent to which his subjects' responses remained the same from Time 1 to Time 2, rather than on the extent to which they changed, and (2) the time interval between test and re-test is usually longer in studies of stability. The question of value stability may be divided into two separate experimental questions: (1) What is the degree of value stability of each person--who are the people with the most stable values and who are the least stable people? and (2) What is the degree of stability gr 259 Page 14 each value--which values changed the most and which ones the least, in this particular population of people during this particular time interval? ROUTINE STABLE provides results to answer both of these questions. First, it computes and prints an index of value stability for each subject, for each of the rank- ing scales. If both terminal and instrumental value scales were administered, then each subject would have a terminal and an instrumental value stability coefficient. This coefficient is the Spearman rank-order (rho) correlation between his Time 1 and his Time 2 rankings of the set of values. If he ranked them in exactly the same order both times, his stability coefficient will be 1.00-- a "perfect score." To the extent that his Time 1 and Time 2 rankings were different, his stability coefficient will be reduced, and may go as low as -1.00. In practice we have found that almost everyone in our samples had stability coefficients of 0.0 or higher. An example of the output from this routine is shown on pages 50, 51, 52, 53, and 54. The routine first computes and prints each subject's stability coefficients, followed by frequency distributions, medians, means, and standard deviations of these sets of stability scores. In regard to the second question--the stability of each value--ROUTINE STABLE offers three different measures 260 Page 15 of stability (or change) over time. Here again we have found that different experimenters prefer different types of measures, so we included several and let each user choose whichever one (or more) he wishes to use. The first of these is perhaps more a measure of change than of stability. It is simply the mean change from Time 1 to Time 2. It is computed by subtracting the group mean of the Time 2 responses from the group mean at Time 1. On this measure the sign is reversed so that a change in the direction of greater importance (such as from 8.00 to 6.00) is shown as a positive change. The second stability measure shown on page 53 is the median of the absolute changes. It is computed by taking the group median of the changes from Time 1 to Time 2, each change considered regardless of direction. In the case of the first measure, mean change, there may occur nearly equal numbers of "positive" and "negative" changes which cancel each other out, producing a mean change of 0 or very nearly 0. Thus this measure fails to reflect the amount of "shifting around" which may be occurring. In order to ascertain which values are most and least stable, a change either from 14 or 15 or from 14 to 13 is defined as a change of 1 unit. The median of these 261 Page 16 changes for the entire group provides an accurate estimate of the stability of the value itself. Third, a Pearson product-moment correlation between Time 1 ranking scores and Time 2 scores is computed. Each of the three stability measures is followed by a set of rank numbers showing their ordering from most to least stable. In this way one may compare the results of the different methods. Finally, a significance test is calculated--the Median Test for K Related Groups (A. Mood, Introduction to the Theory of Statistics, McGraw-Hill, 1950). This test is conducted to ascertain whether the set of values under consideration differ significantly in their degree of stability. SUGGESTIONS Some suggestions on efficient use of VALUTEST routines If the user is contemplating "multiple runs" on the same sample of people (i.e., comparisons of males ver- sus females, old versus young people, blacks versus whites, married versus single peOple, etc., etc.) it is to his distinct advantage to run all of these analyses (or as many of them as possible) in one single job, rather than running each one as a separate job. This is because a very substantial portion of the total computer time which 262 Page 17 is used in any job is taken up by the process of reading in the data. The CDC 3600 computer is an extremely fast arith- metic machine but is comparatively slow on input-output . operations. Particularly when samples are large, the ratio of "readin" time to "computation" time goes up rapidly. It will take from 60 to 90 seconds to read in the data from 1,000 subjects, while the computations will normally take only 20 to 30 seconds, at the most. This is true of all the routines in this system. The amount of time needed to read in data is almost entirely a function of the number of subjeCts (records), irrespective of the number of variables per record. Thus the user will save substantial amounts of time by taking advantage of the opportunity to run a whole series of analyses in one single job. Each routine in this system keeps track of the time elapsed during the data read-in phase and the compu- tation phase, and prints out these elapsed times (in seconds) on the right-hand side of the output page. The user may wish to note the time used for reading in his data and the time taken up by the actual computations. One caution--the time-saving advantage of multiple analyses in a single job may be negated if the user makes an error on his control cards which causes the program to 263 Page 18 produce "garbage" results. Almost all types of control card errors will simply cause the program to stop immedi- ately and print out an error diagnosis. However, there are a few types of errors which allow the program to com- . plete its run, but with "nonsense" results. Users should at all times exercise extreme care in preparing control cards. Checking data for errors ROUTINE CHEKSUMS offers a quick and easy method of double-checking one's data for errors before running it on the other routines. Nearly all those who have approached us wishing to use PROGRAM VALUTEST have initially evinced a firm conviction that their data decks contained no errors. However, it has invariably been found that errors are present, most of which can be located by CHEKSUMS. Each of the routines includes an "echo check" option. This is an opportunity for the user to specify that his data be printed out as they are read in. This provides a ready means of verifying that the data were read in correctly and that the recoding was done correctly. In addition, if the program should abort due to an illegal code (such as an alphabetic character in an integer field) an "echo check" would pinpoint exactly where in the data deck this error was encountered. Users are urged to take advantage of this Option. 264 Page 19 ROUTINE ONEWAY Routine description This routine is designed to compute the median rankings of each of k objects by two or more groups of individuals and to test the significance of the difference between the groups for each object ranked. In terms of Rokeach's Value Survey, this routine compares the value rankings of groups (male versus female, open-mineded ver- sus closed-minded, etc.) defined by category variables (sex, dogmatism, etc.) As shown in the example output on pages 27, 28, 29, and 30 the results appear in three sections. First, for each category variable and for each dependent variable there is computed and printed an overall frequency table for the Median Test. Second, a detailed breakdown of the frequencies is printed showing how many people in each group ranked each object (value) first, second, third, etc. Finally, a summary table is produced showing the group medians and rank numbers (rankings of the medians from "highest" to "lowest" within each group) for all groups across all dependent variables. On the right-hand side are shown the size of the Median Test statistic (chi- square) and its probability of occurrence under a two— tailed test. 265 Page 20 Routine size limits ROUTINE ONEWAY will handle a data array of up to 84,000 items. Please note, however, the following routine size limits: Within one job, the number of category vari- ables may not exceed 6, the number of dependent variables may not exceed 36, and the number of subjects (respondents) may not exceed 2,000. Users whose data exceed one or more of these limits must divide their analyses into several separate jobs. For example, a user with 10 category vari- ables may run the first six, and then the remaining four, as two separate jobs. The routine may be used on other data than Rokeach's value scales. However, because it was formulated primarily for use with the value ranking scales, which contain either 12 or 18 items per scale, it assumes that the number of dependent variables will be an even multiple of 6. It also assumes that if there are 18 or less dependent variables, that these constitute one ranking scale, while if there are more than 18 dependent variables, these constitute two scales, and will be divided accordingly. User options 1. Significance test option: The user is offered a choice of either of two significance tests, the Kruskal- Wallis H test or the Median Test for k Independent Groups (for detailed descriptions of these tests, see S. Siegel, 266 Page 21 Non-Parametric Statistics). Experience has shown that the Kruskal-Wallis test is somewhat more sensitive to differ- ences in "shape" of distributions, while the Median Test is more predominantly a test of differences in central tendency. The two tests normally have been found to give nearly comparable results (i.e., if a given difference is significant according to one test, it will also be sig- nificant according to the other, and vice versa). Unless the user has a hypothesis concerning differential shapes of distributions, the Median Test is recommended. 2. Echo check option: The user may specify that his data be printed out after being read in and recoded. This uses slightly more computer time, especially in the case of very large samples, but it provides a useful safe- guard to insure that the data are being read in and recoded correctly. 3. Automatic recoding option: Supposing, for example, that the user has punched each respondent's age on his data cards in its "raw" form. He now wishes to combine those who are in their teens, those in their twenties, in their thirties, etc., into groups for com- parison. ROUTINE ONEWAY offers a simple means of accom- plishing this "recoding" by the insertion of certain con- trol cards specifying which variables (if any) are to be recoded, in what way. 267 Page 22 Although the program will handle up to 10 groups per category variable, experience has shown that it is vastly easier to interpret the results if the number of groups is reduced to three or four. Thus, on variables such as age and social class, on which a large or small number of groups may be used, it may be well to utilize the automatic recoding option to combine and reduce the number of groups. In addition, care should be taken to eliminate groups of N size 5 or less (20 or less in large samples) by combining them with other groups or recoding them as 0's (see Missing Data). 4. Entire sample option: The user may specify whether he wishes to obtain frequency distributions and medians for his entire sample. Normally this option would be utilized on only the first run-through on a particular sample of data. 5. Identification variables option: Data decks on which this routine is used will frequently contain more than one card per subject. It is absolutely essential that these decks be "match-merged" correctly, in order that each subject's independent variables are matched with his own dependent variables, not someone else's. On numerous occasions in the past we have found that match-merged data decks contained cards out of order or cards missing. These cause mis-matches which lead to worthless results. I cannot emphasize too strongly the 268 Page 23 frequency and ease with which this kind of "data foul-up" can occur. The routines in this program contain a safeguard against mismatches due to missing or out-of-order cards. This is done by reading in a "subject identification number" from each card and comparing these to be sure that they are the same for each of that subject's cards. For example, suppose we have 100 subjects, with 4 cards per subject. These are match-merged so that each subject's four cards are together, in order from 1 to 4. Each card has a subject I.D. number punched somewhere on the card. (These numbers need not be consecutive; they need not run from 1 to 100; but they should be in the same columns on all cards). Thus we wish to read in four "I.D." variables. The total number of variables (as specified on the main param- eters card) should include these four variables. An identification variables numbers card is then inserted along with the other control cards which specifies which ones, of all the variables read in, are the identification numbers. The format card will, of course, also reflect this inclusion. If the program finds a subject whose identification numbers do not match, it will print out the "wrong numbers" it found, and stop. This feature may be used on any number of cards per subject, up to 40. 269 Page 24 If the user has only one card per subject, or wishes to "short-circuit" this safeguard, he may insert a blank card in place of the I.D. variables numbers card or put in the same variable number twice (e.g., "4141"). 6. Control variables option: This option offers the opportunity to analyze "subgroups" from within a sample without physically separating their data cards from the rest. Suppose, for example, the user has a large volume of data, either on cards or on tape. He wishes to run a comparison between, say, males and females but for Catholics only. Thus he wishes to use Sex as his category variable and Religion as a "selector variable" or control variable, excluding from the analysis all other religious groups ex- cept Catholics. This is possible through the insertion of a few additional control cards, thus obviating the bother- some necessity of presorting data decks by hand or con- structing special indices. The user may select a group for analysis on the basis of more than one control variable at once, and more than one value per variable. Thus, if he wishes to select just the White, Middle-and-Upper Class Catholics, he would use Race, Social Class, and Religion as "control variables," and specify one value to be retained for Race (white), two values for Social Class (middle and upper), and one value for religion (Catholic). 270 Page 25 Note that the number of subjects used in aegiven analysis may not exceed 2,000. However, the routine may, if necessary, read in up to 9,999 subjects in the process of searching for 2,000 or less subjects to be used in the actual analysis. Missing data All 0's or blanks on the data cards will be treated as "missing data." Any respondent with a zero or a blank on his data card for a particular variable will auto- matically be dropped from the analysis of that variable pply and the N size will be reduced by 1. Thus a 0 or a blank should be used to represent a "No Answer" or "Did not Respond" code. If, for example, on the category variable "sex,' males and females are punched on the cards as 0's and 1's, the user must make use of the automatic recoding option to recode these groups as 1's and 2's. Input Data may be input from punched cards or from tapes. If input is from tape, the logical unit number must be between 1 and 49 inclusive, and must not be 20 or 25. (The program uses tape unit 25 as a scratch unit for input of control cards). Data decks (or tapes) may contain more than one card per subject, but each subject's cards must 271 Page 26 be together (match-merged). One card must contain values of variables for one respondent only, and each variable must appear in the same column on every card. Timing For an "average-sized" run of 300 subjects, 36 dependent variables, and 4 category variables, the program used approximately one minute of computing time. The amount of time needed varies chiefly with the number of subjects, so that with 1,400 subjects, 36 dependent vari- ables, and 2 category variables, approximately 3 minutes is used. Sample outputs from ROUTINE ONEWAY follow on pages 27, 28, 29, and 30. ROUTINE TWOWAY Routine description This routine computes a "non-parametric two-way analysis of variance." It is designed for use with ranked data such as that obtained from Rokeach's Value Survey. It allows the user to study the joint effects of two inde- pendent variables on each of a series of dependent vari- ables. A detailed discussion of the purposes and appli- cations of this routine may be found in the general intro- duction section on pages 8, 9, and 10. 2'72 oooHv.H Page 27 u mru4 mozunu_7uVm roouwxu no wwacwc mzn we cm 73m¢ no >< chzdu wnax doom: o°.o a macaw ,HVom: mVJaxouo< kc mmzmm H LgmaVz<> erozmawa xwm mJnHVm<> >«oomVHu 7<~owt az3m< nmuurmjcwxu Vmu» 2<~owt COCOCC-CCOCOCOCOCCOOC0000000000000...cCC0.000.000...0.000000000000000COO000.00..O0.00.000000000000000...Cooctoocoooouoco .Vmur ouVH».o:» .. wuHaom Vzu auh3m< «o H. msczuu wHHI 1:00: O0.0H . wmoum ,HVnm: maVJ quHVuxm 7. VJmHVst vaozmgmo xmm mJuHVxH> >aoamHHu 7.Vour azuJ muzama ac »« mactmu wqct anon: oo.cH . mmoum 7 byworwauo xmm anHVmH> scoompau ,«Vouz aszo xoswm Q74 msama nqurmaamcu _mu. 24.9w: 2:73 Page 28 >»~xaUmm JHzc_»OH maaHHx mV meHVaH) quazuamo cos.» .nHm.. H. ~ H H n n H H H n c H H m H . ~ H H HHwau Haaoan H»... as»... .. m H H H . H H H . . . H H H H H H H HHH: Hanan» - an «Han zHVawx HHHo» oH HH 0H 0H HH HH ~H HH 3H o o H o m w n ~ H mmHuzmaoux. HznzaH: umzz. mV meHVaH. Hzmazuamo H...» .H.H.. H. H H H H H . H HH H H H . H H H N H H WHHHHH .Haoan one.» no.9.H on o H H o ~ H H ~ H c H ~ m H H v H aH wHHz uaaomn - sun «Han zHVHH: HHHoH .H HH .H nH .H HH uH HH .H o o H H n v n m H mmHuzmaou:. mmwzHuaHz mH mHmHVaH> Hzmazuamo n....~ Has... H. H H H H o H H H H H H n . HH H .H . HH mHszu Hanan" .HHH.~ H..H.. .. H H H H H H H H H H H H H H m H H .H HHHx .Hacmm - >Ha HHS. zHVam: HHHo» .H HH .H mH HH HH ~H HH HH 9 H H H n H H H H mmVuzmaoua. o xoomwcu m. meHVcH> quayunwo H....H Hos... H. H H H m H o H a H H H H m m H H H N HH.zwi .nsomn HHH... HH.... .H H H H H. v H H H H . H H H H . H H H .H.x .HsozH - >un «Hag zHVHH: HHHoH .H HH .H .H .H HH HH HH HH o . H H m H H ~ H mmVuzmaoua. >HVaHomm HHVxHu mV mHmHVuHH quayunmo .HHH.H H..... H. H H n w H . H H H H H H HH H .H H o H HHwau .Haoan Haas.» .H.¢.. .. o H H a H H H N H H ~ o H H m w . H mHHz annex” - sun uHao zHHaux HHHo» .H HH .H nH .H HH ~H HH HH . . H H n H H H H mmVuzmaoua. >HVHH3ow mV mHmHHaH, quayuamo .HHH.H .HH..HH H. H H H H OH H HH H H N H . H H H H H H HHwa. unseen .HH..~ HHHH..H .. H HH 9 H . H H H H H H H H H H H H H HHH: .naoan - awn uHao zHHau: HHHoH .H HH .H .H .H HH HH HH .H o o H H n H n N H mmVuzwacuz. >H3Hmm uo oHco: H mV-wHoHVnH> Hzmazuawo onus.» H.H».. H. o H H n H H H . H H H H H H a HH . . mHmeu unaoan .HHH.H .H.... .n H H N u H ~ H m H H ~ m n H . . H . mHH: Haaoan - sun 3H3. zHVmH: HHHoH .H HH .H mH .H HH HH HH HH . a H H m H n ~ H mmVuzmaoux. muHma HH 94cc: H «V mHmHVuH> quazuam. H.Hn.n .HH... H. H H v H o n H ~ . H m o H H H .n H H HHmeu nanoxn HHH... .H.H.. .n H H m H H a H H H H o H H H . . . ~ wHHz unseen - sun ¢ua «Han zHVaH: HHHo» .H HH .H “H .H HH ~H HH .H a o H H . c H ~ H mmVuzmaoua. mum; onHHuxm 2H m. mHoHHaH. quavuamo 2H... 3...: H. . HH HH H . H H H H H H H H . H H H H. 3:3. 3526 2.... 2...: H. H H . o H H . H H H H H. H H H H H H 3:. :88 - >un cHao zHHauz HHHoH .H HH .H .H .H HH ~H HH .H o . H . n H H ~ H mmVuzmaouzi muVH mHmeaouxou H mH anHVcH. Hzmaaaumo Hum HV mHnHV=.> scoamHHu monHHV>Ho wHVHHHHH ozH .HzHVamz .mzoVHDHVHHHV. Haywoaucu 274 Page 29 ~HH.H nHH.H HHH.H mHH.o HHH.H HHH.H HHu.H HHH.H HHH.H um ooHHO woo.o mm~.o HHH.~ omv.HH moo.o 0Hv.H nvo.o moo.o "Nx Hmm» z<~omz H HH.. H Ho.v HH HH.H n Ho.m HH HH.HH o HH.H NH Hm.o HH HH.H «H HH.HH xzm aw: Ho mHHzmH OO.N oH.v HH NH Ho.HH om.o oo.m NH HN.OH «H HH.HH xza am: on mJHz xmm m. HHHHHHHH HmoomHHu . mewHon asoao mmmzunm<1 roommau >h~130mm >J~t4 oqmoz « mggmtoou< no mmzmm a HHHJ HzHHHuxw 2H mHHJ mHmHHmourou H .z adama whzmoapm amt 275 Page 30 mo~.o ago.” oca.o vmm.o moo.o mho.o ago.” omo.o ccw.o coo.“ Hfio.o no”.m Nmm.o Noo.o umm.» owo.o «co.» moo.o NH ma on ma ¢~.o “0.0 mm.~ no.ma no.0 om.ow oO.Vfi mm.v« o~.m mo.m ma ma ma ed on NN.O uo.m ma.m on.vH no.m om.oH o~.na mm.Vd ma.n mn.m .muumoo uuz4¢m ma3mpHmaumm J«zo_»‘z u>04 mm:»zoz¢wo z<~cmr 2<1» mqu mmuuzwficwmu quu vn and hz»m»owa nun cnm u_Jozhcu Jmo z<_owr o» J waoum wuzhmmz— u w4m<~m4> 773490 zo_o~4wm u w4m<_~<> 10¢ muz<_m<> mo m~m>4wo 2c—omr 241» mmwa mw—uzwfiowmu 44mm no nNu bz»m»oau can cow u—qozhcu 4c2—toz hao>ma z<~ow1 o» J mcoum moz»am:. u msm<_¢.> y’aaou zone—4mm . m4m :oa uuz<~ma> no m.m>4¢7< > no m_m>4.2« >«x.wzc azawz moxie zuuxpmx wuzwzwuha no 323:3; no pm“; 0.00000000005000000000050000050000500350500000o.oooocoocoooooaoococoaooococo0.000005505050000005050000500000500000.0000. OOCOOOCOOOOCCCOOOOOCOOO50000000000000.0050000000.00000000000000009500coo0555000050000000500005500000000000000000.0000... so“ mo~a.afl~« 4.»n» oaco.o« oo ~n-.oooa 2.x».x n~n~.° maac.« .muo.vn w aomo.a¢ zumzpwm .5 Au masaou 2.0: .3 nuz.:am .o zam atasom lcza~z<> no m_m>4.z¢ >41.wzo aim: Sufi—.81 25523: muzmxmuta .3 5.9.33.3; he 5: ooooooooccooooocooocooc45.055000555503000595554055005055550050504.50.505.555050555505505405500550050550050555050.0505... 00000.c00000000000000.0550005....0550..-5000005050.000500050000055000505cocoa...0000005000050.050050000050405000.0000... «cu unwc.oaaw J no m_m>4.24 » hzwczwmun up: 5893:. 2 332.; .ZcuEZuzymozwnuoz. m3: $55 3072?. o» 5.3.3:. . BB.» 8.2.8 .5. 291 Page 46 ROUTINE STABLE Routine description This routine is designed for samples of subjects who have been tested twice. It computes measures of the stability of each person and the stability of each vari- able (i.e., each value) from pre-test to post-test. Beginning on page 50 are sample results from a group of 40 MSU students. The first set of results are simply a listing of each person's stability coefficients for each of the two scales (in this case both the terminal and instrumental scales were used). The second portion of the results consists of fre- quency distributions, medians, means, and standard devi- ations of each of the two distributions of stability coef- ficients. Thus, for "Scale 1" (terminal value scale) we find that 4 students' stability coefficients were between 0.50 and 0.60, 6 were between 0.60 and 0.70, etc. The median stability of this group of students was .77 for terminal values and .71 for instrumental values. Also shown is the correlation (Pearson product-moment) between terminal and instrumental value stability in this group. Next are shown results pertaining to the stability of each value. First, a frequency distribution of the absolute changes from Time 1 to Time 2 is shown. In the example results, 6 people did not change at all on A Com- fortable Life, 8 people changed one unit either "up" or 292 Page 47 "down," etc. The median amount of change is shown at the right, for each value. Next is a display of three different measures of stability or change, for each of the values. These are mean change (Time 1 group mean minus Time 2 group mean), median absolute change (from the previous page), and the Pearson correlation between Time 1 and Time 2 scores for each value. Each change measure is ranked from least to most change. Note that the value Salvation changed the least (was most stable) in this group, according to both the median and correlation measures. There is, however, a substantial difference between the rankings of the two measures. An Exciting Life, for example, is 10th most stable of the values according to the median of absolute changes, but 2nd according to the correlation method. It is anticipated that different users will wish to choose whichever measure they feel is the best estimate of sta- bility. Finally, a median test is computed to ascertain whether there is a significant difference in the sta- bility of the values. Referring again to the sample out- put, on page 54, we see that for A Comfortable Life (here labeled "Treatment 1") there were 24 subjects out of our sample of 40 for whom A Comfortable Life changed more than the median of each subject's change scores. For Salvation, this was true of only 7 subjects. The chi-square value is 293 Page 48 31.4500, which is significant at the 0.0176 level with 17 degrees of freedom. Thus we may conclude that within this set of values and for this group of people, some values are significantly more stable than others. ROUTINE STABLE may be used on one ranking scale only, or on two scales. If 18 variables or less are read in, it assumes that these comprise one scale. If 19 or more (up to 36) are read in, it assumes that these comprise two scales, and divides them accordingly. Of course, it does not matter to the program whether the terminal or the instrumental scale comes first. Routine size limits The total number of variables is limited to 92 (36 Time 1 and 36 Time 2 measures). The number of subjects is limited to 1,000. Users whose data exceed these limits must divide their data into two or more separate jobs. User options This routine offers four user options, which must be specified on the main parameters card. They are: 1. Echo check option: The user may specify whether he wishes his data to be printed out after they are read in and recoded. 2. Test option: The user may specify whether he wishes to obtain results pertaining to just the subject's stability, just the stability of each value, or both. 294 Page 49 3. Identification variables option: This option is described on pages 22 and 23. 4. Control variables option: This option is described on page 24. Input Data may be input from punched cards or from tapes. If input is from tape, the logical unit number must be be- tween 1 and 49 inclusive, and must not be 20 or 25. (The program uses tape unit 25 as a scratch unit for input of control cards). Data decks (or tapes) may contain more than one card per subject, but each subject's cards must be together (match-merged). One card must contain values of variables for one respondent only, and each variable must appear in the same column on every card. Timing A test run on 100 subjects with 36 values (both scales) used approximately 25 seconds to compute both the subjects' and values' stabilities. Sample outputs from ROUTINE STABLE follow on pages 50, 51, 52, 53, and 54. 295 Page 50 MSU STUDENTS N = 40 FOLLOWING IS A LISTING OF STABILITY COEFFICIENTS FOR EACH SUBJECT (SPEARMAN RHO CORRELATIONS BETWEEN TIME 1 AND TIME 2 VALUE RANKINGS) SUBJECT NUMBER SCALE 1 SCALE 2 11 0.7007 0.7049 12 0.8369 0.7998 13 0.8225 0.7090 14 0.8328 0.7626 15 0.7977 0.5831 16 0.8204 0.7358 17 0.7812 0.6987 18 0.6182 0.6987 19 0.8617 0.7523 20 0.7193 0.4469 21 0.5150 0.3230 22 0.8225 0.6367 23 0.8122 0.8163 24 0.5501 0.5955 25 0.6636 0.8741 27 0.6739 0.6512 28 0.8204 0.7007 29 0.8700 0.7049 30 0.7977 0.7069 32 0.8844 0.8844 33 0.8885 0.7110 34 0.6925 0.7110 1 0.8658 0.6760 2 0.7234 0.4489 3 0.7255 0.3540 5 0.9257 0.7234 2596 Page 51 o»«.. a...° .«~.o ~.«.. an~.o ~.~.o .a.a zgmx z¢.nuz O 0 an d o 0 ac u o O 4.»o» o. .s.».° m. an on N 0.... C N 0.... a o.o a a o o n.o ~.o >»_4.c¢»m CO... «.9 4.»:mvaupmz. oz. 4¢mhz- m»zm.u.uumou >».J_m¢»m ho zo_»¢~>mn n¢¢az<»m a,« .z«m: .z¢_amx .20.»:auzpa.= wuzuaomcs cc .2 m—zmoahm an: 2537 Page 52 nude." ,onoo.u mane.“ ..ns.u an-o.o anon.“ suou.u ann..« on...“ 9.9;.“ ouv«.u coco.» enun.u o.-.a ooao.« ~on..~ ona«.« cums.” 50“..» ococ.~ oouo.« coma-N cum... cams.“ “one.“ -~s.¢ anus.“ coon.“ n»n..~. keno.“ cons.“ ”gnu.” mane.“ oo...~ on...» coon.~ z¢.om: a. a. o. a. O. a. a. o. o. a. a. a. o. a. a. o. a. o. Jase» a“ o o o o a o c n o N o a n N Na 0 s o o o o a o o o o n n v c u an o N o c o a o a o o a n N N N m n o o o o o a o a o a a u v N N o 0 cu m a o a o o a o o o o o N N c 1 ma c c o o a o a o a u n a n a v a «a s c o o o u o o a o a u a n o a Na 0 o o o o o o o u n u o n n N 0 ha n a o o o o o c o o N N o v a h an a o o o o o o o n o o o v n a n on c a o o o o a o o o n v v N c o s n o o o a o a o o c o N N .N 0 N« on m o o o a o o N o o a u n v o v on o c o a a o n o n o u o u v r m ed a o o o o o u a o o n N u v v o ou . o a o o N o a N N N N N N o n o o o o a o o n o u a o u u c n «a o r c o o o o o a u o N o N n o o o e on an va on Na an ca 0 o s o n v n N u a wez«xu no >2:ox< zoom—1 m.xmazm_as was» zo_»_zuoumc 44.non puuanmc sta zo.»¢,4o4 agape: >zasccz aura. nnmz.cact xooumcs >»_aaoun >4.:¢g >».4¢:om >~a¢uo so name: 4 mocuu p< o4¢oz ¢ n.4Nyouuc No mnzwu 4 m._4 az_»_uxm 2. on.) aga¢paostou 4 m=4«» .20—rumm—o no mmmJaucowm. wuzmc mJuhu scan as >»_4.o¢»m mbawoarm 3m: 2518 Page 53 cu Nu nu ou nu Nu onv cu N uu xz<¢ oNos.o «cac.o unnv.o ouon.o n..o.. ado... 9030.9 coco.° osu~.° souo.o onum.o -~s.° Nann.o snuo.o ones.e nuoo.o onus.o «oco.a .ucou nu ma nu cu ou Nu nu ou ou 22.x soda." occ..~ onuo.u on~«.~ unno.e cans.u Nouv.u -~s.u anon.” coon." n»no.~ soda.“ eons.“ “pan.“ mane.“ .oo¢.~ an...” coon.~ z¢_nmz nu uu ou nu cu cu n ou Nu nu n o xz<¢ .u...°. enuc.o. on~o.o. .aao.o. .o.».o. .aon.o onao.o. 9....9. .oos.o coau.o ommu.o .m~v.° .o.«.a on-.« onun.o .ono.o. .m.~.° .o...° 7‘0: zoom—t n.1maymncu was» zo_»_zooum¢ 4¢.uou kuuanma 54w» zo.»<>4‘a uaamcwqc >..uaoun J¢zo_»¢2 u>o4 ucapzoy¢.x guzz. anmz~na<2 tooumcs >-¢aumn >4.r.s >».4¢:ou >~=¢mo so a4¢ox a mu :u4- so >».4.o¢»m us» so omuan«mx muss» 299 Page 54 osuo.o »»_J~n.aoan an roomy.“ go mmmaamc ocnv.«n ou am an vN a mu ou cu o“ o. “N cm a“ ou a“ Na ou cu z._amx .puwuaaw. so: zexp «mega; muozazu so .0, uwzcaom .10 cu su ou nu cu nu nu uu ou N u prmxpmnmm On» How mouunuumuw Ouuumsdummcoz .uwmmum amcpum "mousom on.mm mm.om mm.bv b>.mv ON.ov mN.mm mm.mm vm.mN um.mN om.mN O0.0N mv.mu um.ou mm.vu om om.mm mm.mv mm.wv mm.Nv ao.mm vu.mm mv.Nm vm.mN mm.vN av.NN hh.mu uh.hu hm.mu w~.vu aN mm.om NN.mv N¢.mv vm.uv Nm.hm mo.vm mm.um «n.5N mo.MN om.uN vm.mu mm.ou mm.vu om.mu mN mv.mm om.ov vu.vv uu.ov vb.mm um.Nm Nm.om vm.wN NN.NN ON.ON uu.mu mu.mu Nu.vu mm.Nu hN mo.vm em.m¢ mm.N¢ NN.NM wm.mm om.um mN.mN vm.mN mh.uN Nm.mu mN.hu mm.mu uv.mu ON.Nu ON NO.Nm um.v¢ hm.uv mw.hm mm.vm om.om bu.mN qm.¢N NN.ON cm.mu hv.wu uw.vu on.Nu Nm.uu mN NH.Hm ma.Nv NN.o¢ Nv.om ON.mm mm.¢N ou.hN vm.NN vm.mu mo.mH om.mu mm.mu mm.uu om.cu VN mh.m¢ vm.u¢ hm.wm bu.mm Ho.Nm n¢.mm No.mN vm.NN NO.NH mu.hu mm.vu mo.mu aN.uu ON.OH MN NN.Nv mN.ov oo.hm Nm.mm HN.OM om.hN vm.¢N vN.uN ou.mu um.wu vo.vu vm.Nu O0.0H vm.m NN om.mv mm.mm vm.wm no.Nm Nm.mN bu.oN mm.m~ vm.oN NH.NH vv.mu vN.mu mm.uu Nm.m om.m HN Nm.mv hm.hm No.mm uv.um u¢.mN vo.mN NN.NN vm.mu hm.wu wm.vu vv.Nu mm.ou «N.m wN.w ON Nm.mv mu.wm mm.mm vu.om 0N.NN om.MN mm.uN vm.mu mm.mu Nh.mu mw.uu NH.OH nm.m no.» mu um.Nv om.vm mm.Nm NN.NN mm.mN mb.NN om.o~ vm.hu cv.vu mm.Nu mm.ou mm.m um.h No.h NH mn.ov uv.mm oo.um mm.hN sh.VN No.uN um.mu vm.wu mm.mu OO.NH mo.ou no.m ON.5 uv.w bu mN.mm OO.NM mm.mN om.wN vm.m~ wv.o~ Nv.mu «m.mu NO.NH mu.uu um.m om.h um.m um.m mu on.hm mm.om oN.wN oo.mN Hm.NN um.au Nm.hu vm.vu NN.HH um.ou mm.m wN.b mm.m MN.m mu Nu.om vu.mN bw.mN mm.mm mo.uN mu.mu NN.mu vm.mu NN.OH hv.a mh.h hm.w hm.m mm.v vu mm.vm mo.hN hv.mN wm.NN um.mu mm.ou Nu.mu vm.Nu mm.m mw.m v0.5 mm.m wh.v uu.¢ nu um.Nm NN.@N mo.vN mo.uN mm.mu Hm.mu Ho.vu vm.uu mo.m HN.N om.w MN.m mu.v hm.m Nu mN.um Nn.vN NO.NN mm.mu mN.hu mc.vu om.Nu em.ou mu.m mm.m mm.m mm.v uo.m mo.m uu mm.mN HN.NN wu.uN um.mu mm.mu v¢.mu mh.uu vm.m NN.N mu.m om.¢ vm.m mo.m mm.N cu NN.NN hm.uN mm.mu Nm.ou mo.vu «N.Nu om.ou vm.m mm.w mm.m bu.v Nm.m mm.N mo.N m NH.ON mo.ON nu.mu um.mu mm.mu mo.uu mm.m «N.N mm.m mm.v ov.m mm.N mo.N mw.u m Nm.vN mv.mu Nm.wu no.vu NO.NH om.m mm.m mm.m hm.v NN.N NN.N bu.N mm.u vN.u h wv.NN um.wu mo.mu mm.Nu vo.ou mm.m NN.N mm.m mm.m ho.m ON.N vw.u mu.u hm. m Nm.0N ao.mu :mm.mu ho.uu v~.m mN.h 00.0 mm.v co.m vm.N um.u «H.u mh. mm. m m¢.mu mN.mu no.uu mv.m mn.h mm.m mm.v wm.m ON.N mo.u wo.u us. no. on. e hN.wu vm.uu vo.a NN.N mN.w vw.¢ ww.m NN.N Nv.u oo.u mm. mm. mu. Nu. m Nm.mu HN.m Nw.n mm.m oo.v NN.m uv.N mm.u uh. mv. uN. cu. vo. No. N mm.ou ov.m uv.m vm.m HN.N vm.u bo.u mv. mu. coo. muo. mmoo. mwooo. wuooo. u uoo. uo. No. no. cu. ON. on. cm. as. om. om. mm. mm. am. up mumsumnuno M mx was» o: 666:: suuaunmnoum .mumsvmuuno mo monum> umouuuuoll.mlo mamas TABLE D-4.—-Critical values Of t. 324 Level of Significance for One-Tailed Test .10 .05 .025 .01 .005 .0005 df Level Of Significance for Two-Tailed Test .20 .10 .05 .02 .01 .001 1 3.078 6.314 12.706 .31.821 63.657 636.619 2 1.886 2.920 4.303 6.965 9.925 31.598 3 1.638 2.353 3.182 4.541 5.841 12.941 4 1.533 2.132 2.776 3.747 4.604 8.610 5 1.476 2.015 2.571' 3.365 4.032 6.859 6 1.440 1.943 2.447 3.143 3.707 5.959 7 1.415 1.895 2.365 2.998 3.499 5.405 8 1.397 1.860 2.306 2.896 3.355 5.041 9 1.383 1.833 2.262 2.821 3.250 4.781 10 1.372 1.812 2.228 2.764 3.169 4.587 11 1.363 1.796 2.201 2.718 3.106 4.437 12 1.356 1.782 2.179 2.681 3.055 4.318 13 1.350 1.771 2.160 2.650 ,3.012 4.221 14 1.345 1.761 2.145 2.624 2.977 4.140 15 1.341 1.753 2.131 2.602 2.947 4.073 16 1.337 1.746 2.120 2.583 2.921 4.015 17 1.333 1.740 2.110 2.567 2.898 3.965 18 1.330 1.734 2.101 2.552 2.878 3.922 19 1.328 1.729 2.093 2.539 2.861 3.883 20 1.325 1.725 2.086 2.528 2.845 3.850 21 1.323 1.721 2.080 2.518 2.831 3.819 22 1.321 1.717 2.074 2.508 2.819 3.792 23 1.319 1.714 2.069 2.500 2.807 3.767 24 1.318 1.711 2.064 2.492 2.797 3.745 25 1.316 1.708 2.060 2.485 2.787 3.725 26 1.315 1.706 2.056 2.479 2.779 3.707 27 1.314 1.703 2.052 2.473 2.771 3.690 28 1.313 1.701 2.048 2.467 2.763 3.674 29 1.311 1.699 2.045 2.462 2.756 3.659 30 1.310 1.697 2.042 2.457 2.750 3.646 40 1.303 1.684 2.021 2.423 2.704 3.551 60 1.296 1.671 2.000 2.390 2.660 3.460 120 1.289 1.658 1.980 2.358 2.617 3.373 m 1.282 1.645 1.960 2.326 2.576 3.291 Source: Sidney Siegel, Nonparametric Statistics for the Behavioral Sciences (New York: "5‘19 6), p. 248. McGraw-HilI, "I11111111114111leS