{"http:\/\/dx.doi.org\/10.14288\/1.0089280":{"http:\/\/vivoweb.org\/ontology\/core#departmentOrSchool":[{"value":"Arts, Faculty of","type":"literal","lang":"en"},{"value":"Psychology, Department of","type":"literal","lang":"en"}],"http:\/\/www.europeana.eu\/schemas\/edm\/dataProvider":[{"value":"DSpace","type":"literal","lang":"en"}],"https:\/\/open.library.ubc.ca\/terms#degreeCampus":[{"value":"UBCV","type":"literal","lang":"en"}],"http:\/\/purl.org\/dc\/terms\/creator":[{"value":"Hennig, Karl H.","type":"literal","lang":"en"}],"http:\/\/purl.org\/dc\/terms\/issued":[{"value":"2009-07-02T22:16:31Z","type":"literal","lang":"en"},{"value":"1999","type":"literal","lang":"en"}],"http:\/\/vivoweb.org\/ontology\/core#relatedDegree":[{"value":"Doctor of Philosophy - PhD","type":"literal","lang":"en"}],"https:\/\/open.library.ubc.ca\/terms#degreeGrantor":[{"value":"University of British Columbia","type":"literal","lang":"en"}],"http:\/\/purl.org\/dc\/terms\/description":[{"value":"The purpose of this dissertation was to undertake a conceptualization and empirical\r\n\"mapping\" of the ethic-of-care domain - often characterized as self-referential and lacking in\r\nrigor. The current focus is upon conventional forms of care, involving notions of moral\r\n\"goodness\" as self-silencing and -sacrificial.\r\nEmploying a \"super\" circumplex as a prescriptive and descriptive tool, projected item\r\nanalyses were undertaken as a theoretico-structural clarification of existing scale items, along\r\nwith a provisional pool of additional items generated as part of Study 1. Based in part on\r\nproposed circular criteria, the Conventional Care Scales (CCS) were developed and\r\nsubmitted to a conjoint principal components analysis along with the battery of other\r\ncare\/dependency scale items. An examination of item circular distributions, factor loadings,\r\nalpha-contribution plots, and thematic content revealed several factors expressive of two\r\n\"faces\" of conventional care, submissive and ingenuous. These two forms were shown to\r\nhave unique correlates with measures of adjustment, interpersonal competencies, other\r\nfactors of the Five-Factor Model, false-self beliefs, and reported distress in narrated accounts\r\nof rejected care giving. Gender differences in the association between indices of adjustment\r\nand conventional care were also found. The range of conventional care was also extended\r\nthrough the development of scales reflecting other-directed and socially prescribed\r\ndimensions of conventional care. The factor structure for the CSS was also replicated in a\r\nsecond sample (Study 2). Participants for Studies 1 and 2 were composed of undergraduate\r\nstudents (N = 302 in both samples) who completed a battery of questionnaires in the first\r\nstudy and the CCS alone in the second study.\r\nAnticipating future clinical directions, secondary analyses using structural equation\r\nmodelling were conducted on an existing data set (N = 92) which included measures of\r\nconventional care and perfectionism, along with indices of psychological adjustment and\r\neating disordered attitudes. Results indicated that conventional care, for which there is little\r\nresearch, was more predictive of adjustment and eating disordered attitudes than\r\n\r\nperfectionism, for which there exists a large clinical literature. This research contributes to\r\nan understanding of ways in which an ethic of care can \"go awry,\" as well as proposes a\r\nresearch platform upon which the clinical implications of morality and self-ideals can be\r\ninvestigated. These findings speak to both the constraints and prescriptions that can inform a\r\nphilosophical ethic of care.","type":"literal","lang":"en"}],"http:\/\/www.europeana.eu\/schemas\/edm\/aggregatedCHO":[{"value":"https:\/\/circle.library.ubc.ca\/rest\/handle\/2429\/9977?expand=metadata","type":"literal","lang":"en"}],"http:\/\/purl.org\/dc\/terms\/extent":[{"value":"16935876 bytes","type":"literal","lang":"en"}],"http:\/\/purl.org\/dc\/elements\/1.1\/format":[{"value":"application\/pdf","type":"literal","lang":"en"}],"http:\/\/www.w3.org\/2009\/08\/skos-reference\/skos.html#note":[{"value":"MAPPING T H E C A R E DOMAIN: CONCEPTUALIZATION, ASSESSMENT, A N D R E L A T I O N TO EATING DISORDERS by K A R L H . HENNIG B E d . , University of British Columbia, 1982 M.C.S., Regent College, 1994 M.A. , University of British Columbia, 1995 A DISSERTATION SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE D E G R E E OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES Department of Psychology We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A October 1999 \u00a9 Karl H. Hennig, 1999 In presenting this thesis in partial fulfillment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Psychology The University of British Columbia Vancouver, Canada Date: 13 October 1999 A B S T R A C T The purpose of this dissertation was to undertake a conceptualization and empirical \"mapping\" of the ethic-of-care domain - often characterized as self-referential and lacking in rigor. The current focus is upon conventional forms of care, involving notions of moral \"goodness\" as self-silencing and -sacrificial. Employing a \"super\" circumplex as a prescriptive and descriptive tool, projected item analyses were undertaken as a theoretico-structural clarification of existing scale items, along with a provisional pool of additional items generated as part of Study 1. Based in part on proposed circular criteria, the Conventional Care Scales (CCS) were developed and submitted to a conjoint principal components analysis along with the battery of other care\/dependency scale items. An examination of item circular distributions, factor loadings, alpha-contribution plots, and thematic content revealed several factors expressive of two \"faces\" of conventional care, submissive and ingenuous. These two forms were shown to have unique correlates with measures of adjustment, interpersonal competencies, other factors of the Five-Factor Model, false-self beliefs, and reported distress in narrated accounts of rejected care giving. Gender differences in the association between indices of adjustment and conventional care were also found. The range of conventional care was also extended through the development of scales reflecting other-directed and socially prescribed dimensions of conventional care. The factor structure for the CSS was also replicated in a second sample (Study 2). Participants for Studies 1 and 2 were composed of undergraduate students (N = 302 in both samples) who completed a battery of questionnaires in the first study and the CCS alone in the second study. Anticipating future clinical directions, secondary analyses using structural equation modelling were conducted on an existing data set (N = 92) which included measures of conventional care and perfectionism, along with indices of psychological adjustment and eating disordered attitudes. Results indicated that conventional care, for which there is little research, was more predictive of adjustment and eating disordered attitudes than ' i i i perfectionism, for which there exists a large clinical literature. This research contributes to an understanding of ways in which an ethic of care can \"go awry,\" as well as proposes a research platform upon which the clinical implications of morality and self-ideals can be investigated. These findings speak to both the constraints and prescriptions that can inform a philosophical ethic of care. iv T A B L E OF CONTENTS Abstract i i List of Tables viii List of Figures ix Acknowledgements xi Chapter 1: Introduction 1 Kohlberg and the Justification of Norms 5 Gilligan and the Voice of Care 8 Empirical Investigations 11 Women's Interdependent Self 11 The \"Voices\" of Care versus Justice 13 Silencing the Self 15 Dependency, Attachment, and Care 20 Future Directions and Finding Common Ground 24 Clinical Applications 30 Summary and Overview 33 Chapter 2: Mapping and Scale Development Studies 37 Introduction 37 Construct Explication 39 Provisional Item Development 40 Method 40 Participants 40 Demographic Information 42 Factor Structure Measures 42 Dependency and Conventional Care Measures 44 Outcome Measures 47 V Results 51 Creating the \"Super\" Circumplex Space 53 Projecting Outside Variables onto Circular Space 57 Circular Analysis of Scale Items 61 Scale Development and Fidelity to Circular Structure Criteria 69 Conventional Care Scale Development 72 Preliminary Orientation to Scales and their Projections 81 Introduction to the Conjoint Principal Components Analyses 85 Conjoint Principal Components Analysis: Octant HI (Unassured-Submissive) 86 Conjoint Principal Components Analysis: Octant JK (Unassuming-Ingenuous) 101 Further Mapping: Octant NO (Gregarious-Extraverted) 109 Further Mapping: Octant F G (Aloof-Introverted) 113 Beyond the Self-directed Dimension 114 The Super Circumplex as a Nomological Net: Octant D E (Cold-hearted) 120 The Super Circumplex as a Nomological Net: Octant F G (Aloof-Introverted) 124 The Super Circumplex as a Nomological Net: Octant HI (Unassured-Submissive) 125 The Super Circumplex as a Nomological Net: Octant JK (Unassuming-Ingenuous). . \u2022 127 The Super Circumplex as a Nomological Net: Octant NO (Gregarious-Extraverted) 128 Interpersonal Competencies 131 Beyond the Circumplex 136 Gender Differences: Factor Structure Measures 139 Gender Differences: Care\/dependency and Outcome Measures 143 Gender Differences: False-self 145 Gender Differences: Motives for Helping 147 vi Regression Analyses Predicting Adjustment 150 Discussion 154 Chapter 3: Conventional Care and Eating Disorders 161 Introduction 161 Eating Disorders: Epidemiology and Characteristics 163 The Thinness Ideal: Why Women? Why Now? 167 Developmental Aspects: Why Young Women? 174 Care, Dependency, and Eating Disorders 177 Perfectionism 179 Method 182 Participants 182 Measures 182 Results 185 Structural Equation Modelling 185 Modelling Adjustment 187 Modelling Weight\/shape Concerns 188 Discussion 189 Chapter 4: General Discussion 194 Clinical Implications 200 Differences in Gender or Power\/dependency? '. 201 Conventional Care and Differentials of Power 202 Conventional Care and Dependency 204 Implications for a Theory of Authentic Care 207 Limitations and Future Directions 209 Summary 216 Philosophical Excursus 218 References 224 vii Appendix A: Levels of Justice and Care 247 Appendix B: Study 1 Measures 249 Appendix C: Study 2 Conventional Care Scales 278 Appendix D: Conjoint Principal Components Analysis for the Self-directed Care\/dependency Scales . . . . ; 283 Appendix E: Conjoint Principal Components Analysis for the Preliminary Anger-in Scale 286 Appendix F: Regression Analyses 287 Appendix G: DSM-IV Eating Disorder Diagnostic Criteria 292 Appendix H: Study 3 Measures 294 Appendix I: Correlations and Descriptive Statistics for the Eating Disorders Analyses 299 LIST OF T A B L E S Table 1: Comparison between Self-silencing Subscales and Measures of Care and Dependency 19 Table 2: Comparison of Scale Circular Statistics for Dependency Measures 67 Table 3: Comparison of Scale Circular Statistics for Measures of Conventional Care . . 68 Table 4: Factor Loadings for Self-directed Conventional Care Scales 75 Table 5: Factor Loadings for the Other-directed and Socially Prescribed Conventional Care Scales 76 Table 6: Descriptive Statistics for the Conventional Care Scales 77 Table 7: Intercorrelations among Care Scales 83 Table 8: Summary of the Conjoint Principal Components Analysis 108 Table 9: Factor Loadings for Non-self-directed Care Scales 118 Table 10: Correlations between Care\/Dependency Scales and Interpersonal Competencies 133 Table 11: Correlations between Care Scales and Self-orientations 147 Table 12: Correlations between Care Scales and Care Narratives 149 Table A - l : Kohlberg's Six Stages of Moral Judgment 247 Table A-2: Gilligan's Three Levels of Care 248 Table F - l : Hierarchical Regression Analyses Predicting Adjustment (Depression and Self-esteem) for Octant D E 287 Table F-2: Hierarchical Regression Analyses Predicting Adjustment (Depression and Self-esteem) for Octant HI 288 Table F-3: Hierarchical Regression Analyses Predicting Adjustment (Depression and Self-esteem) for Octant JK 289 Table F-4: Curve Estimation using Regression Analyses for Care\/dependency Scales Predicting Adjustment (Depression and Self-esteem) 290 ix LIST OF FIGURES Figure 1: Dependency as an Aspect of Attachment-detachment and Directiveness-receptiveness Dimensions 22 Figure 2: Structural Representation of Agency and Communion 25 Figure 3: Items and Circular Locations for the Interpersonal Adjectives Scales 28 Figure 4: Three Dimensions of Conventional Care and their Respective Scales 41 Figure 5: A Comparison of IASR and IIP-C Scale Items from Octant JK (Unassuming-Ingenuous) within Super Circumplex Space 54 , Figure 6: Structure of the IASR and the IIP-C in \"Super\" Circumplex Space 56 Figure 7: Projection of an Outside Variable onto an Interpersonal Space 58 Figure 8: Circular Dispersion of Items for the IBT - Demand for Approval (dependency) Subscale 63 Figure 9: Circular Dispersion of Items for the STSS, UCS, and the SPS-SD scales . . . . 64 Figure 10: Circular Dispersion of the Provisional Pool of Self-directed Conventional Care Items 65 Figure 11: Item Contribution to Alpha as a Function of Deviation from the Circular Mean for the IBT - Demand for Approval Subscale 71 Figure 12: Projection of Self-directed Conventional Care Scales onto the Super Circumplex 82 Figure 13: Typical Individual or Group Profile for the IASR or IIP-C Circumplex Data . . 85 Figure 14: Item Contribution to Alpha as a Function of Circumplex Angular Location for the Self-directed Social Perfectionism Scale 88 Figure 15: Item Contribution to Alpha as a Function of Circumplex Angular Location for the External Subscale 93 Figure 16: Item Contribution to Alpha as a Function of Circumplex Angular Location for the Silence Subscale 94 Figure 17: Item Contribution to Alpha as a Function of Circumplex Angular Location for the Care Subscale 104 Figure 18: Item Contribution to Alpha as a Function of Circumplex Angular Location for the Unmitigated Communion Scale 106 Figure 19: Characteristic Attitudes and Behaviors in Circular Space 110 X Figure 20: Super Circumplex Scale Projections for the Other-directed Social Perfectionism, Conventional Care, and Socially Prescribed Conventional Care Scales 117 Figure 21: Personality and Interpersonal Correlates of the Care and Dependency Scales for Octant HI (Unassured-Submissive) 137 Figure 22: Personality and Interpersonal Correlates of the Care and Dependency Scales for Octant JK (Unassuming-Ingenuous) 138 Figure 23: Personality and Interpersonal Correlates of the Care and Dependency Scales for Octant D E (Cold-hearted) 139 Figure 24: IASR Profiles for Women 141 Figure 25: IASR Profiles for Men 142 Figure 26: Scatterplot of Anxious Concern and Depression, with Linear and Quadratic Trend-lines 154 Figure 27: Path Diagram Showing the Structural Model for Perfectionism and Global Self-silencing (Conventional Care) in Predicting Adjustment 187 Figure 28: Path Diagram Showing the Structural Model for Perfectionism and Global Self-silencing (Conventional Care) in Predicting Weight\/shape Concerns . . .188 Figure 29: Schematic Summary of the Results of Structural and Substantive Analyses for the Universe of Care Content 196 XI A C K N O W L E D G E M E N T S Without the contributions of a great many wonderful and supportive persons the completion of this degree would not only have been impossible, but never aspired to. I thank Drs. Carl Armerding and Elmer Dyck for their initial encouragement and inspiration, members of my supervisory committee, Drs. Josie Geller, Wolfgang Linden, and Michael Chandler, for their assistance and mentorship. I especially want to thank Dr. Larry Walker for having truly been with me throughout those many challenging years and significant transitionary moments. He has been the very best friend, godfather, mentor, and colleague. I would like very much to thank the British Columbia Health Research Foundation and the Social Sciences and Humanities Research Council of Canada for fellowships supporting my doctoral studies. I also acknowledge the financial support for my dissertation research provided by a British Columbia Health Research Foundation grant and a U B C Humanities and Social Sciences research grant (to Dr. Walker). I had the good fortune of having several hard-working student research assistants. Above all I would like to acknowledge the patience, love, and support of my wife, Mary, without whom the receiving of this degree would have been left as an unfulfilled dream. 1 CHAPTER 1: INTRODUCTION The general purpose of this dissertation was to undertake an empirical investigation -a \"mapping\" - of the domain of conventional care; best understood as the intersect of both terms, \"conventional\" and \"care.\" It is the hope that the field of moral psychology might make an important contribution to both personality and clinical assessment through the theoretical and empirical selection\/development of a final set of measures most soundly representing the domain of conventional care. The term \"conventional,\" while potentially evoking a range of meanings, is here intended to make explicit reference to the work of Lawrence Kohlberg (e.g., 1981, 1986) which over the past several decades has dominated the field of moral psychology. Reportedly arising \"in part as a response to the Holocaust\" (Kohlberg, 1981, p. 470), Kohlberg's efforts were of a piece with social psychology's studies in compliance (e.g., Milgram, 1963) and work of the Frankfurt School (Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950), all of which sought an evaluative base for examining the validity claims of social norms - thus offering a critical theory of conventional society. Kohlberg, in his now classic 1971 article, \"From Is to Ought: How to Commit the Naturalistic Fallacy and Get Away with It\" proposed that moral validity claims could be justified by pointing to their place in the unfolding course of human development. Later stages of development were said to provide an increasingly adequate basis for moral decision-making. In the current context, the term \"conventional\" takes its meaning from the second of three levels within Kohlberg's stage theory, first emerging around early adolescence. Kohlberg referred to this developmental level as a \"good boy\/girl\" morality, characterized by a mutual understanding of relationships perspective. It was with respect to the conventional level of development that Kohlberg framed a third, post-conventional level of moral reasoning, anchored by the Platonic conception that \"virtue is ultimately one, not many, and ... the name of this ideal form is justice\" (Kohlberg, 1981, p. 30). A fuller explication of Kohlberg's levels and stages of moral development is found in Appendix A. 2 Kohlberg's stages of moral development were specifically stages of justice reasoning. The second term in the phrase \"conventional care\" takes its reference from Carol Gilligan's so called \"ethic-of-care.\" Initiating what came to be called the Gilligan-Kohlberg debate, Gilligan's seminal 1982 book, In a Different Voice, was levelled not only against justice as the final arbiter of the moral good, but also against the individualistic conception of selfhood which Kohlberg's theoretical framework was thought to assume. Discussions in the area initiated by Gilligan have now rippled far and beyond their initial point of entry into the theoretical literature on moral development. Her \"voice\" metaphor has effectively become a stereotypic catch phrase within certain feminist circles, along with her supposedly more interpersonal conception of selfhood. The ethic of care became both a source of empowerment (by affirming it as a valid morality distinct from justice concerns), as well as a target for (feminist) social analysis in examining \"conventions of goodness where the good woman is 'selfless' in her devotion to meeting others' needs\" (Gilligan, 1990, p. 9). From the perspective of Gilligan's initially proposed developmental schema and its three levels of care, the current dissertation undertakes an examination of her Level 2 care. See Appendix A for a fuller description of Gilligan's three levels of care. Skoe's sketch of Gilligan's Level 2 \"conventions of goodness,\" or what is here referred to as conventional care, will serve as an initial working definition: This perspective is characterized by a strong emphasis on responsibility and maternal\/paternal morality that seeks to provide care for the dependent and unequal. \"Good\" is equated with self-sacrificing care for others. The person adopts societal values, and conventionally-defined goodness becomes the primary concern because survival is now seen to depend on the acceptance of others. \"Right\" is defined by others and responsibility for defining it rests with them. The person has a strong need for security and avoids taking responsibility for choices made. S\/he feels responsible for the actions of others whereas others are responsible for the choices she or he makes. The strength in this position lies in its capacity for caring; the limitation lies in the prohibition of self-assertion. Conflict arises specifically over the issue of hurting and others are helped or protected often at the expense of self-assertion, (p. 13) While both Kohlberg and Gilligan began with the notion of the \"conventional\" as 3 potentially oppressive and something to be transcended, providing a moral vantage point from which it might be critiqued, they did so by moving in different directions. Kohlberg was interested in universal moral judgments, whereas Gilligan moved toward the more contextual and particular, especially towards the party or identity politics of women's unique moral concerns. Gilligan speaks most frequently about two aspects of care, avoiding harm and pursuing care; or what will here be referred to as the two faces of conventional care, \"submissive care\" and \"ingenuous care,\" respectively. Rather than view the two as mutually defining or reflecting a singular global self-silencing construct (Jack, 1991), my emphasis is upon an exploration of the construct's multifaceted nature. Five specific goals guided this dissertation. The first purpose was to provide a circular structure examination and clarification of the conventional care construct. The universe of semantic content was composed of existing scales, as well as an additional provisional pool of items generated as a part of this dissertation. What is meant by a circular structure examination will require some explanation in what follows, but the bulk of its detailing will be undertaken in Chapter 2. The construct of dependency is of particular comparative interest, on the expectation that it should share much of the same semantic space as conventional care. Following Pincus and Gurtman (1995), conventional care and dependency scales were examined within a circular \"super\" factor space circumscribed by the two axes of dominance-submissiveness and nurturance-cold heartedness. The circular factor space in which scales were plotted was employed, both in its descriptive role as well as its prescriptive role: descriptive, in that circular space can function as a nomological net in which to undertake the task of construct validation. Its prescriptive role is operative in two ways: (a) by prescribing a location within circular space where mature care can be found, linking this with a more philosophical discussion undertaken in the final chapter of this dissertation; and (b) a more narrowly psychometric 4 construal in prescribing a set of circular criteria for use in test development. Second, the process of structural clarification, or \"mapping,\" of the conventional care domain was continued through the development of additional scales from the provisional \u2022 pool of items, entitled the Conventional Care Scales. The development of these scales made explicit use of the proposed circular criteria - in addition to the more usual substantive item-total correlations and factor loadings - in their final item selection. Third, the process of construct validation was extended to include the familiar full Five-Factor Model and a variety of outcome constructs: interpersonal competencies, depression, self-esteem, false-self attitudes, and the narrated experience of distress as a rejected caregiver. Finally, and in line with the recent shift in moral psychology (Walker & Hennig, 1997) toward an examination of real-life over abstract hypothetical dilemmas, this dissertation was intended to locate itself within an applied behavioral context by examining the role of conventional care in predicting problems for which women are primarily at risk, eating disorders. This investigation takes the form of some preliminary structural equation modelling of an existing data base. The purpose of this preliminary investigation was to examine, within a largely eating-disordered sample, the comparative relation between conventional care (for which there is a sparse literature) and perfectionism (for which there exists an extensive literature) in predicting adjustment and weight\/shape concerns. By way of an overview to what follows, the notion of a Kohlbergian convention will be examined in more detail, followed by a similar expansion on the work of Gilligan and her ethic of care. The sketch of Gilligan's ideas will be followed by an empirical literature review reporting evidence in favor of her proposal: (a) that women possess a more interdependent, less bounded and individuated sense of self than men; and (b) that women construe moral dilemmas more from within a care orientation than a justice orientation in comparison to men. On Gilligan's (1982, p. 2) interpretation, it is precisely this interdependence of the female self that accounts for women's interpersonal sensitivity to 5 issues of responsiveness and care, those very elements said to be absent from male-individuated construals of both the self and the moral domain. Receiving the greatest emphasis in the present investigation is Jack's (1991) empirical operationalization of Gilligan's Level 2 ethic of care, or what is here referred to as \"conventional care.\" Jack's Silencing the Self Scale is the most broadly employed scale of Gilligan's Level 2 ethic of care, and will be discussed in more detail, although in Study 1 several other care-like scales will be examined as well. The expectation that will be fleshed out is that the ever-burgeoning development of \"care\" scales might be usefully clarified and constrained by considering their relation both to each other, and to the dependency literature. It is hoped that this analysis might mitigate the charge of self-insularity that has been leveled against Gilligan and her colleagues. The use of a structural model, such as that employed in the current investigation, is particularly suggestive, allowing for the comparison and contrast of scales within a well-recognized semantic space. Kohlberg and the Justification of Norms Norms involve shared behavioral expectations. Let's suppose that person A does something that person B considers wrong in some broad moral sense. Person B may say, \"Why did you do that?,\" or, if person A is found in the midst of doing what they shouldn't have, \"What do you think you're doing?\" In response, a person could: (a) point to mitigating circumstances (e.g., \"Normally I wouldn't do that but....\"); or (b) claim ignorance or offer a counter-justification (e.g., \"What's wrong with that?\"). Note that the question itself calls for a response, which the accused feels some force to rejoin. Some give-and-take of reasons and justifications ought to naturally ensue. These sorts of interactions take place regularly. \"Why have you come home late? The supper is cold and you didn't phone.\" Asserting the expectation and the felt need for such a response both reflect the intersubjective norms which make up the tacit nature of communication and mutual compliance\/consensus-seeking. Coming home late may be 6 wrong because one should show consideration and respect for others, or because agreements are important in relationships, or being able to count on someone is important. Some normative claim rooted in interpreted needs and interests (Rehg, 1994) is made by one upon the other. It was Kohlberg's seminal effort to demonstrate how the process of individually validating norms proceeded in an age-related sequential fashion. Kohlberg's Moral Judgment Interview is an open-ended discussion of standard hypothetical dilemmas frequently encountered within moral philosophy. Responses are assigned to one of six stages. For example, in one of Kohlberg's hypothetical dilemmas, the son Joe is asked by his father to render up his hard earned paper-route money so the father can go on a fishing trip. \"How contemptible!\" it might be said of the father. Or on the other side, \"After all the father has done for the boy it is the least the son can do!\" How one comes down on the issue, and the justifications or judgments offered in defense, are all potentially scorable according to Kohlberg's elaborate coding manual (Colby & Kohlberg, 1987). This particular hypothetical dilemma pits the son's property rights (i.e., entitlement) against his duty\/obligation to the father in their complementary roles as father (responsible authority) and son. A scorable conventional Stage 3 judgment would be: \"The most important thing a son\/father should consider is to try to understand the other, respect the other's feelings, see each other's point of view, be willing to listen to each other, or think what it is like to be a child or parent.\" This moral judgment best reflects the \"balancing perspectives or role taking\" developmental capacities that emerge around early adolescence, reflective of the justice ideal at conventional Stage 3. The insight emerges at Stage 3 that self and other can simultaneously consider each other's viewpoint. What the interviewer seeks to elicit from the research participant over the course of the approximately half to three-quarter hour interview is reasoning on both sides of the issue, reasoning regarding both entitlements and obligations. A final stage score is assigned based on the type of moral judgments generated across all scorable judgments. Little research has 7 been done examining those cases where individuals repeatedly produce material on only one side of the bipolar entitlement-obligation dimension. Among the rare exceptions, Trevethan and Walker (1989) found that psychopaths used considerably more egoistic utilitarian judgments in their discussions of real-life dilemmas than did delinquents, reflective of their focus on self-interest. More indirectly, much has been said within cultural criticism about the way the language of classic liberal rights has constituted a greater social expectation of entitlement in the West, having lost sight of the reciprocal obligatory pole (Elshtain, 1993). Overall, the work of Gilligan could be seen to consist of a discussion of the (complmentary) obligatory side of Kohlberg's conventional Stage 3, by attempting to understand Stage 3 structure as expressive of a desire to \"preserve relationships.\" It was Gilligan's initial argument that women's \"concerns with preserving relationships\" had originally resulted in their being downgraded as morally immature relative to men in Kohlberg's scoring system. She argued that Kohlberg's model resulted in more women being stuck at the Stage 3 relationships perspective whereas more men went on to \"higher\" stages of principled moral reasoning. Following a similar line of argument, Jack (1991), of whom more will be said in what follows, also linked conventional care with the (woman's) desire to preserve relationships, which she, in turn, further relates to constructs such as \"anxious attachment\" and \"compliant connectedness.\" Whether these hypothesized relations actually hold are empirical questions the current study sought to address. Picking up on Kohlberg's acknowledgement that his cognitive-developmental account may fail \"to map the entire moral domain\" (Kohlberg, 1986, p. 500, italics added), Gilligan edited a book entitled, Mapping the Moral Domain (1988). Kohlberg's moral judgment coding manual (Colby & Kohlberg, 1987) makes no claim to having provided a complete listing of all possible moral judgments that persons are capable of making in defense of an action. This \"mapping\" project was intended to link her ethic of care and Kohlberg's justice ethic, calling \"attention to moral judgments that did not fit the [Kohlbergian justice] definition of 'moral' and to self-descriptions at odds with the [interpersonal] concept of self\" 8 (p. xvii). The current research continues the mapping process, with a particular eye to assessing existing constructs and further developing new ones inherent within the \"domain\" of care - or more specifically (Gilligan's Level 2) conventional care where \"the 'good' woman becomes caring by becoming selfless\" (Gilligan, 1998, p. 138). Where this work diverges from Kohlberg's and Gilligan's approaches is in moving toward the examination of continuous variables. The problem with any typology or categorical variable where an individual is described as not\/being of Type X , be it a developmental one or not, is that considerable variance is lost. According to Skoe (1993), for example, Level 2 global stage descriptions, of the sort quoted above, an individual would be coded at a conventional care stage by justifying not leaving a relationship, for example, so as \"not to let people down,\" or because \"it would not be the religiously sanctioned thing to do.\" The stage structure may well be present, but the degree to which a care focus on the other involves a repudiation of self is lost. A good deal of codable variability is especially needed to provide the necessary ceiling room for the creation of clinical scales. Secondly, an examination of the facets - or stage \"elements\" to use Kohlbergian language - may prove them to be empirically quite distinct, even though they share an underlying common structural background. That is, facets of the conventional care construct may prove to have very distinct sets of correlates, too easily passed over when relying on a notion of a global stage. As elements within a stage level, the multiple facets of the construct of care must, it is argued, also be examined in relation to associated constructs such as dependency and attachment. G i l l i g a n a n d the V o i c e o f C a r e Gilligan's (1982) proposal of the existence of two fundamentally distinct and gender-related moral orientations, care and justice, sparked a large controversy that has not yet abated. Gilligan's move was to link a (personalist) conception of morality - a conception that has historically long stood in opposition to Kohlberg's (impartialist) tradition - with both gender and conceptions of identity associated with care and justice. 9 Gilligan argued that the masculine sense of self is bounded and separable, not containing the other as part of its identity; whereas the feminine sense of self is bound up with the other. Women, on this account, define themselves in relation to others, and men do not. Gilligan articulates the Kohlbergian justice\/rights orientation in terms of separateness and the care\/response orientation in terms of connectedness. \"Responsibility,\" according to Gilligan, is understood in terms of responsiveness to the needs of the other, rather than obligation\/duty based on deduction from abstract principles. The separate-self judges according to reciprocity, rules, and roles. In much of the same voice, Haan (1986) caricatures the Kohlbergian moral agent as consulting the stage structure of their development to locate the relevant general rule or principle that will decide the issue by transcending its content and its social-personal context.... When the additional judgment of self-responsibility is made, people will act in accordance with their stage.... In this view actors have a certain detachment from moral conflict, (p. 1272) Gilligan's etiological account of the gender differences is based on a same-sex parental identification process more familiar within certain psychodynamic accounts of development (Chodorow, 1978). She argues that boys are required to separate from the mother in order to identify with their father, whereas the developmental experience of girls is different, in that gender identification does not require separation from the mother. On this classical account, two distinct experiences arise for the boy and the girl. The boy, once having separated, comes to value autonomy and objectivity, whereas the girl comes to value connectedness and caring relationships. As young girls reach adolescence, however, they come to a fork in their pathway; a point in development while pressing girls to separate also heightens their sensitivity to relationships and disconnection. As Gilligan argues, Girls, because of their more acute personal encounter with disconnection at adolescence [become] alerted to problems of connection at a time in history when relationships in general have become corrupt, (p. 11) Based on a more qualitative or collaborative \"listening\" to 100 adolescent girls from a boarding school in the Midwest, Gilligan concludes that, for girls, adolescence constitutes a 10 special \"impasse in female development\" (1990, p. 9). I felt at times that I was entering an underground world, that I was led in by girls to caverns of knowledge.... What I heard was at once familiar and surprising: girls' knowledge of the human social world, a knowledge gleaned by seeing and listening, by piecing together thoughts and feelings, compelling in its explanatory power and often intricate in its psychological logic, (p. 14) On the basis of such evidence, Gilligan reports that \"much of what psychologists know about relationships is also known by adolescent girls\" (p. 24). Girls at this point are taught to give up their authentic voice (what they really know regarding their valuation of connection) in order to fit into a world where they are required to be \"nice,\" \"good,\" \"caring,\" \"helpful,\" \"undemanding,\" and \"unselfish.\" In response to such pressures, girls come to answer with the sign of repression, \"I don't know,\" in response to questions that were once spoken with self-confidence. Consequently, the clinical interviewing approach, according to Gilligan, is necessary in order to move through the introjected voices of conventional morality. I also begin to listen with girls to the voices which they are taking in. Opening their ears to the world, listening in, eavesdropping on the daily conversations, girls take in voices which silence their relational knowledge.... Voices which intentionally or unintentionally interfere with girls' knowing, or encourage girls to silence themselves, keep girls from picking up or bringing out into the open a series of relational violations which they are acutely keyed into, such as not being listened to, being ignored, being left out, being insulted, being criticized. (1991, p. 19) The voices which precipitate this process of dissociative self-silencing involve the (patriarchal) sanction against selfishness, where \"goodness\" is understood as self-sacrificial care. The wall that keeps memory [read: one's authentic voice] from seeping through these covers may be the wall with the sign which labels body, feelings, relationships, knowing, voice and desire as bad. (Gilligan, 1991, p. 23) Gilligan's beliefs regarding female adolescent development directly build on her earlier formulations regarding the centrality of women's interdependent sense of self. Gilligan's In a Different Voice (1982) has now become canonical in the study of gender differences; introducing the notion of \"voice\" to contemporary feminist scholarship (Davis, 1994). \"Voice\" came to express a feminine morality, since it both referred to an ethic of care 11 and was expressed to a greater extent by women. Gilligan's work became a rallying point for feminist scholarship, transforming the very qualities historically used to declare women as inferior into virtue. Empirical Investigations Do women experience a sense of self distinct to that of men, one less bounded and more relationally defined? Several studies have sought to examine the questions raised by Gilligan and others. In what follows, empirical studies relating to the two parts of Gilligan's argument will be examined: (a) the existence of an interdependent self; and the (b) the presence of an unique \"voice\" or ethic of care. Women's Interdependent Self A number of researchers have pursued the question of gender differences in self-construal. McGuire and McGuire (1982), using an open-ended format with school children aged 7 to 17, found that more girls than boys expressed self-conceptions that were fundamentally social. Girls described themselves in terms of other people 50% more often, and including more spontaneous references to significant others, whereas boys included more spontaneous references to people in general. Similarly, using an autobiographical method, Clancy and Dollinger (1993) found women included within their photo albums more pictures of themselves with others and more pictures including family members. By contrast, men tended to have more pictures of themselves alone than did women. In descriptions of their ideal and undesired selves, women were also found to be more likely than men to include relationships (Bybee, Glick, & Zigler, 1990; Ogilvie & Clark, 1992). In a related vein, when participants were asked to respond to experimenter-selected attributes, men were more likely to evaluate themselves positively on those representative of independence (e.g., power and self-sufficiency), whereas women regarded themselves positively on more interdependent attributes (e.g., likability or sociability; for reviews see Maccoby & Jacklin, 1987; Simmons, 1987). In addition to gender differences found in the content of self-descriptions, using both 12 spontaneous self-descriptions and fixed questionnaire formats, men and women are also reported to differ in the centrality or importance of certain values. In a study of adolescents' self-concepts, for example, girls tended to regard interpersonal harmony and sensitivity as more important than did boys. In contrast, boys in this study regarded social dominance and toughness as more important than did girls (Rosenberg, 1986; see also Eccles, Wigfield, Flanagan, Miller, Reuman, & Yee, 1989). Similarly, in a study examining the relation of adult roles to identity, women ranked relational aspects (e.g, spouse, friend, son or daughter) as more central to their identity than did men (Thoits, 1992). Gender differences also have been found by those using various information-processing approaches of the so-called \"self-referential effect.\" The assumption of this line of research is that individuals have better memory recall for words encoded with respect to the self. As such, words that are more readily recalled are assumed to have been more \"deeply processed\" (Greenwald & Pratkanis, 1984). Following these dictates, Josephs, Markus, and Tafarodi (1992) studied the hypothesis that men's and women's self-esteem is grounded in different sources. Based on the assumption that self-esteem is derived from succeeding in what is valued within a particular socio-cultural niche, they hypothesized that women with high self-esteem would recall a significantly greater number of words that involved associations with a close friend. In general, their results are said to support the view that women [with high self-esteem] have highly elaborated structures of knowledge about important others and ... the information encoded with respect to these others can be used to produce a rich, highly memorable encoding of the stimulus words in these conditions, (p. 396) By extension, one would think that the memory of women for people and relational events should be more accurate than that of men, who are assumed to have a more independent self-construal. Consistent with such expectations, a meta-analytic review conducted by Hall (1984) found that women had better recall for faces than did men, d = .34. This same gender difference in face recognition has also been found in children as young as 13 4 years old (Feldstein, 1976). In comparison with men, women recall more details of persons casually encountered in the street (Yarmey, 1993), remember more high school classmates' names and faces in the years following (Bahrick, Bahrick, & Wittinger, 1973), and recall more vivid and detailed accounts of relational events (e.g., a vacation, an argument, or first date; Ross & Holmberg, 1992). Similarly, female counsellors are reported to recall more details of their clients than do male counsellors (Buczak, 1981). By contrast, it was found that men have a greater recall of historical events (Storandt, Grant, & Gordon, 1978). Similarly, men are more likely to over-estimate the degree to which they consider their own characteristics or attributes as unique and unshared by others, a phenomenon referred to as the \"false uniqueness bias\" (Goethals, Messick, & Allison, 1991). Women also reported that not being forgiven by a friend would have a greater impact on their self-esteem than did men (Hodgins, Liebeskind, & Schwartz, 1996). The \"Voices\" of Care versus Justice Direct empirical investigations of Gilligan's claims using semi-structured interviews of real-life and hypothetical dilemmas (e.g., \"Should Heinz steal the drug that would save his wife?\") have not only partly confirmed her hypotheses, but have opened up new avenues for their interpretation (Walker, 1991). On the strong form of Gilligan's account - that care and justice orientations divide along strict gender lines - studies examining intraindividual consistency have found that the number of both women and men who used the same orientation across two real-life dilemmas did not differ from chance (Pratt, Golding, Hunter, & Sampson, 1988). Such findings do some damage to Gilligan's claim that \"most people ... focus on one orientation and minimally represent the other\" (1986, p. 10). Others report, however, that women produce a proportionately greater number of care responses on real-life dilemmas than do men (Walker, 1989; Langdale, 1986). When the type of dilemma (personal versus impersonal) is controlled, however, gender differences lose their significance (Walker, 1995). Personal dilemmas (i.e, those involving conflicts among persons with an ongoing relationship) elicit more care reasoning for both men and women. 14 Impersonal dilemmas (i.e., those involving conflicts with strangers or institutions) were found to similarly elicit more justice reasoning. What began as a possibly strict division between care and justice types of moral reasoning now appears to turn on the kind of dilemma the participant generates. The further question is then begged: Why are more women raising personal dilemmas? Another possible contributing factor to possible gender differences in rights\/care orientations concerns the potentially polarizing effect inherent in the parenting role. Walker (1989; Walker, de Vries, & Trevethan, 1987) found gender differences in moral orientation among parents, but not children. Pratt et al. (1988) compared two samples of same-aged adults, half of whom were parents and half, not. This pattern was only evidenced among the parents, indicating \"that the sex difference in moral orientations is of rather limited generality\" (Walker, 1995, p. 97). Various meta-analytic investigations of the moral development litersture have suggested that gender differences are rapidly disappearing, supporting the socially constructed nature of gender stereotypes. Knight and colleagues (Knight, Fabes, & Higgins, 1996), however, question to what extent these causal inferences may not be an artifactual result of changes in research methodology. Based on their own meta-analyses of aggression - one of the most stereotypic of gender differences in favor of men - they conclude, \"gender differences in aggression appear to be remarkably stable when changes in study characteristics over time are controlled\" (p. 410). In general, Gilligan's critics have charged her theory as deficient in that it is reifying past gender stereotypes, and fails to give sufficient weight to related matters of ethnic and socioeconomic differences. In other words, Gilligan is said to have confused prescriptivity for descriptivity (Bebeau & Brabeck, 1989; Okin, 1989; Puka, 1991). It has been argued that a great deal of what has been said about women is equally true of poor people, who place less emphasis on work and self-advancement, are deferential to .those occupying higher positions of status, and appear to others as more sensitive and intuitive than rational. In 15 Pollitt's (1992) critique of Gilligan, the moral ideal of care seems to function more as a rhetorical means of advancing women's status by portraying them as morally superior. Miller (1986) has argued that because of the relative powerlessness of women in our culture, despite gains made by the Women's Movement, women must be sensitive and responsive to others, especially others who have power over them. Persons in positions of power may promote rights and rationality, whereas those in subordinate positions will use more indirect means of attaining ends by advocating connection and concern (Hare-Mustin & Marecek, 1988). In summary, recent investigations of Gilligan's understanding of women's interdependent self-construal and moral orientation of care, using open-ended questions, have been partly supportive; women do define themselves more interpersonally and produce more care reasoning when discussing real-life dilemmas than do men. By contrast, other research has highlighted a number of possible confounds: being in the role of a parent may have a polarizing effect on the type of orientation expressed, and particular types of dilemma (e.g., personal versus impersonal) may \"pull\" (Krebs, Vermeulen, Carpendale, & Denton, 1991) for one orientation over another. Studies using standardized questionnaires have generally failed to find gender differences (Friedman, Robinson, & Friedman, 1987; Walker, 1995). Gilligan's (Brown, Debold, Tappan, & Gilligan, 1991; Brown & Gilligan, 1992) more recent move to a \"narrative\" or qualitative approach to investigating the voice of care, has been found \"vague and unreliable\" by some (Wark & Krebs, 1996). The problem of successfully operationalizing the \"care\" construct has been an ongoing one. Efforts, like those of Jack (1991), to develop an objective measure of Gilligan's care construct are needed if the many confounds which invade the empirical investigation of gender differences are to be better understood. Silencing the Self With Jack's book, Silencing the Self (1991), and the introduction of a paper-and-pencil measure of self-silencing (The Silencing the Self Scale) or what has here been called 16 conventional care, Gilligan's ideas have been used to explore a variety of clinical problems ranging from depression to drug abuse. At present some 10 articles, 26 conference papers, and almost 50 theses and dissertations have included Jack's measure (D. Jack, personal communication, April 1998). At her Kohlberg Memorial Lecture Gilligan stated, In Silencing the Self: Depression and Women (1991), Dana Jack brought the process of dissociation into the centre of this conversation by showing how depressed women silence themselves.... we realise that we can not know what we know, not feel what we feel, not say what we mean, nor care about what we care most deeply about. (1998, p. 132) Elsewhere, and in a special feature responding to her critics, Gilligan cites Dana Jack's study of depression and women (Silencing the Self 1991) and the scale derived from that study... [as being] among the first of a growing number of innovative, empirical studies in developmental and clinical psychology that have been stimulated by my work on voice. (1994, p. 421) Jack's Silencing the Self Scale, however, has undergone very little in the way of construct validation and is typical of scales in the field. A number of points will require clarification before this burgeoning population of studies approaches intelligibility. The first point involves a question about the sorts of theoretical interpretation that are made of the self-silencing \"data.\" On the one hand the Silencing the Self Scale (STSS) is said to \"not overtly refer to the feminine role but only to its imperative\" (Jack & Di l l , 1992, p. 98); yet, on the other hand, it is described as \"an instrument to investigate gender-specific schemas ... about how to create and maintain safe, intimate relationships [which] lead women to silence certain feelings, thoughts, and actions\" (pp. 97, 98; italics added). Jack argues that the centrality of relationships for women's self-identity, in conjunction with gender role norms urging women to be compliant and unselfish, places women at unique psychological risk. While the results across several studies using mixed samples of both men and women report gender differences in favor of men doing the greater proportion of self-silencing (e.g., Cowan, Bommersbach, & Curtis, 1995; Jack & Di l l , 1992; Thompson, 1995), discussion continues to revolve around engendered self-in-relation theory: \"High scorers [on the Silencing the Self Scale] ... reflect greater pressure to fulfill norms of the 'good woman'\" 17 (Jack & Di l l , 1992, p. 99). Are men who self-silence to be understood as fulfilling norms of the \"good woman\"? Interpretation of various findings of high levels of self-silencing has turned on the conditional adverbial clause, \"When I am with my friends,\" that appears in many of the stems within the STSS. The assumption, it would appear, is that these behaviors are for the purpose of maintaining already existing relationships, albeit ones in which individuals are not very contented. Jack speaks of \"compliant connectedness\" (1991, p. 40) in a way that resembles anxious attachment. Depressed women's statements such as, \"I have learned, don't rock the boat with my partner\" ... show their conscious awareness of making themselves appear passive or compliant for an intended effect: to keep outer harmony, to preserve relationship. (Jack, in press, p. 9) The attachment literature, however, regards behaviors involving compulsive caretaking and anxious attachment as focused primarily upon a particular close attachment figure. Would the above description not also fit for the dependent woman of an abusive partner with the attendant motivation not to seek intimacy, but simply to maintain harmony for the preservation of life and limb? Attachment and anxious attachment, in particular, are also different constructs reflecting some general strong inclinations towards \"affiliation.\" There seems to be some potential for confusion between the supposedly distinct domains of self-silencing and \"anxious attachment,\" or other more general forms of social anxiety (Bartholomew & Battel, 1998). In a to-be-published chapter (Jack, in press), much of the discussion of the role of self-silencing revolves around women's vulnerability within our culture and the relational-self. On the few brief occasions when men's self-silencing is spoken of, it is in the context of a yet broader domain - that of work. Dan's [one of the interviewees] litany of shoulds contain not only an implied perfectionism, but also an image of selflessness focused on [his] professional role: he should be \"selflessly\" married to medicine, (p. 29) Self-silencing seems here to be no longer for the purpose of maintaining relationships but 18 occurs in the more impersonal context of individuals at work. Dan says, \"I feel that I need to put on my professional face and my caring face\" when he is with his patients (p. 29). Jack is surely correct when she writes, \"researchers must look behind self-silencing for its gendered meanings and its relational intent\" (p. 9). Regarding men's greater tendency of \"withdrawal through silence or passive resistance\" (Gottman, 1994), Jack's extracted intent is, \"when men self-silence, they may intend to create distance and control interactions in relationships\" (p. 9). If men's \"stonewalling\" is also a form of self-silencing then this represents a very different form of behavior from how self-silencing is described elsewhere; this represents a cold-avoidant form of interpersonal behavior. It may be that self-silencing reflects a great many interpersonal behaviors beyond anxious attachment and general social anxiety, but self-in-relation theory provides no suitable non-gender-specific interpretation. Part of the confusion enters when the designator \"self-silencing\" is intended to reflect both the global construct indicated by the title of the scale, Silencing the Self, as well as one of the subscales, Silence. Should self-silencing (i.e., conventional care) prove to be a multifaceted construct, this only adds to difficulties in gaining interpretive clarity. A second point of concern has to do with the dangers of any theoretical\/empirical enterprise becoming too insular and self-referential. If men also self-silence, then a broader conceptual framework will be needed to give some coherent account of this fact. Generally, what is needed is to locate the Silencing the Self Scale, and other scales of this sort, not only in conversation with one another, but more importantly within a nomological net of broader constructs whereby the instrument could either be verified as an independent contributor, or found redundant with existing variables. It is here suggested that the dependency and attachment literatures form just such a starting point for such an investigation. From Table 1 it can be seen that the thematic content of items within the various STSS subscales (in the left column) show considerable resemblance with those of other dependency\/care\/attachment measures (in the right column) such as the Sociotropy Autonomy Scale - Pleasing Others Subscale (SAS-PO; Beck, Epstein, Harrison, & Emery, 1983), the I] 19 Table 1 Comparison between Self-silencing Subscales and Measures of Care and Dependency Self-silencing SAS-PO \u2022 When friends' opinions conflict with \u2022 People can pretty easily change me mine, rather than asserting my own point even though I thought that my mind was of view I usually end up agreeing with already made up on a subject. them. IBT \u2022 I find it hard to go against what others think. Care UCS \u2022 In my friendships my responsibility is \u2022 For me to be happy, I need others to be to make the other person happy. happy. External SPS-SD \u2022 I often feel responsible for my friends' \u2022 I consider myself a failure if I can't act feelings. the way others want me to. Unmitigated Communion Scale (UCS; Helgeson & Fritz, 1998), the Social Perfectionism Scale - Self-directed (SPS-SD; Wiebe & McCabe, 1998), and the Irrational Beliefs Test -Demand for Approval Subscale (IBT; Jones, 1969). What is the relationship between self-silencing (i.e., conventional care) and the other dependency\/attachment constructs? There would appear to be a constellation of constructs here that invite comparison and contrast. Test and theory developers must be continually open to the possibility that the operationalized ground that their constructs are intended to furrow may be narrated from very different theoretical viewpoints. A structural clarification is needed to examine both the consistency within the proposed constructs, as well as their interrelations within a larger nomological net. For example, an item used in the current study reads \"I am afraid of making mistakes in conversations\" (Self-directed Social Perfectionism Scale). While the item was intended by authors Wiebe and McCabe (1998) to be a motivational construct aimed at seeking social perfection, it could just as easily be associated with avoiding relationships, or a dependency construct such as Fear of Negative Evaluation 20 (e.g., \"I feel very upset when I commit some social error\"; Watson & Clark, 1984). The relationships among the constructs of care, dependency, and attachment will require considerable clarification. Dependency, Attachment, and Care The constructs of care, dependency, and attachment all have been variously used with overlapping reference. Thompson states that Gilligan's notion of compliant connectedness \"resembles anxious attachment, [and] is characterized by compulsive caretaking, pleasing the other, and inhibition in self-expression\" (Thompson, 1995, p. 338). From Bowlby's perspective, however, the construct of compulsive caretaking was seen to be distinct from anxious attachment, and both attachment constructs were thought to be markedly distinct from dependency (Bowlby, 1973\/1991, 1980\/1991). What care, dependency, and attachment constructs do share, however, is a conception of persons as fundamentally seeking attachment objects and relations, as opposed to attachment as a drive derivative (Freud, 1930\/1955), or as a generalized association with sex and\/or food (e.g., behaviorism). The construct of dependency has its roots in the psychoanalytic image of the infant-caregiver relationship. Freud believed that, an excess of parental affection does harm by causing precocious sexual maturity and also because, by spoiling the child, it makes him [sic] incapable in later life of temporarily doing without love or of being content with a smaller amount of it. (1894\/1955, p. 223) In this light dependent persons are characterized as immature and childlike. Atop this basic metaphorical picture of the child's material dependency needs, when referred to adults the importance of identifying additional component constructs has been frequently recognized. The range and type of components or facets used to define the whole, however, varies. Zuckerman, Levitt, and Lubin (1961) defined dependency as involving succorance (attention and approval seeking), deference (a tendency to subordinate oneself to others and inhibit self-assertion), and abasement (involving self-blame and guilt). The Navran (MMPI) Dependency Scale possesses largely the self-critical facet of the construct 21 (Birtchnell, 1984). The list of constructs that has come to be associated with dependency is a lengthy and overlapping one: suggestibility, interpersonal compliance, conflict-avoidance, pessimism, self-doubt, deference, abasement, emotional reliance upon others, lack of social self-confidence, conformity, help seeking, need for approval, need for guidance, passivity, and pleasing others. Dependent persons are said to be interested primarily in avoiding responsibility and interpersonal conflict; they try to compensate for their sense of helplessness by obtaining support of others through occupying a lower status or follower position. (Assor, Aronoff, & Messe, 1981, p. 790) Against this largely disparaging psychological backdrop it was important for Bowlby (1969\/1991) to make a distinction between dependency and attachment. Logically, the word \"dependence\" refers to the extent to which one individual relies on another for his [sic] existence, and so has a functional reference; whereas attachment as used here refers to a form of behavior.... whereas dependency is maximum at birth and diminishes more or less steadily until maturity is reached, attachment is altogether absent at birth and is not strongly in evidence until after an infant is past six months. The words are far from synonymous, (p. 228, italics added) Bowlby sought to reserve the concept of attachment for behaviors directed towards one or a few particular attachment figures. The early parent-child attachment forms the blueprint for subsequent relationships, guiding behaviors and expectations in later romantic relationships in particular. Similar to children, securely attached adults will seek contact with a significant other in the face of stress, be comforted by the presence of the attachment figure, and experience distress at the threat of losing the attachment figure. On this account, the role of an attuned parental figure needed to be distinguished from the notion of \"spoiling,\" which is associated with dependency. It may, however, be questioned whether Bowlby's definition of dependency, as involving material \"existence\" (see above italics), isn't excessively narrow (Birtchnell, 1984). Bowlby (1969\/1991) proposed that dependency should be regarded as a fusion of Murray's (1938) need for affiliation (\"to form friendships and associations,\" p. 83) and need for succorance (\"to seek aid, protection, and sympathy,\" p. 83). Bowlby associated 22 Figure 1 Dependency as an Aspect of Attachment-detachment and Directiveness-receptiveness Dimensions Note. From \"Attachment-detachment, Directiveness-receptiveness: A System for Classifying Interpersonal Attitudes and Behaviour,\" by J. Birtchnell, 1987, British Journal of Medical Psychology, 60, p. 18. Copyright 1987 by the British Psychological Society. overdependency with insecure, more specifically anxious, forms of attachment. Most relevant to the current research, Birtchnell (1987) proposed understanding dependency as a broad multifaceted construct locatable within two fundamental dimensions of human functioning: directiveness versus receptiveness and attachment versus detachment (see Figure 1). As such, Birtchnell broadly locates dependency between \"the tendency to stay close\" and an \"inclination to assume a submissive attitude\" (p. 18). Drawing from attachment theory, Birtchnell defines attachment as \"a need for proximity and a fear of being alone,\" directiveness as \"an inclination for giving or doing things to others ... characterized by a fear of being controlled or taken over,\" and receptiveness as \"an inclination to assume a recipient attitude towards others and is characterized by a fear of assuming responsibility\" (p. DIRECTIVENESS (inclination to assume a dominant attitude) DETACHMENT (tendency to remain distant) ATTACHMENT (tendency to stay close) DEPENDENCY RECEPTIVENESS (tendency to assume a submissive attitude) .25). 23 One of the problems with the above proposal is that, in mapping out the domain of dependency as involving an entire quadrant, and some area in the quadrant immediately above (as will later be discussed), Birtchnell contains within the one construct, variables which are orthogonal (i.e., attachment and directiveness are orthogonal dimensions). One solution, and the one taken here, is to propose a division of constructs based on their position within further divisions of the quadrant. More will be said of this proposal below. Motivated in a similar way to Bowlby, Gilligan sought to distance the ethic of care and women's associated interdependent sense of self from Freud's conceptions of morality and dependence. Because they were judged to be destitute as a result of an incomplete resolution of the Oedipus complex and thus not having a properly developed superego, Freud inferred that women \"show less sense of justice than men, that they are less ready to submit to the great exigencies of life, [and] that they are more often influenced in their judgements by feelings of affection or hostility\" (1925, pp. 257-258). It was through Chodorow's (1978) re-interpretation that aspects of Freud's dependency and moral inferiority became the virtues of care: \"Girls emerge with a stronger basis for experiencing another's needs or feelings as one's own\" (Chodorow, 1978, p. 167). In light of what has been said above, what is to be made of the rapidly expanding data using Jack's Silencing the Self Scale and scales of its kind? Critics of Gilligan have charged her with being overly schematic. Tavris (1994) is doubtful that: girls and women speak consistently across situations, that adolescent patterns should determine later adult outcome, that self-silencing is unique to or common among women, that \"power\" is male and \"powerlessness\" female, and that self-silencing is strictly a bad thing. Are men who self-silence to be understood as fulfilling norms of the \"good woman\" or should the conventional sanction against selfishness be moored elsewhere? Questions multiply, with an absence in the field of sound measures by which to undertake their investigation. It can be anticipated that dependency, attachment, and (conventional) care are locatable within the same semantic space. If these constructs and their respective scales are 24 not located within some broader nomological net and the offerings of Gilligan's self-in-relation theory not interpreted within a larger psychological framework, only further polarization and a lack of clarity will likely result. Problems arise when any interpretive community becomes too insular. Contratto comments, \"her [Gilligan's] writing and that of the other members of the Harvard Project has become increasingly self-referential\" (1994, p. 368). The current study sought a structural clarification across a superordinate set of terms, or higher point of view, by which dependency, care, and attachment might be discussed simultaneously. Future Directions and Finding Common Ground One of the primary correctives that Gilligan introduced to moral psychology was to stress the importance and complexity of the interpersonally embedded particular, as captured in the metaphor of \"voice\" - over against the tendency to moral abstractions and universal generalizations. Favoring localizations of \"voice\" over the universal and general, however, risks losing sight of possible common grounds for discussion. There has been considerable consensus regarding the broader framework from within which the constructs of attachment, dependency, and care should be studied. Beck (1983) speaks of sociotropy versus autonomy, Blatt (1990) interpersonal relatedness versus self-definition, and Birtchnell (1987) attachment versus directiveness. Earlier Bakan (1966) had proposed the existence of two independent modalities, agency and communion. Agency refers to the character of being a differentiated individual, and is expressed in strivings for mastery and power. Communion refers to the character of being connected with a larger social or spiritual whole, and is expressed in strivings for intimacy and connectedness with the larger whole. Similar to Birtchnell's figure (see Figure 1 above), Figure 2 presents a schematic view of the two dimensions of agency (A+) and communion (C+) and their conceptual opposite poles, passivity (A-) and disconnection (C-). These two, agency and communion, can be found to represent fundamental dimensions across a variety of worldviews, languages, and psychological theorizations (Wiggins, 1991). 25 Figure 2 Structural Representation of Agency and Communion II. hostile dominant (C-) DISCONNECTION (hostility, disaffiliation) III. hostile submissive (A+) AGENCY (mastery, power) I. friendly dominant (C+) COMMUNION (intimacy, union) IV. friendly submissive (A-) PASSIVITY (vulnerable, submission) Beyond the speculative figures of Birtchnell and Bakan, Timothy Leary (1957) was one of the first to actually operationalize the \"interpersonal circle\" - which would later go through a number of subsequent revisions (Benjamin, 1974; Carson, 1969; Kiesler, 1983). There is considerable agreement among interpersonal theorists that the best structural model for representing personality dispositions and interactions is a two-dimensional circumplex in which variables are distributed continuously around the orthogonal dimensions relabelled dominance versus submission (DOM) in place of AGENCY, and nurturance versus hostility (NUR) in place of COMMUNION (Carson, 1969; Kiesler, 1982; Leary, 1957; Sullivan, 1953). Employing both rational and empirical approaches, Wiggins (1979) classified 817 of Goldberg's (1978) 1710 adjective traits into 16 interpersonal clusters theoretically suggested by the circumplex model. Interpersonal adjectives were those determined to describe: \"dyadic interactions that have relatively clear-cut social (status) and emotional (love) 26 consequences for both participants (self and other)\" (p. 398). The Wiggins model has been validated and employed across numerous studies (Pincus & Gurtman, 1995). Although several other broad structural models of personality have been proposed (e.g., Eysenck, 1994; Tellegen, 1985), the interpersonal focus of the current study fits well with the Interpersonal Circumplex (Leary, 1957; Wiggins, 1979), a model which has additionally shown convergence with the Five-Factor Model (McCrae, 1989). Wiggins and his colleagues, as well as McCrae and Costa (1989), have demonstrated that two of the more robust factors of the Five-Factor Model, Extraversion and Agreeableness, but for a modest transposition of axes, are isomorphic with the Dominance and Nurturance dimensions, respectively. What divides the two most broadly employed models, the Five-Factor Model and the Interpersonal Circumplex, is their understanding of the additional three factors (i.e., Openness, Conscientiousness, and Neuroticism) regarded as primary to the former. In the Five-Factor Model they represent three additional factors, equal in their own right as factors alongside the other two. From the perspective of the Interpersonal Circumplex, these additional three factors are considered as imparting an additional \"coloration\" or character upon the more central circumplex \"dimensions\" of dominance and nurturance (Trapnell & Wiggins, 1990). However the conceptual relations among the five super-factors are understood, their uncovering represents significant conceptual and empirical progress in the field of personality psychology. Whereas the 1980s saw a tremendous disavowal of even appearing to be a \"trait theorist\" - regarded \"like witches of 300 years ago\" as Jackson and Paunonen (1980, p. 523) derisively observed - recent convergences are said to now provide \"a theoretical structure of surprising generality\" and \"a good answer to the question of personality structure\" (Digman, 1990, pp. 418, 436). The five factors seem to cover a vast conceptual space, integrating Cattell's 16 factors, Eysenck's \"big three,\" Murray's 20 needs, Guilford's temperaments, Jung's types, and Block's California Q-set (McCrae, 1989). Not only do these five factors provide a compelling framework for building personality measures 27 that are capable of representing the domain of personality broadly and systematically, but such a framework enables researchers to locate a manifold of constructs and measures within a meaningful conceptual space thereby enabling their comparison and contrast, and bringing clarification to what Adelson (1969) referred to as \"a disconcerting sprawl.\" The possession of such a model will also provide a good starting point for investigations into the biological substrate of human universals (Livesley & Jang, 1993). Numerous cross-cultural studies have to varying degrees largely supported the Five-Factor Model in at least six different languages (English, German, Japanese, Chinese, Tagalog [Filipino], and modern Hebrew). Goldberg writes: They [the five factors] suggest that those who have contributed to the English lexicon as it has evolved over time wished to know the answer to at least five types of questions about a stranger they were soon to meet: (1) Is X active and dominant or passive and submissive (Can I bully X or will X try to bully me)? (2) Is X agreeable (warm and pleasant) or disagreeable (cold and distant)? (3) Can I count on X (Is X responsible and conscientious or undependable and negligent)? (4) Is X crazy (unpredictable) or sane (stable)? (5) Is X smart or dumb (How easy will it be for me to teach X)? Are these universal questions? (1981, p. 161) Within a hierarchical arrangement the interpersonal circle may be conceived as integrating a series of circularly arranged subdimensions within a superordinate factor or interpersonal DOM\/NTJR space. The number of subdimensions - reflected in the angles of separation between the various vectors - are understood as representing meaningful differences in interpersonal behavior (e.g., Strong, Hills, Kilmartin, DeVries, Lanier, Nelson, Strickland, & Meyer, 1988). While some investigators have found even small angles of separation to be meaningful (Benjamin, 1974; Kiesler, 1996), most investigations have opted for a circumplex division into octants (i.e., eighths). Figure 3 shows how various interpersonal traits are distributed around the circumplex axes of dominance (DOM) and nurturance (NUR) in Wiggins' (1995) Interpersonal Adjectives Scales; angular locations for each are given with the right pole of the x-axis (NUR) representing 0\u00b0. The circumplex approach describes healthy interpersonal functioning, including the ability to care, as marked by the flexible enactment of any of the trait descriptors when the 28 Figure 3 Items and Circular Locations for the Interpersonal Adjectives Scales (DOM) Assured-Dominant (PA) 95* Firm (.269) 107* Dominant (.423) 108* Forceful (.377) 114' Domineering (.342) Arrogant-Calculating (BC) 130* Cocky (.254) 136* Crafty (.393) 137* Cunning (.456) 140* Boastful (.200) 142* WDy (.385) 143\" Calculating (.205) 144' Tricky (.427) 146' Sly (.452) Cold-Hearted (OE) 159' Ruthless (.291) 175' Ironhearted (.327) 179' Hardhearted (.364) 179* Uncharitable (.286) 180* Cruel (.292) ) 186' Cold-hearted (.358) 187* Unsympathetic (.369) 196' Warmthless (.373) Aloof-Introverted (FG) 215* Uncheery (.358) 215\" Unneighborly (.336) 217* Distant (.337) 218* Dissocial (.429) 2 2 1 ' Unsociable (.452) 222* Antisocial (.389) 229* Unspariding (.352) 242' Introverted (.406) 89* Persistent (.171) 87* Assertive (.443) 84 ' Self-confident (.398) 8 1 ' Self-assured (.391) (NO) Gregarious-Extraverted 66' Extraverted (.460) 57' Outgoing (.570) 50\" Enthusiastic (.357) 45' Perky (.328) 40' Jovial (.303) 34* Neighborly (.308) 33 ' FriefxSy(.471) 30' Cheerful (.414) (LM) Warm-Agreeable 9* Kind (.229) 6* Sympathetic (.310) 6* Tender (.368) 4* Charitable (.290) 360' Tenderhearted (.376) 360' Gendehearted (.390) 356\" Accommodating (.182) 349' Soft-hearted (.337) (JK) Unassuming-Ingenuous 321 ' Unsly(.319) 318* Uncunning (.348) 316' Unwily(.279) 313* BoasUess(.171) 311' Uncrafty (.306) 305' Uncalculating (.213) 295' Undemanding (.200) 294* Unargumentative (.198) (NUR) (HI) Unassured-Submissive 255' Timid (.330) 283' Unaggressive (.399) 256' Bashful (.305) 279\" Untold (.314) 258* Shy (.373) 276' Unauthoritative (.321) 260' Meek (.306) 272' Forceless (.314) Note. From Interpersonal Adjectives Scales: Professional Manual, by J. Wiggins, 1995, p. 23. Copyright 1995 by Psychological Assessment Resources. context deems them appropriate; neither chaotic nor rigid. Well-being is characterized by a moderate amount or \"intensity\" of any given behavior. Intensity is operationalized as the distance (i.e., vector length) from the center of the circumplex. Lorna Benjamin adds a 29 qualitative dimension in preferring the Western cultural ideal of the friendly-dominant quadrant (upper right): \"behaviors that are friendly and moderately enmeshed or moderately differentiated are normal.... and [Quadrant I] represents an optimal position in relation to others\" (1996, p. 253; see Figure 2). While hostile behaviors can be normal in particular contexts, normality in the ideal or prescriptive sense would be located in the Quadrant I. So Bakan (1966) believed that if gone unmitigated by a positive sense of communality, a strong sense of agency would be harmful to the individual and society. Similarly, the desirable aspects of agency must mitigate against an overpowering sense of communality if the individual is to function successfully. Within the interpersonal circumplex, the first quadrant (0\u00b0-90\u00b0) represents the friendly-dominant region and is comprised of a \"blend\" of agency and communion. Prescriptivity is never value-free, nor need \"adaptiveness\" be taken as the strict ideal. Empirical support for the hypothesis that flexible responding is highly functional\/ healthy (i.e., behavioral flexibility) is far from conclusive (Paulhus & Martin, 1988). In addition to accommodating both the descriptive and morally prescriptive use of the circumplex, as a superordinate semantic viewpoint it also has the potential to synthesize previously conflicting constructs. Gilligan has been criticized for contrasting connectedness and autonomy: \"the message that readers often take from this view is that autonomy has no meaning for women - is somehow beneath them, beyond them, or unnatural to them\" (Berlin & Johnson, 1989, p. 79). Rather than viewing autonomy and connectedness as bipolar dimensions marking men and women, respectively, Berlin and Johnson argue that autonomy is necessarily a part of mature relatedness, that it is precisely the mix of autonomy and warm bondedness that transforms resentful self-sacrifice and submission into generous, attuned, and mutual interaction, (p. 79; italics in original) These authors also consider the first quadrant as representing ideal healthy functioning. Gilligan has written about the importance of women bringing their needs and abilities into relationships but has avoided the word autonomy which she tends to equate with disconnection. 30 Of the two most widely used structural models of personality, the Five-Factor Model and the Interpersonal Circumplex, the latter with its orthogonal dimensions of dominance and nurturance seems ideally suited to the detailed investigation of the conventional care constructs. The present study goes beyond circumplex analyses with the inclusion of the three additional super-factors, Conscientiousness, Neuroticism, and Openness; understanding, however, the two interpersonal dimensions as most salient. In addition to providing a parsimonious description of interpersonal behaviors and a moral prescription of what interpersonal behaviors are deemed desirable, investigators have also begun to suggest an important prescriptive role for the circumplex in test development. This study will employ the interpersonal circle in both these aspects, discussing the latter in more detail in the following chapter. Clinical Applications Gilligan's seminal efforts were intended as a means of better understanding the experiences of women, who according to Gilligan had not been adequately represented in the psychological literature; especially experiences which might provide clarity to problems which uniquely or predominantly affect women, such as abortion and eating disorders. Bruch's (1973) seminal work in the field of eating disorders conceptualizes anorexia nervosa as the \"desperate struggle for a self-respecting identity\" (p. 250). Sounding very much like the care construct of external self-perception (Jack, 1991), for example, Bruch describes eating disordered women as relying upon the ideas of others to explain their own behavior. Alienated from their own experience, they are constantly: \"double-tracking ... pursuing one's own thought while trying simultaneously to figure out parental motives and reactions and continually monitoring others' responses\" (p. 62). The dissociated voices of eating disordered women sound not dissimilar to women Gilligan reports interviewing. One of Bruch's clients said, \"I really feel I am not myself and this is really sick, this not being myself, not my body or person, not really a human being\" (p. 71). Bruch argued that the basic misperceptions of these women involves a fundamental 31 alienation from self and others. As one woman stated: \"I have this unbelievable fear of people not liking me\" together with its counterpart, \"I'm not worthy of their liking me\" (p. 95). One of the links among the studies of gender, identity formation, and the problems that seem to largely afflict women, is the puzzle regarding the decline of self-esteem among young girls in early adolescence. Whereas judgments of self-worth remain largely stable for boys, some studies have shown a marked and gradual decline among girls (Harter, 1992). One of the major developmental tasks of adolescence involves the emergence of a sense of personal identity, including a sense of what it is to be a false-self or \"not real\" (Blasi & Milton, 1991; Harter & Lee, 1987). Notions of a true inner-self emerge, including a sense that one should \"be true to\" one's self. Adolescents speak of being a \"phoney,\" \"not expressing their true feelings,\" or \"playing a roler\"'There is a concomitant set of insights that emerges as to the motivational components of such behavior. The first motivational cluster views false-self behavior as an attempt to gain the approval of others, impress peers, and advance one's social relations. Another motivational cluster adolescents mention involves the attribution of others not liking them, their not liking themselves, and doubting their own true self. In one study comparing adolescents who reported high levels of false-self behavior with adolescents who did not report a need to conform for the purpose of gaining acceptance, Harter (1992) found that false-self behaviors were largely self-defeating. Rather than gaining the approval of peers, adolescents with high false-self behavior reported less peer approval and greater hopelessness about being liked or ever becoming likable. False-self adolescents also reported lower self-esteem and reported not knowing their true self. Correlational data, however, do not allow us to determine whether becoming a false-self undermines one's self-worth or whether poor self-worth and self-trust motivate an individual to adopt others' perceptions as one's own. What is striking is the relationship between self-worth and evaluations of personal 32 appearance with respect to the thinness ideal - now regarded as one of the definitive \"core\" psychological factors in eating disorders. Correspondence between the outer physical self which others see and the inner psychological self are closely tied. In study after study, at any developmental level we have examined, including older children, adolescents, college students, and adults in the world of work and family ... we have repeatedly discovered that self-evaluations in the domain of physical appearance are inextricably linked to global self-esteem. (Harter, 1992, p. 117) Correlations across the life-span remain very strong (rs = .70 to .80); self-identity has virtually become narrowed to that shallow reflection contained in the mirror. Harter (1992) finds that while ratings of perceived appearance remain stable for males from Grades 3 to 11, females' evaluations drop steadily across the same period from initially similar levels to those of the boys in Grade 3. Self-esteem, while dropping for females across the same period, does not do so as dramatically as perceived appearance. The relationship, or more specifically, the perceived direction of the relationship between physical appearance and self-esteem may be the mediating factor. Adolescent participants were asked to decide which of two orientations was most descriptive of them: \u2022 If others like or approve of me (first), then I will like and approve of myself. \u2022 If I (first) like and approve of myself, then others will like and approve of me. (Harter, 1992, p. 123) Those who chose the first orientation as self-representative - their self-esteem was an adoption of the judgments of others, or what Susan Harter calls the \"looking-glass self\" (Cooley, 1902\/1964) - reported lower levels of self-esteem. Those who chose the second option as most representative of themselves - self-acceptance would come from the self first -reported significantly higher levels of global self-esteem. Level of self-esteem has proven the only consistent predictor of treatment outcome for eating disorders, both at the end of treatment and at repeated follow-ups (Fairburn, Kirk, O'Connor, Anastasiades, & Cooper, 1987). The links among inner self, outer self, and peer perceptions can make relationships with peers intense. Seligman (1995\/1996) provides a prototypic illustration. 33 Christine jumps to conclusions when she sees her friends passing notes. She assumes that the notes say bad things about her.... When Laurie tries to talk with her, she rejects her.... In addition she refuses to look at the notes, she remains sure that the notes were about her.... If Christine continues to accuse her friends of being cruel and then withdraws, regardless of the accuracy of her accusations, her friends will retreat.... Boys, on the other hand, often get into physical fights, (p. 233) With the cognitive-developmental emergence during early adolescence of more sophisticated role-taking tasks comes the capacity for greater moral and relational elaboration. Kohlberg's (1986) conventional moral Stage 3 emerges at this time, allowing for the simultaneous and mutual conception of entitlements and obligations within relationships. But while the emergence of subsequent stages brings with them advantages, they also bring with them certain risks as well. An increase in self-consciousness and potential false-self behavior emerges. Taking up the perspective of the \"other\" - and by the \"other\" I mean to include the broader cultural conception of one's various engendered role identities - can be very self-effacing. Where those cultural standards are unrealistic, as in the thinness ideal held out for young women, the discrepancy between the actual and the ideal will result in lowered self-esteem. Further, insofar as women can be said to have a more interdependent sense of self, a culture's lack of appreciation of those attributes will likely also be encountered as repressive. Bruch's (1973) observations of women with eating disordered behaviors reflects a very similar picture of self-consciously \"double-tracking,\" judging oneself from the perspective of the other, and false-self behavior where one's sense of self-worth is made contingent upon the viewpoint of the other. Where self-worth is contingent upon the other, considerable energy will be spent in seeking others' approval. In Chapter 3 the specific application of these ideas to eating disorders will be picked up in more detail. Summary and Overview The construct of conventional care is best understood at the intersect of both terms, and more broadly reflective of what has been called the Kohlberg-Gilligan debate. The contextualism and emphasis on care in relationships which Gilligan stresses was already present in the wider philosophical debate raging at the time. But what distinguishes her work 34 from the moral philosophy that preceded it is her linking of the contextual (or personalist) ethic, focusing on empathy and care within relationships largely, with the psychological emphasis on an interdependent self. Her final move is in associating the interdependent self and care with gender. Evidence was reviewed supporting the gender-relational self link as well as the gender-care orientation link. Recent research, while confirming these associations, has also pushed the question back further opening up the explanation of gender-differences to other potential factors beyond a strictly developmental same-sex identification paradigm. Continuing efforts to operationalize Gilligan's constructs have taken place outside of a process of more rigorous psychometric evaluation and have been thus criticized for being insular. The empirical component of the current research seeks to locate itself within a larger framework where the superordinate dimensions of agency versus communion, directiveness versus attachment, or dominance versus nurturance, are viewed as providing fundamental anchors not only of personality, but also of human development (Blatt, 1990; Kegan, 1982). The largely independent care, attachment, and dependency literatures suggest a variety of subordinate constructs that are in need of construct examination in relation to one another, over a third superordinate set of terms which the Interpersonal Circumplex is here suggested as providing. As an overview framing the considerable detail that is to follow, largely in Chapter 2, five broad goals or tasks will be delineated. The first task then, involves an exploration of the construct of conventional care, Gilligan's Level 2, as it relates to the construct of dependency (and thus submissiveness). In that dependency has come to be viewed as multifaceted- Pincus and Gurtman's (1995) own exploration was published under the title \"The Three Faces of Dependency\" expressed within three octants of circular semantic space -some such similar division is anticipated for the construct of conventional care. (The term \"circle\" or \"circular\" will be used synonymously with \"circumplex,\" referring to the semantic space created by the two axes of dominance and nurturance.) Indeed the parallel tracks of 35 conventional care and dependency are likely to cross paths at various points, with the hope of mutually informing one another. This is a move of considerable importance given the frequent critique of insularity within the care literature. The second global task of the present study involves a rather detailed exploration and assessment following the proposal of circular structure test criteria to be used in conjunction with traditional criteria in scale development. While such criteria have been employed in not dissimilar ways and even suggested to test developers, none have thus far been proposed in conjunction with traditional substantive approaches, and none prior to the current study been employed in the actual construction of test measures. By these criteria several scales, entitled the Conventional Care Scales, will be developed and assessed together with related care and dependency scales which have already been introduced. The current research involves a continuation of the mapping process, with a particular eye to assessing existing constructs and further developing new ones inherent within the domain of care - or more specifically conventional care where \"the 'good' woman becomes caring by becoming selfless\" (Gilligan, 1998, p. 138). In addition to extending the number of facets subordinated within the traited or self-directed dimension, two further dimensions of conventional care will be examined: other-directed and socially prescribed conventional care. These three dimensions are those frequently employed within interpersonal theory and clinical practice (Benjamin, 1974, 1996; Sullivan, 1953). What distinguishes the three dimensions are the different objects towards which expectations of care are directed: towards one's self (self-directed), towards others (other-directed), and the perception of care expectations from others (socially prescribed). Each dimension is expected to have unique correlates. Third, in addition to the examination of conventional care in relation to dependency constructs, further construct validation and exploration will be undertaken examining conventional care in relation to measures of well-being, interpersonal capacities, relationship satisfaction, and the false-self. The first three goals will be undertaken in a lengthy first 36 study reported in Chapter 2. A second study is intended to confirm the factor structure of the scales developed in Study 1. Fourth, and in line with the recent shift to the examination of real-life over abstract hypothetical dilemmas (Walker & Hennig, 1997), this dissertation seeks also to locate itself within an applied behavioral context examining the role of conventional care in predicting problems with which more women are identified, eating disorders. The role of conventional care, which has not previously been studied in the eating disordered population outside of this data set, is examined against perfectionism, for which there is a large literature demonstrating its association with eating disorders. One focal area of developmental concern involves the large number of young women whose sense of self-worth is constituted largely on the perception they have of their own external appearance, particularly weight and shape. Judgments of self-worth become the more punitive as culture itself extols an increasingly unrealistic ideal of female beauty and thinness. One aspect of this is the press towards, and the meanings that get attached to, an unrealistically thin female body. Dieting has become virtually normative in Western culture with women reporting highly negative self-views largely based on weight and shape. Identity development has long been viewed as the core personality issue involved in eating disorders, facilitating the internalization of our culture's fixation on thinness. Deeply implicated is the possession of a false-self expressive of the belief that for one to like and accept oneself, others have to like and accept them first. Gilligan's discussions of Level 2 conventional care were intended to address precisely these sorts of issues. Finally, having already begun with the philosophical discussion above, some further reflection will be undertaken in the final General Discussion chapter to spell out the findings more philosophically and with an eye to some description of a mature ethic of care. 37 C H A P T E R 2: M A P P I N G A N D S C A L E D E V E L O P M E N T STUDIES Introduction The first major goal of Study 1 was to provide a structural clarification and examination of the conventional care construct for which Jack's Silencing the Self Scale (1991) was taken as illustrative in the previous chapter. The circumplex is ideally suited to these ends, given its emphasis upon interpersonal functioning and its empirical success in simplifying a manifold of data into precisely these two orthogonal dimensions of interest (i.e., dominance and nurturance). As a second goal, the process of mapping was continued through the development of additional scales, entitled the Conventional Care Scales (CCS). It was anticipated that the domain of conventional care would occupy a similar region of semantic space as dependency, within Quadrant IV (Friendly-Submissive; see Figure 2 above for a structural representation of agency and communion, otherwise dominance and nurturance). The present study also extends analyses beyond previous self-directed scales to include other interpersonal dimensions hitherto unexamined. The Conventional Care Scales were distilled from a large pool of items reflecting each of three dimensions: self-directed, other-directed, and socially prescribed conventional care. These three dimensions are those frequently employed within interpersonal theory and clinical practice (e.g., Benjamin, 1974, 1996: introject, self-focus, other-focus; Sullivan, 1953). What distinguishes the three dimensions are the different objects towards which expectations of care are directed: towards one's self (self-directed), towards others (other-directed), and the perception of care expectations from others (socially prescribed, see Figure 4 below). The hypothesis that conventional care arises in response to expectations from others to be nice, demure, and silent (i.e., socially prescribed conventional care) is one for which there currently exist no measures. It has only been assumed that the self-directed posture of conventional care arises as the introjected mirror of others. Each dimension was expected to have unique correlates. The notion of mapping, however, was here understood to involve self-directed 38 interpersonal constructs only. Whereas other dimensions can be examined within the Interpersonal Circumplex, the meaning of their projection changes, as is yet to be clarified. Within the self-directed dimension, several constructs added to or clarified current conceptions of the domain of care: anxious concern, self-sacrificial care, and perfectionistic care. The relations among these scales were examined along with other conventional care-like scales, including Jack's Silencing the Self Scale and dependency, employing the Interpersonal Circumplex in its descriptive use as a nomological net. Additionally, a principal components analysis was undertaken to examine the substantive relations among the scale items in an effort to determine the extent of overlap\/redundancy among constructs. Such a descriptive clarification was expected to be useful in improving personality and clinical assessment, evaluating the theoretical models behind measures such as the Silencing the Self Scale, and generating linkages to broader conceptual domains such as dependency and attachment constructs. Criteria used for the final selection of Conventional Care Scale items were based on a combination of the more usual substantive criteria (item-total correlations and factor loadings, etc.; Jackson, 1970) and a number of proposed circular structure criteria. Empirical evidence for the use of circular criteria (i.e., alpha-contribution as a function of angular proximity to a scale's central tendency) was examined in the context of the scales used in this study. In short, both substantive (i.e., thematic content) and circular structure criteria were used, both to develop scales as well as to investigate their final relations. Further discussion of structural versus substantive criteria follows. Owing to these additional efforts in test construction, the Conventional Care Scales proved superior to most of the already existing scales in a number of respects. Evidence will be examined suggesting that the two primary facets of conventional care are Octants HI (Unassured-Submissive) and JK (Unassuming-Ingenuous). While an insufficient number of items existed to test final scales, some speculation about the inclusion of a Selfishness and an Intrusive Care scale, reflecting Octants F G (Aloof-Introverted) and NO (Gregarious-Extraverted), respectively (to be 39 described in detail later), among the Conventional Care Scales, will also be examined. The third goal of Study 1 was to extend the process of construct validation across the other three super-factors (i.e., Conscientiousness, Neuroticism, and Openness), as well as to examine their relation with interpersonal capabilities, relationship satisfaction, the false-self, and well-being (i.e., depression and self-esteem). Study 2 replicated the factor structure of the Conventional Care Scales in a second sample. Finally, a third study (reported in a subsequent chapter) involved a secondary analysis of a previous study and examined the comparative predictive validity of conventional care and perfectionism in an eating disordered sample. The preliminary steps involved in the development of the final Conventional Care Scales will first be discussed before examining the final results of Study 1. Construct Explication Stating more specifically what was said in Chapter 1, care, as investigated in the present study, is defined from a \"conventional\" perspective and fundamentally involves an imbalanced focus on norms as they are instantiated within the complementarity of social roles. Specific measures were developed, expressing the constructs of: anxious concern over being uncaring to others; self-sacrificial care, where the needs of others are considered more important than one's own; and striving to be perfectly caring. Additional preliminary constructs and respective items were identified, suggesting future research directions. While it was argued that many currently existing scales in the care literature have not paid sufficient attention to the dependency literature, the dependency literature would be well served by examining the tacit normative matrix in which relationships operate. While care is a strong social ideal involving concern for others, the qualification of \"conventional\" is meant to describe both an aspect of intensity (e.g., over\/y-nurturant or compliant) and a qualitative interpersonal aspect in reflecting (largely) the discrete friendly-submissive (lower right; see Figure 1 or 2 above) Quadrant IV of the circumplex; intensity\/flexibility and circumplex location being the two basic indicators of healthy functioning mentioned in Chapter 1. 40 Provisional Item Development After defining the construct of interest, the next step involved the creation and editing of a pool of items comprising the set from which the final Conventional Care Scales were developed. A provisional set of 137 items was created by item-writers and the author based on a general description of (excessive) care and a review of constructs in the literature. Item writers were composed of a handful of colleagues, lay persons, counsellors, and clinical psychologists. Each was given a sheet of lined paper headed by the construct definition and asked to write at least five items for each of the three dimensions: self-directed, other-directed, and socially prescribed conventional care. A considerably larger sample of self-directed items was included to allow for sufficient items for the series of scales nested within the self-directed dimension. Editing involved the removal of redundant items, correcting for clarity, and rephrasing where necessary to ensure a sufficient number of negative instances. Potential scale items were identified by their thematic content and grouped a priori from among the self-directed items, forming three distinct conventional care scales: Anxious Concern, Self-sacrificial Care, and Perfectionistic Care (see Figure 4). Method Participants Study 1. Participants were 110 male and 192 female undergraduate students drawn from the University of British Columbia, Department of Psychology's subject pool (M = 19.3 years, SD = 1.98). The ethnic composition of the sample was 61.3% Asian (including Chinese, East Indian, Korean, Japanese, Filipino, Vietnamese, Malaysian, and Polynesian), 34.8% White, 2.6% Arabic, and 1.3% unspecified. The gender and ethnic composition of the sample is typical of the department of the university where data were collected. A l l participants received course credit for their involvement. Test administration involved the completion of a questionnaire package composed of twelve randomly ordered self-report measures in addition to a general demographics form. 41 Figure 4 Three Dimensions of Conventional Care and their Respective Scales Perfectionistic Care Self-sacrificial Anxious Care Concern ti ti ti Questionnaire packages were administered in groups, required approximately 75 minutes to complete, and contained the measures described below (see Appendix B). Study 2. The sample of 302 participants in the factor-replicating study was composed of 90 male and 208 female undergraduate students (and four others who failed to indicate their gender and age) drawn from the University of British Columbia, Department of Psychology's subject pool (M = 19.7 years, SD = 2.54). The ethnic composition of the sample was 56.6% Asian, 37.7% White, 1.3% Black, .7% Arabic, .3% each of Hispanic and North American Indian, and 3.0% unspecified. Participants completed a questionnaire 42 containing 87 conventional care items, which required approximately 15 minutes (see Appendix C). Note that all the findings reported below are based on the sample of Study 1, with the exception of the coefficient of congruence statistic reporting the replicability of the factor structure across the samples of both Study 1 and 2. Demographic Information Participants were asked to provide information regarding their age, gender, ethnicity, length of Canadian residency if born outside Canada, year in university, academic major, and average grade (e.g., A+, A , A- , etc.). Information was also requested regarding relationships. If the respondent was currently in an intimate relationship, they were asked to provide its current length and indicate on a 5-point Likert scale (1 = extremely dissatisfied; 3 = neutral; 5 = extremely dissatisfied) their satisfaction and their perception of their partner's satisfaction with the relationship. Factor Structure Measures Revised Interpersonal Adjectives Scales - Big Five (IASR-B5; Wiggins, 1995). The IASR-B5 is an extended version of the Revised Interpersonal Adjective Scales (IASR; Wiggins, Trapnell, & Phillips, 1988). The IASR consists of 64 trait-descriptive adjectives (e.g., \"dominant\") assessing the eight octants of the Wiggins (1979) circumplex that together are expressive of the coordinates of Dominance and Nurturance. The eight subscales, each reflecting one of the octants, are Assured-Dominant (PA), Arrogant-Calculating (BC), Cold-hearted (DE), Aloof-Introverted (FG), Unassured-Submissive (HI), Unassuming-Ingenuous (JK), Warm-Agreeable (LM), and Gregarious-Extraverted (NO); located relative to the positive pole of the Nurturance axis at 90\u00b0, 135\u00b0, 180\u00b0, 225\u00b0, 270\u00b0, 315\u00b0, 0\u00b0, and 45\u00b0, respectively. The IASR-B5 supplements the 64-item measure with the inclusion of 20-item markers for each of the three Five-Factor Model factors of Conscientiousness, Neuroticism, and Openness. Wiggins reports alpha coefficients in a large self-report sample of undergraduates 43 ranging from .73 to .86 across subscales. The IASR has been shown to have acceptable stability and validity (Wiggins et al., 1988); and has been used in a variety of studies with both normal and clinical populations (Wiggins, 1995). The IASR is generally viewed as the standard for ideal circumplex criteria. Using an 8-point Likert scale from 1 (Extremely Inaccurate) to 8 (Extremely Accurate), respondents rate each of 124 adjectives for self-descriptive accuracy. An individual circumplex profile can be constructed according to their T-score for each of the eight circumplex scales and, by taking the D O M and NUR factor scores as x and y Cartesian coordinates, the mean length and directionality can also be plotted onto circumplex space. Individuals can then be investigated according to overall profile and\/or octant category. Additionally, a score for each of the other three Five-Factor Model factors -Conscientiousness, Neuroticism, and Openness - can be derived. Circumplex Inventory of Interpersonal Problems (IIP-C; Alden, Wiggins, & Pincus, 1990). The IIP-C is a 64-item measure of a variety of interpersonal problems that people report experiencing and which fit around a circumplex about the axes of dominance (DOM) and nurturance (NUR). The longer 127-item Inventory of Interpersonal Problems (IIP; Horowitz, Rosenberg, Baer, Ureno, & Villsefior, 1988) was revised to circumplex criteria by Alden et al. (IIP-C) and has been demonstrated to structurally converge to the Revised Interpersonal Adjective Scale (IASR; Wiggins et al., 1988). The IIP-C is divided into two sections according to the type of items included: \"things you find it hard to do with other people\" and \"things that you do too much.\" Participants rate each of the 64 statements on a 5-point Likert scale from 0 (Not at all) to 4 (Extremely), yielding eight octant raw-scores varying from 0 to 32. Each octant is represented by four items for each of the two types, for a total of eight items per octant. The eight subscales, each reflecting one of the octants, are Domineering (PA), Vindictive (BC), Cold (DE), Socially Avoidant (FG), Nonassertive (HI), Exploitable (JK), Overly Nurturant (LM), and Intrusive (NO); located relative to the positive pole of the x or NUR axis at 90\u00b0, 135\u00b0, 180\u00b0, 225\u00b0, 270\u00b0, 315\u00b0, 0\u00b0, and 45\u00b0, respectively. 44 The authors report alpha coefficients across the eight scales ranging from .72 to .85. The IIP has demonstrated test-retest reliabilities across a 10-week interval ranging from .80 to .90, and amongst a clinical sample demonstrated sensitivity to change as a result of psychotherapy, and was meaningfully related to other concomitant measures of psychopathology (Horowitz et al., 1988). The IIP-C has been used by a number of investigators to examine individual differences in interpersonal style and has been shown to have good structural convergence across self- and peer-ratings (e.g., Wagner, Kiesler, & Schmidt, 1995). Dependency and Conventional Care Measures Irrational Beliefs Test - Demand for Approval Subscale (IBT; Jones, 1969). The IBT - Demand for Approval Subscale measures the extent to which individuals assert 10 of the core irrational beliefs described by Albert Ellis, the originator of Rational Emotive Therapy. Factor analyses have confirmed the structure of the total measure (Lohr & Bonge, 1981) and test-retest reliability over an 8-week period has demonstrated considerable temporal consistency. In the present study, the alpha coefficient for this scale was .78. The measure has been shown to predict a number of clinical problems including depression (Nelson, 1977) and anxiety (Lohr & Bonge, 1981). Nelson (1977) found gender differences for the relationship between the Demand For Approval Subscale and depression, in favor of women showing the stronger relationship (z - 2.55, p < .05). The current study used only the 12-item Demand for Approval Subscale (e.g., \"I want everyone to like me\") which correlates .80 with the total scale. Participants rated their item agreement on a 5-point Likert scale from 0 (Disagree) to 4 (Agree), yielding a range from 0 to 60. High values indicate a high need for approval from others. Conventional Care Scales (CCS). Participants rated each of 137 provisional items on a 7-point Likert scale from 1 (Strongly Disagree) to 7 (Strongly Agree). A combination of circular and traditional criteria (Jackson, 1970) was used to determine final item inclusion in their respective theoretically derived scales. Items from the provisional pool were dropped 45 if: (a) they were highly skewed (i.e., less than 7% of endorsements were located in three adjacent rating points on the 7-point scale), (b) they were undiscriminating (i.e., standard deviation less than 1), (c) their corrected item-total correlations were less than .40 on their respective subscales as defined a priori by thematic content, and (d) in a final principal components factor analysis, they loaded less than .40 on their respective factor, and did not load above .40 on more than one factor. Criteria for item inclusion within the three self-directed Conventional Care Scales involved additional circular criteria to establish the circular meaningfulness of the construct. Items were dropped if their angular projections were outside their target octant (i.e., octant midpoint \u00b1 30\u00b0). A slightly broader range beyond the bounds of the octant was used so as to maintain a balance between constructs that were octant-defining, taking the octant as next superordinate level, and not unduly constraining the item range for those constructs whose circular mean was nearer the octant boundary than the center point. Those eight items with the highest factor loadings were finally retained. The circular factor loading was defined as (see Equation 1 below for more detail): V. * cos(0, - <\/>) (1) where