The Mismeasure of Love: How Self-Doubt Contaminates Relationship Beliefs

http://psp.sagepub.com/ Bulletin Personality and Social Psychology http://psp.sagepub.com/content/27/4/423 The online version of this article can be found at: DOI: 10.1177/0146167201274004 2001 27: 423Pers Soc Psychol Bull Sandra L. Murray, John G. Holmes, Dale W. Griffin, Gina Bellavia and Paul Rose The Mismeasure of Love: How Self-Doubt Contaminates Relationship Beliefs Published by: http://www.sagepublications.com On behalf of: Society for Personality and Social Psychology can be found at:Personality and Social Psychology BulletinAdditional services and information for http://psp.sagepub.com/cgi/alertsEmail Alerts: http://psp.sagepub.com/subscriptionsSubscriptions: http://www.sagepub.com/journalsReprints.navReprints: http://www.sagepub.com/journalsPermissions.navPermissions: http://psp.sagepub.com/content/27/4/423.refs.htmlCitations: What is This? - Apr 1, 2001Version of Record >> at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ http://psp.sagepub.com/content/27/4/423 http://www.sagepublications.com http://www.spsp.org/ http://psp.sagepub.com/cgi/alerts http://psp.sagepub.com/subscriptions http://www.sagepub.com/journalsReprints.nav http://www.sagepub.com/journalsPermissions.nav http://psp.sagepub.com/content/27/4/423.refs.html http://psp.sagepub.com/content/27/4/423.full.pdf http://online.sagepub.com/site/sphelp/vorhelp.xhtml http://psp.sagepub.com/ http://psp.sagepub.com/ PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN Murray et al. / SELF-DOUBT AND RELATIONSHIP BELIEFS The Mismeasure of Love: How Self-Doubt Contaminates Relationship Beliefs Sandra L. Murray State University of New York at Buffalo John G. Holmes University of Waterloo Dale W. Griffin University of British Columbia Gina Bellavia Paul Rose State University of New York at Buffalo The authors argue that individuals with more negative models of self are involved in less satisfying relationships because they have difficulty believing that they are loved by good partners. Dating and married couples completed measures of self-models, perceptions of the partner’s love, perceptions of the partner, and relationship well-being. The results revealed that individuals troubled by self-doubt underestimated the strength of their part- ners’ love. Such unwarranted insecurities predicted less positive perceptions of their partners. In conjunction, feeling less loved by a less-valuable partner predicted less satisfaction and less opti- mism for the future than the partner’s feelings of love and com- mitment warranted. A dependency regulation model is described, where feeling loved by a good, responsive partner is thought to represent a sense of felt security that diminishes the risks of interdependence and promotes closeness. Of all the forms of caution, caution in love is perhaps the most fatal to true happiness. —Bertrand Russell, The Conquest of Happiness Why are individuals troubled by self-doubt typically involved in less satisfying and less stable relationships than individuals with a stronger sense of self-worth (e.g., Fincham & Bradbury, 1993; Hendrick, Hendrick, & Adler, 1988; Karney & Bradbury, 1997; Kelly & Conley, 1987; Murray, Holmes, & Griffin, 1996a, 1996b)? Rus- sell’s lament may provide the key. His words point to a curious tension in romantic rela- tionships. On one hand, happiness seems to involve throwing caution somewhat to the wind and making inferential leaps of faith. For instance, dating and mar- ried individuals are happier in their relationships when they see virtues in their partners that are not apparent to others (Murray, Holmes, Dolderman, & Griffin, in press) or to their partners themselves (Murray et al., 1996a). They are also happier when they optimistically believe that the future of their own relationship is rosier and more secure than the future of most other relation- ships (Murray & Holmes, 1997). On the other hand, the dependency and vulnerability inherent in romantic relationships make some degree of caution or circumspection a necessity. After all, there is perhaps no other adult context where the possibility of another’s rejection is more self-threatening and the pos- sibility of acceptance more self-affirming (Hazan & Shaver, 1994). In the face of such stakes, individuals need a sense of felt security based in the belief that if they allow themselves to fully trust in their partners and rela- tionships, they will not face eventual disappointment. 423 Authors’ Note: We are grateful for the assistance of Lisa Allard, Chris- tine Celnar, Dan Dolderman, Mary Dooley, and Rebecca Gulliford in the conduct of this research. This research was prepared with the sup- port of a Social Sciences and Humanities Research Council of Canada research grant to John Holmes and Sandra Murray and a National In- stitute of Mental Health and National Science Foundation grant to Sandra Murray. Correspondence concerning this article should be ad- dressed to Sandra Murray, Psychology Department, Park Hall, State University of New York, Buffalo, NY 14260-4110; e-mail: smurray@ acsu.buffalo.edu. PSPB, Vol. 27 No. 4, April 2001 423-436 © 2001 by the Society for Personality and Social Psychology, Inc. at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ In this article, we argue that individuals troubled by self-doubt are involved in less satisfying relationships because they are excessively cautious, unable to find suf- ficient reason for felt security in even the most accom- modating behavioral realities. Figure 1 presents a model illustrating this dynamic. It reflects our contention that a sense of felt security that tempers the need for caution requires sustaining two beliefs: (a) that the partner loves the self and thus is willing to be available and caring and (b) that the partner is a good, responsive person and thus is capable of fulfilling one’s needs. In the remainder of this introduction, we justify this working definition of felt security and then describe why more negative mod- els of self might make these conjunctive requirements for felt security difficult to achieve. THE REPRESENTATIONAL BASIS OF FELT SECURITY Attachment theorists believe that the capacity to find a sense of felt security in close attachments develops early in childhood in the context of actual interactions with a primary attachment figure (e.g., Bartholomew, 1990; Bowlby, 1982; Collins & Read, 1994). Through interactions with consistently available and responsive caregivers, children are thought to develop generalized models of themselves as worthy of love and models of others as generally caring and responsive. In contrast, interactions with unavailable or inconsistently respon- sive caregivers are believed to foster uncertainty in one or both models (see Hazan & Shaver, 1994, for a review). From an attachment framework, then, a predisposition to experience felt security, conceptualized as a secure attachment style, is associated with working models that support positive inferences about both the disposi- tions of others and one’s own worthiness of love (Barth- olomew, 1990). What are the implications of this conceptualization for the actual experience of felt security in a specific romantic relationship? We believe that felt security is rooted in the beliefs that a good, responsive partner loves and is committed to the self. This definition comes from our translation of the beliefs and feelings thought to underlie general working models of self and other into their relationship-specific counterparts. Feeling loved by a particular partner has conceptual parallels to general feelings of worthiness of love (models of self), whereas perceiving a good, responsive partner has con- ceptual parallels to general beliefs about the trustworthi- ness of others (models of other). Together, these related but logically distinguishable expectations may support positive inferences about the likelihood of a specific romantic partner being both responsive and available, now and in the future (e.g., Holmes & Rempel, 1989). Why should these two beliefs in particular be critical? Our emphasis on the importance of feeling loved is con- sistent with recent arguments that the need for interper- sonal acceptance is a fundamental human motivation (Baumeister & Leary, 1995; Leary, Tambor, Terdal, & Downs, 1995). In romantic relationships, the motivating influence of belongingness needs is particularly evident because perceptions of a romantic partner’s attraction to the self seem to be a more potent force in triggering attraction and love than considerations of this person’s qualities per se (Aron, Dutton, Aron, & Iverson, 1989; Hazan & Diamond, in press). To experience felt security, however, individuals also may need to believe that their loving partner is not only willing but able to meet their needs (Clark & Reis, 1988). Such inferences may be based in part on observations of a partner’s past responsiveness to needs in specific cir- cumstances (Wieselquist, Rusbult, Foster, & Agnew, 1999). However, confident expectations about future responsiveness also may necessitate going beyond the behavioral evidence (Murray et al., 1996a). Most people think in lay dispositionalist terms, believing that behav- iors require the existence of certain underlying traits to occur (Gilbert, 1998). As a result, perceiving a good, responsive partner, someone who is kind but not critical and warm but not distant, may foster confident predic- tions about a partner’s behavior in situations requiring responsiveness to needs, now and in the future (Holmes & Rempel, 1989; Murray et al., 1996a). Once established, a sense of felt security then may diminish the perceived risks of interdependence and temper the need for caution, thus providing the basis for relationship well-being. Accordingly, we expected dating and married individuals to report greater satisfaction in the present and greater optimism for the future when they felt more loved and perceived their partners more positively. Furthermore, being with a partner who pos- sesses this sense of felt security should have similar benefits. Models of Self and the Perception of Love Why might negative models of self make these con- junctive requirements for felt security difficult to 424 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN Figure 1 The representational basis of felt security. at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ achieve? The most obvious possibility is that low-self- esteem individuals experience relationship realities that are objectively less conducive to felt security. For instance, the self-perceived faults that trouble low self- esteem individuals may be just as apparent to their part- ners, making it difficult for even the most tolerant of partners to express unconditional love. Ironically, though, the failings that so trouble low self-esteem individuals may not be at all obvious to their partners. Low and high self-esteem individuals do not seem to differ in their status on consensually held crite- ria for worth, such as intelligence or physical attractive- ness. Instead, they differ in their subjective beliefs about their possession of such qualities (Brown, 1998). Cou- pled with this source of ambiguity, perception in roman- tic relationships is itself subjective. Satisfied individuals see virtues in their partners that are not apparent to their partners or to others (Murray et al., 1996a, 2000). Such considerations suggest that individuals may be loved regardless of their own self-perceived virtues and faults. Is such love as likely to be perceived by individuals troubled by self-doubt? Rather than accurately monitor- ing others’ reactions to the self, individuals typically act like naive realists, assuming that most others see them in the same way as they see themselves (Kenny, 1994; Shrauger & Schoeneman, 1979). In our prior research, we focused on one aspect of reflected appraisals, percep- tions of how the partner sees the self on specific interper- sonal qualities (Murray, Holmes, & Griffin, 2000). We asked dating and married couples to describe them- selves, their partners, and how they believed their part- ners saw them on a variety of interpersonally oriented virtues and faults. Individuals who described themselves more negatively on these qualities incorrectly believed that their partners also saw them negatively. In contrast, higher self-esteem individuals correctly assumed their partners saw them positively. Among low self-esteem individuals, such inaccurate but nonetheless self-verifying, perceptions need not compromise perceptions of a partner’s love. In fact, per- ceptions of a partner’s love and regard for the self on specific traits could be orthogonal. Feeling loved is thought to be largely an emotional experience, one that is not easily articulated or justified by specific reasons (Brickman, 1987). Psychological wisdom also suggests that individuals feel truly loved when they believe their partners see them as they really are, proverbial warts and all, but love them anyway (e.g., Brickman, 1987; Reis & Shaver, 1988). Moreover, low self-esteem married inti- mates report greater intimacy the more accurately their partners perceived their faults (Swann, Hixon, & De La Ronde, 1992). If such validation of the self is the key to feeling loved, then individuals with more negative mod- els of self may hold inaccurate beliefs about their part- ners’ perceptions of their specific attributes but still accurately perceive a partner’s feelings of love. Despite the romantic appeal of a partner’s love being beyond reason, we suspect that individuals believe that their partners love them at least in part because of the qualities they see in them. As a result, more negative models of self may make it harder to construct a logical basis for confident inferences about a partner’s caring (Path A in Figure 1). After all, low self-esteem individuals believe that others’ love and acceptance is contingent on their possessing valued attributes (Baldwin & Sinclair, 1996) and that their partners see as many faults in them as they see in themselves (Murray et al., 2000; Murray, Holmes, MacDonald, & Ellsworth, 1998). Thus, a more mixed and uncertain store of self-knowledge may make a partner’s love more of a mystery to lows, leaving them less certain of its existence. However, the requisite appre- ciation of a partner’s love may be a more readily attain- able goal for individuals with more positive models of self because it is so easy for them to generate reasons why their partners should care for them. A partner’s love is easily explained by the assumption that their partners see as many virtues in them as they see in themselves. The Regulation of Interpersonal Dependency Why might doubts about a partner’s love compromise perceptions of a partner’s basic worth or value, as Path B in Figure 1 illustrates? After all, there is no necessary logi- cal relation between these beliefs. In romantic relation- ships, however, the risks of interdependence are such that these perceptions are likely to be related in practice. Rather than being naively trusting, intimates are thought to regulate feelings of attachment and commit- ment in a self-protective fashion (e.g., Berscheid & Fei, 1977; Bowlby, 1982; Holmes & Rempel, 1989). In this vein, we contend that individuals only allow themselves to risk closeness (and thus dependence) when the risks of rejection are perceived to be minimal (Murray et al., 1998, 2000). In terms of the dynamics presented in Fig- ure 1, intimates may only feel safe believing that they have found that special person when they feel confident of their partners’ continued, reciprocated affections. Our prior research provided an initial test of this dependency regulation model (Murray et al., 2000). In this study, dating and married individuals rated their partners more positively on a variety of interpersonal qualities the more positively they believed their partners saw them on these same traits. Such trait-specific reflected appraisals also mediated the link between self- esteem and perceptions of the partner. This suggests that low self-esteem individuals see their partners less positively in part because they (incorrectly) believe their partners also see them in a relatively negative light. How- ever, perceptions of the partner’s regard for the self on Murray et al. / SELF-DOUBT AND RELATIONSHIP BELIEFS 425 at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ specific traits might have predicted perceptions of the partner on these same traits because evaluations tend to be reciprocated (not because concerns about rejection constrain idealization processes, as we hypothesized). More crucial, perceptions of a partner’s regard for the self on specific traits need not correspond to perceptions of a partner’s love, as we argued earlier. For these reasons, a basic but still untested assump- tion of the dependency regulation model and attach- ment theory itself is that individuals actively regulate per- ceptions of a partner’s basic worth with a sense of confidence in a partner’s love. Regardless of self-esteem, perceptions of a partner’s love may constrain attach- ment in ways that leave individuals who feel less loved defensively perceiving their partners less positively than individuals who feel more loved. By valuing their part- ners less, then, individuals who feel less loved may pro- tect themselves against the sting of rejection in advance (because they perceive less to lose). Such caution is logical and self-protective in instances where the partner’s love and commitment really is in question. Ironically, however, intimates with more nega- tive models of self may be reserving judgments about their partners’ worth in response to largely unwarranted doubts about their partners’ love. In this way, the dynam- ics of dependency regulation may then interfere with the second condition for felt security, the perception of a good, responsive partner (Path B). For individuals with more negative models of self, then, the operation of an overly sensitized sociometer and the normal operation of the dependency regulation system may make it doubly difficult for them to find sufficient reason for felt secu- rity in their relationships. In reality, however, individuals who are troubled by self-doubt may possess romantic partners who are just as loving and committed as the partners of individuals with a stronger sense of self-worth. This led us to the ironic, and disturbing, conclusion that dating and married indi- viduals with more negative models of self may be temper- ing projections for the future of their relationships in response to largely unwarranted concerns about their partners’ love and commitment. If this occurs, the rela- tionship inferences of these individuals may be miscalibrated in a further sense: Individuals with more negative models of self may be less optimistic about the future of their relationships than their partners’ level of commitment actually warrants. METHOD Participants Married sample. This sample included 105 couples involved in marital or cohabiting relationships at least 2 years in length; 77 couples were married and 28 couples were cohabiting. Their mean age was 38.5 years, and the average duration of their relationships was 10.9 years. Our participants received $20 payment and chances to win a $100 restaurant gift certificate. Dating sample. This sample included 121 dating cou- ples. Their mean age was 19.5 years and they had been dating 19.0 months on average. They received either course credit or $6 payment.1 Procedure Married sample. In recruiting the married sample, we posted street and newspaper advertisements across Kitchener-Waterloo inviting couples to participate in a study on thoughts and feelings in close relationships. At the 2-hour session, the experimenters first introduced the study to the volunteers and then gave them packets containing the questionnaires and instructions. Couples were asked to complete the measures without compar- ing their responses. The experimenters then conducted a workshop session on conflict resolution skills. Dating sample. We invited introductory psychology stu- dents who were in dating relationships to participate in a study on close relationships. If both members of the cou- ple were present in the laboratory, we placed them at sep- arate tables. If only one member of the couple attended, their partners were sent questionnaires and letters invit- ing them to participate. They received an explanation of the study and $6 payment upon receipt of their question- naires. We again cautioned them to complete the mea- sures without discussing their responses. Measures Self-esteem. Rosenberg’s (1965) 10-item measure assessed global self-evaluations (e.g., I feel that I am a person of worth, at least on an equal basis with others). Dating ( = .85) and married ( = .88) participants responded on 4- and 7-point scales, respectively (1 = strongly disagree, 4 = strongly agree). Interpersonal Qualities Scale (IQS). This 22-item mea- sure included attributes from the interpersonal circle, a model based on the dimensions of warmth-hostility and dominance-submissiveness (see Murray et al., 1996a, for further details). Example attributes included kind and affectionate, open and disclosing, patient, critical and judgmental, thoughtless, and irrational. Dating ( = .75) and married ( = .79) participants described themselves on these traits using a 9-point scale (1 = not at all character- istic, 9 = completely characteristic). We reverse-scored nega- tive traits such that higher scores represented more favorable perceptions. Attachment. Bartholomew and Horowitz’s (1991) mea- sure asked participants to rate how well four prototypic styles of attachment described their general relationship 426 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ styles (1 = not at all like me, 7 = very much like me). We cre- ated indices of the positivity of models of self (secure and dismissing minus fearful and preoccupied ratings) and models of other (secure and preoccupied minus dismiss- ing and fearful ratings) as in Griffin and Bartholomew (1994). Perceived love. This scale tapped a sense of security in the level and constancy of a partners’ love (e.g., I am completely confident that my partner loves me; Some- times I wonder whether my partner feels as strongly for me as I feel for him or her). Participants responded to these items on a 9-point scale (1 = not at all true, 9 = com- pletely true). Dating participants ( = .74) completed a 5-item version; married participants ( = .75) completed a 4-item version. Perceptions of the partner’s regard. Participants also described how they thought their partners saw them on the IQS attributes. Dating ( = .78) and married ( = .84) participants made these ratings using 9-point scales, and negative attributes were reverse-scored. Perceptions of the partner. Dating ( = .83) and married ( = .85) participants also described their partners on the IQS attributes using the 9-point scales described above. Love. This scale tapped feelings of love for the partner (e.g., I am very much in love with my partner; I feel ex- tremely attached to my partner). Participants responded to these items on a 9-point scale (1 = not at all true, 9 = com- pletely true). Dating participants ( = .85) completed a 4-item version; married participants ( = .84) completed a 3-item version. Commitment. This two-item scale tapped feelings of commitment to the relationship (i.e., I am very commit- ted to maintaining my relationship; I have made a firm promise to myself to do everything in my power to make my relationship work). Dating ( = .67) and married ( = .85) participants responded to these items on a 9-point scale (1 = not at all true, 9 = completely true). Optimism. This measure assessed people’s perceptions of the likelihood of a variety of positive (e.g., the love my partner and I share continuing to grow) and negative (e.g., my partner and I discovering areas in which our needs conflict in a serious way) events occurring in their own relationships relative to other relationships of com- parable length (Murray & Holmes, 1997). Participants responded to these items on a 9-point scale (1 = much less likely to occur, 9 = much more likely to occur). Dating partici- pants ( = .89) completed 11 items; married participants ( = .90) completed 16 items. Responses to negative events were reverse-scored. Satisfaction. This 4-item scale assessed dating ( = .87) and married participants’ ( = .94) global evaluations of their relationships (e.g., I am extremely happy with my current romantic relationship; I do not feel that my cur- rent relationship is successful). RESULTS As the first step in our analyses, we created a models of self composite by standardizing and then summing scores on the Rosenberg, interpersonal qualities, and attachment model of self indices. Higher scores on this composite reflect more positive models of self. We focused our analyses on this composite because we believed that combining intercorrelated (average = .59) but nonetheless diverse sources of self-regard best captured the scope of models of self, defined broadly as an internalized sense of one’s own self-worth and worthi- ness of acceptance. As the second step in our analyses, we examined the link between self-models and relationship perceptions (to establish that there was indeed an effect to be explained). These zero-order correlations revealed that dating and married men with more positive models of self reported greater relationship satisfaction, r(119)D = .31, p < .001, r(102)M = .36, p < .001, and greater optimism, r(119)D = .29, p < .001, r(103)M = .46, p < .001. For women, a parallel pat- tern of results emerged for both satisfaction, r(119)D = .19, p < .05, r(102)M = .39, p < .001, and optimism, r(119)D = .25, p < .01, r(103)M = .41, p < .001. In the rest of our analyses, we test the hypothesis that more negative models of self interfere with positive and optimistic relationship inferences because such self- doubts make a sense of felt security difficult to achieve. To this end, we first examine whether more negative models of self predict miscalibrated inferences about a partner’s love. Next, we construct mediational models linking self-models to perceived love (Path A in Figure 1) and perceived love to perceptions of the partner (Path B in Figure 1), and finally, we link the conjunctive sense of being loved by a good partner to positive and optimistic inferences about the relationship (Paths C and D in Fig- ure 1). We conclude by examining whether individuals with more negative models of self are unduly pessimistic (or insufficiently optimistic) about their relationships. Miscalibration in the Perception of a Partner’s Love We hypothesized that more negative models of self predict less generous and optimistic relationship infer- ences because an impoverished sense of self fosters unwarranted doubts about a partner’s love, a basic requirement for a sense of felt security. Was this the case? Before getting into the nitty-gritty of the results, we describe the logic governing the analysis that we used to test this miscalibration hypothesis (Edwards, 1994; Grif- fin, Murray, & Gonzalez, 1999). Generally, this method Murray et al. / SELF-DOUBT AND RELATIONSHIP BELIEFS 427 at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ involves treating perceived and actual love as separate outcome variables, allowing us to distinguish the expected within-subject interaction (i.e., discrepancy) effects from main effects across the outcome variables. Specifically, we first estimated simultaneous regression lines predicting two dependent variables, the perceiver’s perception of the partner’s love and the partner’s actual love, from the continuous measure of the perceiver’s models of self using the structure equation modeling (SEM) package in AMOS. The left half of Figure 2 presents these regression lines for dating men as an example, but in all cases, the patterning of the slopes and intercepts of these lines for dating and married men and women revealed the inter- action that we expected; namely, less convergence or greater miscalibration as self-models become more neg- ative. This interaction pattern comes from two aspects of these lines: First, perceived love is more tied to models of self than the partner’s actual love (differing slopes), and second, the partner’s actual love is stronger or more pos- itive than the perceiver’s beliefs about that love (differ- ing intercepts). We then tested the difference between the slopes and intercepts of the lines using SEM within AMOS. As our first step, we tested for gender differences in the size of the slopes and intercepts for the perceived and actual love regression lines within dating and married samples. We did this by first obtaining the coefficients for a model that predicted four dependent variables (men’s and women’s perceived and actual love) from the two inde- pendent variables (men’s and women’s models of self). Because the slopes and intercepts for the perceived and actual love regression lines were strikingly similar for men and women, we then fit models estimating common or pooled perceived and actual love intercepts and slopes across gender. To test the difference in the pooled slopes for the per- ceived and actual love regression lines, we then com- pared the fit of models that constrained the slopes of these regression lines to be equal to the fit of models that left these slopes free to vary. If the slopes are significantly different, then the goodness of fit for the models estimat- ing separate coefficients would be significantly better (i.e., a smaller chi-square) than the goodness of fit for the models estimating common coefficients (a 1 df test). The complementary test of the difference between pooled perceived and actual love intercepts examines whether the lines have different elevations at the point where the independent variable equals zero. This test makes little practical sense here because models of self have no natural zero point. Therefore, we scaled models of self in a way that made the test of the intercept psycho- logically and theoretically meaningful. First, we cen- tered women’s (and men’s) models of self so that zero represented women (or men) at the bottom 20th per- centile on self-regard. The comparison of the pooled perceived and actual love intercepts at this percentile examines whether intimates who possess relatively nega- tive models of self incorrectly believe that their partners love them less than they actually do. Next, we centered women’s (and men’s) models of self so that zero repre- sented women (or men) at the top 20th percentile on self-regard. The test of these pooled intercepts compares the levels of perceived and actual love for individuals with relatively positive models of self. The top half of Table 1 contains the slopes, intercepts, and chi-square tests for the lines depicted in Figure 2. For dating and married samples, the slope of the pooled perceived love regression line was significantly higher than the pooled actual love regression line. That is, the perceived strength of the partner’s love was much more a function of the perceiver’s self-models than the strength of the partner’s actual feelings of love (which was not at all contingent on self-perceived worthiness of love). Second, the intercept for the pooled actual love regression line was significantly higher than the inter- 428 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN Figure 2 Predicting perceived and actual love and the perceiver’s optimism and the partner’s commitment from the perceiver’s models of self. at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ cept for the pooled perceived love regression line for intimates at the bottom 20th percentile in self-regard in all four cases. The intercepts of these lines did not differ significantly at the top 20th percentile in self-regard for either sample. In combination, these regression lines illustrate the exact pattern of miscalibration we hypothesized: Inti- mates with more negative self-models were considerably more likely to feel that they were loved less than was actu- ally warranted by the strength of their partners’ actual sentiment. Significant patterns of miscalibration also were evident when we examined the constituent ele- ments of the model of self composite individually.2 Alternative explanations? We reasoned that such pat- terns of miscalibration exist because certain intimates have difficulty stepping outside of their own self-doubts and appreciating their partners’ potentially differing perspectives of them and their relationships. However, at least two possible alternative explanations for these pat- terns exist. The first possibility is that individuals with more nega- tive models of self underestimate their partners’ love because they are generally distrusting of others. To explore this possibility, we analyzed a further set of regression plots predicting perceived and actual love from the perceiver’s generalized model of other as assessed on the Bartholomew and Horowitz (1991) scale. Crucially, these analyses did not reveal any systematic evi- dence of miscalibration interactions in either sample. In fact, the slopes of the perceived and actual love lines were both near zero and did not differ significantly as a function of models of other. The second possibility is that we obtained the patterns simply because we were comparing the reports of one person against another. As a result, any person-specific artifacts, such as response bias or susceptibility to social desirability concerns, might inflate the relation between the perceiver’s self-models and relationship inferences as compared to the partner’s reports. If that were simply the case in the present data, we would expect to find simi- lar patterns of miscalibration regardless of the predictor variable being examined or the content of the percep- tions being compared. This was not the case. First, models of other did not predict miscalibration (even though these analyses involved comparing the perceptions of one partner to another). Second, a fur- ther set of regression plots predicting the perceiver’s and the partner’s satisfaction from the perceiver’s self-models suggest that miscalibration depends on the content of the perceptions being compared. We exam- ined perceiver and partner satisfaction in these analyses because reports of one’s own satisfaction do not involve imagining the partner’s perspective on the self, but they seem at least as susceptible to response bias and social desirability concerns as measures of perceived love. When we created these plots, we did not find any system- atic evidence of miscalibration. In fact, the slopes and intercepts of these lines did not differ significantly for dating women or married men or women. These find- ings suggest that miscalibration effects are specific to comparisons between meta-perceptions of a partner’s love and his or her actual sentiments. Satisfying Felt Security Needs: The Inferential Bases for Satisfaction and Optimism In this next section, we examine the conceptual model illustrated in Figure 1 in more detail, testing the hypothesis that perceived love predicts positive evalua- tions of the partner, and that the conjunctive experience of feeling loved by a good partner predicts greater satis- faction in the present and optimism for the future. Fig- ure 3 presents the model testing these predictions. To test this model, we used the SEM program within the AMOS procedure of SPSS. Because of its ability to test the fit of competing models, SEM allowed us to test for gender and relationship status differences in our Murray et al. / SELF-DOUBT AND RELATIONSHIP BELIEFS 429 TABLE 1: Chi-Square Tests Comparing the Slopes and Intercepts of the Perceived and Actual Love Regression Lines and the Perceiver’s Opti- mism and Partner’s Commitment Regression Lines Perceiver’s Models of Self Dating Married Criterion Slopea Bottom 20th Intercept Top 20th Intercept Slopea Bottom 20th Intercept Top 20th Intercept Perceived love 0.72 6.6 7.6 0.73 6.6 7.8 Partner’s love –0.02 7.6 7.9 0.22 7.4 8.0 Chi-square test 29.9** 43.0** 3.4† 7.7** 27.1** 1.6 Perceiver’s optimism 0.46 6.5 7.2 0.57 6.2 7.2 Partner’s commitment –0.04 7.4 7.8 0.15 7.6 8.4 Chi-square test 10.3** 47.4** 24.9** 7.6** 64.7** 41.5** a. Nonstandardized coefficients are presented. †p < .10. *p < .05. **p < .01. at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ path models by constraining certain paths to be equal (Kenny, 1996). Within each sample, for instance, we tested whether men and women were equally likely to regulate positive perceptions of the partner with per- ceived love. We did this by comparing the fit of models that estimate common perceived love paths for men and women (equality constraints placed on paths d and d in Figure 3) with the fit of models that estimate separate coefficients (paths d and d free to vary). If men and women did differ, then the goodness of fit for the models estimating separate coefficients would be significantly better (i.e., a smaller chi-square) than the goodness of fit for the models estimating common coefficients (a 1 df test). The same approach can be used to test for differ- ences between married and dating samples. As a general strategy, we first fit models estimating sep- arate path coefficients for men and women within each sample. In all cases, the size of the paths was similar for both genders. Therefore, we fit models estimating pooled or common path coefficients for men and women within each sample. In most cases, the size of the corresponding (pooled by gender) paths was similar for dating and mar- ried samples, so we fit a third set of models estimating pooled path coefficients for married and dating men and women. In reporting our results, we present standardized path coefficients that are pooled across both gender and rela- tionship status for the majority of the paths. We place these coefficients on only the nonprimed paths. For cer- tain paths, however, the model test described above revealed significant differences between dating and mar- ried couples. In such cases, we put the coefficients for the dating sample on the nonprimed paths (subscripted D) and the coefficients for the married sample on the primed paths (subscripted M). We present separate paths only when the reduction in chi-square was at least mar- ginally significant (p < .10). Figure 3 also contains the pooled, standardized path coefficients and fit statistics for this model collapsed across dating and married samples. First, the similarity 430 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN Figure 3 The role of perceived love and perceptions of the partner in mediating the link between self-models and satisfaction. NOTE: Estimates for the residual correlations between women’s and men’s perceptions of their partners’ love, women’s and men’s perceptions of their partners, and women’s and men’s satisfaction also were included in the estimation of the model. Goodness of Fit Index = .97, Comparative Fit Index = 1.00, 2(34, N = 220) = 26.35, ns. †p < .10. *p < .05. **p < .01. at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ between partners’ self-images, as indexed by the zero- order correlation, was minimal, r(119) = .13, ns, for dat- ing couples, and r(103) =.20, p < .05, for married cou- ples. Turning to perceived love, paths a and a were strong and positive, supporting our hypothesis that self- models predict perceptions of a partner’s love for the self. Individuals also felt more loved the more positive their partners’ self-models, although paths b and b were only marginally significant. Turning to perceptions of the partner, dating and married individuals reported more positive impressions of their partners the more positive their partners’ mod- els of self. However, these reality effects (paths e and e ) were significantly stronger in the married than the dat- ing sample, 2(1, N = 220) = 9.00, p < .01. More impor- tant, perceived love predicted perceptions of the partner as the pooled paths d and d were strong and significant in both samples. Intimates who felt more loved per- ceived their partners more positively, whereas individu- als who felt less loved perceived their partners less posi- tively. This dynamic was most pronounced among the married couples, 2(1, N = 220) = 3.3, p < .10. Moreover, perceptions of a partner’s love partially mediated the link between self-models and perceptions of the partner. The direct effect for self-models (paths c and c ) was still significant in the mediational model, but it was very sig- nificantly reduced in comparison to its total effect, 2(1, N = 220) = 39.5, p < .001.3 Turning to the intrapersonal sources of satisfaction, feeling loved and perceiving a good, responsive partner each had statistically independent or unique effects in predicting satisfaction. That is, individuals were happier in their relationships the more love they perceived (paths g and g ) and the more positively they evaluated their partners (paths h and h ). The additive effects of feeling loved by a good partner also completely medi- ated the link between self-models and satisfaction in dat- ing couples and partially mediated this relation for mar- ried couples, 2(1, N = 220) = 5.1, p < .05. That is, the direct effect of self-models (paths f and f ) was nonsignificant for dating couples and marginal for mar- ried couples. Moreover, the magnitude of these paths was significantly reduced compared to their total effects for married couples, 2(1, N = 220) = 4.4, p < .05. Turning to the interpersonal effects, individuals were more satis- fied the more loved their partners felt (paths k and k ) and the more positively their partners actually viewed them (paths j and j ). Table 2 presents a parallel pattern of results for optimism.4 Alternative explanations? These results support our rea- soning that individuals troubled by self-doubt are less sat- isfied and less optimistic in part because they have trou- ble finding sufficient reason for a sense of felt security in their relationships. Moreover, further analyses revealed that these mediational patterns also were evident when we examined the constituent elements of the model of self composite individually. At least three possible alternative explanations for these results exist, however. The first alternative is that perceived love is simply redundant with perceptions of how the partner sees the self on traits, the type of re- flected appraisal we examined in our prior research (Murray et al., 2000). In the present data, perceptions of love were indeed based on the qualities individuals believed their partners saw in them; for men, r(119)D = .38, p < .001, r (103)M = .45, p < .001, for women, r(119)D = .43, p < .001, r(103)M = .49, p < .001. However, the moder- ate size of these correlations suggests that feeling loved is not isomorphic with feeling positively regarded. But we also conducted a much more stringent test of the importance of perceived love in a further set of mediational models. In these analyses, we included trait reflected appraisals as a further exogenous variable in the model depicted in Figure 3. Attesting to the impor- tance of feeling loved, these analyses revealed unique effects of self-models and trait-specific reflected apprais- als on perceived love, unique effects of perceived love and trait-specific reflected appraisals on perceptions of the partner, and unique effects of perceived love and perceptions of the partner on satisfaction and optimism. The second alternative is that individuals with more negative self-models feel less loved and see their partners less positively simply because they are involved in less sat- isfying relationships than highs. In other words, self- models, perceived love, and perceptions of the partner might only be related through a spurious correlation with satisfaction (i.e., a third variable model). To explore this possibility, we included satisfaction as an exogenous variable in the model illustrated in Figure 3. When we did Murray et al. / SELF-DOUBT AND RELATIONSHIP BELIEFS 431 TABLE 2: The Role of Perceived Love and Perceptions of the Part- ner in Mediating the Link Between Self-Models and Opti- mism Dating Married Coefficients Coefficients f and f : Perceivers’ self-models .01 .16** i and i : Partners’ self-models –.09* –.09* g and g : Perceivers’ perceived love .25** .25** k and k : Partners’ perceived love .20** .20** h and h : Perceivers’ view of partner .43** .28** j and j : Partners’ view of perceiver .08† .08† NOTE: Goodness of Fit Index = .95, Comparative Fit Index = .99, 2(33, N = 220) = 41.31, ns. See Figure 3 for paths. Estimates for the residual correlations between pairs of corresponding men’s and women’s paths also were included in the estimation of the models. Paths f and f and h and h were significantly different between samples, 2(1, N = 220) = 3.7, p < .10, 2(1, N = 220) = 6.65, p < .01, respectively. †p < .10. *p < .05. **p < .01. at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ this, we still found the predicted effects of self-models on perceived love and perceived love on perceptions of the partner. This suggests that dispositional insecurities play a unique role in undermining perceived love and gener- ous perceptions of the partner, one that is not simply redundant with satisfaction. The third alternative is that perceived love or gener- ous perceptions of the partner only seemed to play a mediating role because of biases associated with self-reports. For instance, individuals with more positive models of self might report feeling more loved and describe their partners more positively because they are more likely to depict themselves and their relationships in a socially desirable light. However, global reports of satisfaction undoubtedly capture individuals’ tenden- cies to distort their experiences to a public audience (or even to themselves). And when we controlled for satis- faction in the analyses described above, perceived love still mediated the link between models of self and part- ner perceptions. Unfounded Pessimism: Miscalibration in Relationship Inferences The results to this point shed some light on why indi- viduals with more negative models of self experience greater relationship difficulties. Is their inferential cau- tion warranted? Or do the dynamics of dependency reg- ulation result in excessive caution, or undue pessimism, for individuals with more negative self-models? As a first step in exploring this question, we needed to choose an appropriate benchmark for detecting undue pessimism, or insufficient optimism. Interdependence theorists argue (at least implicitly) that one intimate’s forecasts for the relationship ought to be calibrated with certain evidential bases; namely, the partner’s level of commitment (e.g., Kelley, 1983). According to this anal- ysis, individuals who possess more committed partners are reasonably justified in making optimistic forecasts for their futures. Our dependency regulation framework, though, sug- gests that such perceptions ought to show systematic pat- terns of miscalibration as a function of the perceiver’s model of self. To examine this hypothesis, we returned to the regression plot analysis we described earlier. We first estimated regression lines predicting the perceiver’s feelings of optimism from the perceiver’s self-models. We then estimated regression lines predicting the part- ner’s actual expressions of commitment from the con- tinuous measure of the perceiver’s self-models (pooling across gender). Next, we tested the significance of the difference between the slopes of these lines and the intercepts of these lines using the procedure we described earlier. The right half of Figure 2 presents example regres- sion lines, and the bottom half of Table 1 contains the slopes, intercepts, and chi-square tests. The pattern of results was similar for men and women in both dating and married samples. First, as we expected, the slope of the optimism regression line was generally significantly higher than the partner’s commitment regression line. That is, optimism was much more a function of the perceiver’s self-models than the partner’s actual level of commitment. Second, the intercepts for the partner’s commitment regression lines also were significantly higher than the intercept for the optimism regression line at both intercept points, but this discrepancy was much more pronounced at the bottom 20th percentile on self-models. In combination, these regression lines illustrate the miscalibration pattern we expected: Inti- mates with more negative self-models made less optimis- tic forecasts about the future of their relationships than their partners’ actual levels of commitment seemed to warrant.5 The Mismeasure of Generality? The present analyses extend our prior work with these samples by showing, first, that self-doubt seems to com- promise perceptions of a partner’s love; second, that feeling loved and perceiving a good, responsive partner are both critical for satisfaction and optimism; and third, that self-doubt predicts more cautious inferences about the future than a partner’s actual feelings of commit- ment seem to warrant. Although all of these effects were statistically independent of the effects of trait-specific reflected appraisals, some readers might still wonder whether we are somehow capitalizing on chance by investigating novel hypotheses within these samples. Accordingly, we conducted independent tests of these hypotheses within new samples of dating (N = 99) and married (N = 94) couples participating in daily experi- ences studies at the State University of New York at Buf- falo (Murray, Bellavia, & Rose, 2000). These samples in very large part replicated the findings we presented earlier. DISCUSSION The experience of romantic love tempts individuals with the hope of satisfying basic needs for belongingness and thus affirming the worthiness of the self. However, the process of becoming attached simultaneously leaves individuals vulnerable to rejection and thus poses an implicit threat to the self. In the face of such stakes, some degree of circumspection is natural, even reasonable, but the present findings suggest that individuals with more negative models of self may be unduly cautious, need- lessly compromising the quality of their relationships. 432 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ Seeking Felt Security With a Miscalibrated Sociometer We believe that the conjunctive sense of feeling loved by a good, responsive partner provides the core of the representational structure subjectively experienced as felt security. Although attachment theorists typically define and measure felt security as a property of a per- son’s general working models, our findings suggest that this sense of trust in the partner’s availability and respon- siveness might be best considered a property of a specific relationship (Holmes & Rempel, 1989). After all, per- ceived love and perceptions of the partner predicted well-being, controlling for more general models of self. Rather than being isomorphic with felt security, models of self may be more general representations that can color perceptions of specific partners and relationships (and thus the potential for experiencing felt security) in a relatively top-down fashion. Cast in these terms, the results shed further light on why self-doubt seems to pose a relationship vulnerability. Perhaps most critically, negative models of self predict difficulty in attaining the hypothesized primary compo- nent of felt security—faith in the partner’s love. Rather than stemming from general models of other or simply reflecting a partner’s actual feelings, perceptions of a partner’s love are in large part projections, reflecting self-perceived worthiness of love (Path A in Figure 1). Thus, individuals with more positive models of self cor- rectly perceived loving partners, whereas individuals with more negative models of self needlessly doubted the strength of a partner’s love. Why might perceptions of a partner’s love be so inex- tricably linked to perceptions of one’s own worth? The present data suggest that perceiving a loving partner is partly contingent on believing that the partner sees posi- tive qualities in the self. That is, individuals feel more loved the more strengths they believe their partners see in them. For individuals with more negative models of self, an impoverished appreciation of the strengths their partners see in them and working models that stress the contingencies of interpersonal acceptance may make it difficult to construct or sustain the type of propositional structure that would support optimistic inferences about a partner’s love. For individuals with more positive mod- els of self, however, a resource-rich self-concept and working models that stress the unconditional nature of others’ regard easily support optimistic inferences about the partner’s regard and love. When individuals are troubled by self-doubt, the pro- cess of dependency regulation itself may then compro- mise the hypothesized secondary component of felt security—the perception of a good, responsive partner. The present findings suggest that individuals describe their partners more positively when they feel more loved. Moreover, perceptions of a partner’s love partially mediated the link between models of self and percep- tions of the partner in both samples. Troubled by unwar- ranted doubts about their partners’ love, individuals with more negative models of self seemed to react to this vulnerability by reserving judgment, perhaps defensively perceiving their partners in a less positive light. In con- trast, feeling more loved, individuals with more positive models of self could safely risk greater vulnerability, see- ing their partners in a more positive light than individu- als troubled by self-doubt. Consistent with our hypothesis that feeling loved by a good partner is the essence of felt security, dating and married individuals reported greater satisfaction in their relationships and expressed greater optimism about their futures when they felt more loved and perceived their partners more positively. Individuals also reported greater current satisfaction and future optimism when their partners felt more loved and perceived them more positively. Moreover, the additive effects of feeling loved by a good, responsive partner mediated the link between perceivers’ self-models and satisfaction and optimism in both samples.6 Such findings would not be at all surprising if individ- uals with more negative models of self actually attracted less valuable, less loving partners than individuals with more positive models of self. But the correlation between intimates’ self-models was minimal in both sam- ples, suggesting that individuals with more negative models of self do not necessarily possess more troubled partners than individuals with more positive models of self. Most crucial, individuals troubled by self-doubt were loved just as much as individuals more secure in their worth. The correlation between the perceiver’s self- models and the partner’s actual feelings of love was near zero in both samples. Instead, the findings suggest that individuals with more negative models of self are less happy in their relationships in part because they inter- pret relationship realities that could actually afford a sense of felt security in ways that mirror their own self-doubts. Not surprisingly, miscalibration in the perception of love and acceptance seemed to result in individuals with more negative models of self being overly cautious or conservative in the inferences they were willing to draw about the future of their relationships. That is, individu- als troubled by self-doubt reported less optimism about their futures than their partners’ actual feelings of com- mitment seemed to warrant. In contrast, the relationship inferences of individuals with more positive models of self were better calibrated to match their partners’ actual feelings of commitment. As positive, optimistic thinking predicts continued satisfaction and greater stability in dating relationships (Murray & Holmes, 1997), such Murray et al. / SELF-DOUBT AND RELATIONSHIP BELIEFS 433 at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ undue pessimism may have the ironic consequence of making further relationship difficulties (and thus the actual experience of rejection) more likely for individu- als troubled by self-doubt, even though it may function to protect them from the potential sting of rejection in the short term. Are We the Miscalibrated Ones? The present findings suggest that individuals trou- bled by self-doubt experience relationship difficulties in part because it is relatively hard for them to find suffi- cient reason to trust in their partners and relationships (even when this reason exists). Alternate causal models might provide reasonable explanations for the results, however. For instance, low self-esteem might first dimin- ish generosity in perceiving the partner and then tarnish perceptions of a partner’s regard for the self. On concep- tual grounds, however, perceiving a less-valuable partner should increase, rather than decrease, confidence in a partner’s love for the self if basic social exchange notions are considered. The opposite was the case in the present research. A more serious threat to our preferred inter- pretation of the results is that self-models, perceived love, and perceptions of the partner are related only through a shared, perhaps spurious, correlation with sat- isfaction. If that were true, then the links among these variables should disappear when satisfaction was placed first in the model (as an exogenous variable). This was not the case. Undoubtedly, though, the true causal dynamics are more complicated than the ones we have depicted. For instance, an interpersonal theorist on the self might argue that relationship perceptions should be portrayed as a cause rather than effect of self-esteem (e.g., Leary et al., 1995). From this perspective, relationships provide a looking glass in which self-concepts are formed (e.g., Cooley, 1902; Mead, 1934). Our prior research even speaks to this possibility. A longitudinal study of dating couples revealed that individuals incorporate their part- ners’ idealized images of them into their self-concepts as time passes (Murray et al., 1996b). Given the potential for such effects, why, then, did we depict self-models as the originating cause rather than perceptions of a part- ner’s love or the quality of the relationship itself? In our view, it is more logical to depict self-models as an exogenous variable given that a sense of self-worth also may originate in early experiences and attachments that preceded the current romantic relationship. Once established, self-concepts are remarkably immune to social influence: Self-perceptions generally play a much greater role in structuring the imagined appraisals of others than the actual appraisal of others plays in struc- turing self-perceptions (see Kenny, 1994, for a review). Models of self also predicted perceived love and percep- tions of the partner in analyses that controlled for satisfaction in the present research, further supporting our contention that models of self are not simply rooted in current relationship experiences. The link between self-models and relationship perceptions is likely to be a reciprocal one, though, and continuing to untangle these complex causal dynamics will require further research. A Self-Fulfilling Prophecy? Perhaps the greatest potential benefit of close roman- tic attachments comes from the satisfaction of the need for belongingness (Baumeister & Leary, 1995). For indi- viduals with more negative models of self, though, this need may be difficult to fulfill even in relationships where loving, committed partners are actually available. Why might this occur? Is it because individuals with more negative models of self really want their romantic partners to confirm their own self-doubts about their worthiness of love (e.g., Swann et al., 1992)? Our research suggests that doubts about a partner’s regard may surface even though lows seek and desire a sense of acceptance. For instance, even individuals with lower self-esteem want their partners to see them much more positively than they see themselves (Murray et al., 2000), and they are actually happier in their relationships when they feel more positively regarded (Murray et al., 2000) and feel more loved by their partners. Ironically, it may be precisely their need for accep- tance that oversensitizes and thus miscalibrates the sociometers of individuals troubled by self-doubt (e.g., Nezlek, Kowalski, Leary, Blevins, & Holgate, 1997). A basic tenet of the sociometer model is that individuals with low self-esteem are particularly needy of others’ acceptance (because they are operating at a chronic inclusion deficit). However, the sociometer is calibrated to be more sensitive and reactive to signs of rejection than acceptance, meaning that rejection poses a greater loss to the self than acceptance poses a gain (Leary et al., 1995). As a result, low-self-esteem individuals tend to be more interested in protecting the self against loss than high-self-esteem individuals, taking only those opportu- nities for self-enhancement that seem safe and sure to affirm the self and avoiding those that run the risk of threatening an impoverished sense of self (see Baumeister, 1993, for a review). Our dependency regulation perspective suggests that intimates react to (perceived) signs of a partner’s disaf- fection by finding greater fault in their partners and dis- tancing themselves from their relationships (Murray et al., 1998, 2000). The operation of an overactivated and thus miscalibrated sociometer may thus create a fur- 434 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ ther difficulty for individuals troubled by self-doubt— the erosion of the evidential basis for felt security, the existence of a loving, committed, responsive partner. Consistent with this analysis, women who anticipate rejec- tion in relationships seem to behave in ways during con- flicts that actually elicit less-accepting behaviors from their romantic partners and may even set the stage for dissolution (Downey, Freitas, Michaelis, & Khouri, 1998). In such ways, being overly cautious in the quest for felt security may have the eventual, and ironic, effect of cre- ating the troubled romantic reality that individuals with more negative models initially, incorrectly, perceived. NOTES 1. Both samples participated in the studies described in Murray, Holmes, and Griffin (2000), and the dating couples participated in the study described in Murray, Holmes, and Griffin (1996b). No overlap- ping findings are presented in the current article. 2. The test of the slopes is normally based on the assumption of equal variances in the dependent variables. This assumption was met for dating women and married men and women. However, dating men’s perceptions of their partners’ love was significantly more vari- able than the measure of their partners’ actual love. Crucially, though, we obtained the same pattern of results for dating and married men and women, suggesting that the greater variability of the perceived love measure for dating men did not disproportionately influence the results. 3. We tested for partial mediation by comparing the fit of a model that left paths c and c (in Figure 3) free to vary (i.e., the model we esti- mated) to the fit of a model that constrained these paths to be equal to the size of their total effects. Partial mediation is suggested if the good- ness of fit for the model that left these paths free to vary is significantly better (i.e., a smaller chi-square) than the goodness of fit for the model that set these paths equal to their total effects. 4. We also found a negative effect of the partner’s self-models pre- dicting both satisfaction and optimism. We suspect this represents a suppressor effect because the zero-order correlations of the positivity of the partner’s self-models with the perceiver’s optimism and satisfac- tion were near zero in the dating sample and all positive and significant in the married sample. 5. The assumption of equal variances in the dependent variables was not met because the measure of the perceiver’s optimism was usu- ally less variable than the measure of the partner’s commitment (due to a natural ceiling effect for the latter scores). We believe these analy- ses are still appropriate because the greater variance in commitment works against the observed effects. 6. We found a parallel pattern of mediational results when we used feelings of love toward the partner as the outcome measure. Individ- uals with more positive models of self felt more loved, and feeling more loved predicted greater love for the partner. REFERENCES Aron, A., Dutton, D. G., Aron, E. N., & Iverson, A. (1989). Experiences of falling in love. Journal of Social and Personal Relationships, 6, 243-257. Baldwin, M. W., & Sinclair, L. (1996). Self-esteem and “if . . . then” con- tingencies of interpersonal acceptance. Journal of Personality and Social Psychology, 71, 1130-1141. Bartholomew, K. (1990). Avoidance of intimacy: An attachment per- spective. Journal of Social and Personal Relationships, 7, 147-178. Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category model. Journal of Personality and Social Psychology, 61, 226-244. Baumeister, R. F. (1993). Self-esteem: The puzzle of low self-regard. New York: Plenum. Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497-529. Berscheid, E., & Fei, J. (1977). Romantic love and sexual jealousy. In G. Clanton & L. G. Smith (Eds.), Jealousy (pp. 101-109). Englewood Cliffs, NJ: Prentice Hall. Bowlby, J. (1982). Attachment and loss: Vol. 1. Attachment. London: Hogarth. Brickman, P. (1987). Commitment, conflict, and caring. Englewood Cliffs, NJ: Prentice Hall. Brown, J. D. (1998). The self. New York: McGraw-Hill. Clark, M. S., & Reis, H. T. (1988). Interpersonal processes in close rela- tionships. Annual Review of Psychology, 39, 609-672. Collins, N. L., & Read, S. J. (1994). Cognitive representations of attach- ment: The structure and function of working models. Advances in Personal Relationships, 5, 53-90. Cooley, C. H. (1902). Human nature and the social order. New York: Schocken. Downey, G., Freitas, A. L., Michaelis, B., & Khouri, H. (1998). The self- fulfilling prophecy in close relationships: Rejection sensitivity and rejection by romantic partners. Journal of Personality and Social Psy- chology, 75, 545-560. Edwards, J. R. (1994). Regression analysis as an alternative to differ- ence scores. Journal of Management, 20, 683-689. Fincham, F. D., & Bradbury, T. N. (1993). Marital satisfaction, depres- sion and attributions: A longitudinal analysis. Journal of Personality and Social Psychology, 64, 442-452. Gilbert, D. T. (1998). Ordinary personology. In D. T. Gilbert, S. T. Fiske, & G. Lindsey (Eds.), The handbook of social psychology (Vol. 2, pp. 89-150). Boston: McGraw-Hill. Griffin, D. W., & Bartholomew, K. (1994). Models of the self and other: Fundamental dimensions underlying measures of adult attach- ment. Journal of Personality and Social Psychology, 67, 430-445. Griffin, D. W., Murray, S., & Gonzalez, R. (1999). Difference score cor- relations in relationship research: A conceptual primer. Personal Relationships, 6, 505-518. Hazan, C., & Diamond, L. (in press). The limits of sexual strategies the- ory and the promise of attachment theory for explaining human mating. Review of General Psychology. Hazan, C., & Shaver, P. R. (1994). Attachment as an organizational framework for research on close relationships. Psychological Inquiry, 5, 1-22. Hendrick, S. S., Hendrick, C., & Adler, N. L. (1988). Romantic relation- ships: Love, satisfaction, and staying together. Journal of Personality and Social Psychology, 54, 980-988. Holmes, J. G., & Rempel, J. K. (1989). Trust in close relationships. In C. Hendrick (Ed.), Review of personality and social psychology: Close relationships (Vol. 10, pp. 187-219). Newbury Park, CA: Sage. Karney, B. R., & Bradbury, T. N. (1997). Neuroticism, marital interac- tion, and the trajectory of marital satisfaction. Journal of Personality and Social Psychology, 72, 1075-1092. Kelley, H. H. (1983). Love and commitment. In H. H. Kelley, E. Berscheid, A. Christensen, J. H. Harvey, T. L. Huston, G. Levinger, E. McClintock, L. A. Peplau, & D. R. Peterson (Eds.), Close relationships (pp. 265-314). New York: Freeman. Kelly, E. L., & Conley, J. J. (1987). Personality and compatibility: A pro- spective analysis of marital stability and marital satisfaction. Journal of Personality and Social Psychology, 52, 27-40. Kenny, D. A. (1994). Interpersonal perception: A social relations analysis. New York: Guilford. Kenny, D. A. (1996). Models of non-independence in dyadic research. Journal of Social and Personal Relationships, 13, 279-294. Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. L. (1995). Self- esteem as an interpersonal monitor: The sociometer hypothesis. Journal of Personality and Social Psychology, 68, 518-530. Mead, G. H. (1934). Mind, self and society. Chicago: University of Chi- cago Press. Murray et al. / SELF-DOUBT AND RELATIONSHIP BELIEFS 435 at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/ Murray, S. L., Bellavia, G., & Rose, P. (2000). The daily experiences of dat- ing and married couples. Unpublished data, University at Buffalo, The State University of New York. Murray, S. L., & Holmes, J. G. (1997). A leap of faith? Positive illusions in romantic relationships. Personality and Social Psychology Bulletin, 23, 586-604. Murray, S. L., Holmes, J. G., Dolderman, D., & Griffin, D. W. (2000). What the motivated mind sees: Comparing friends’ perspectives to married partners views of each other. Journal of Experimental Social Psychology, 36, 600-620. Murray, S. L., Holmes, J. G., & Griffin, D. (1996a). The benefits of posi- tive illusions: Idealization and the construction of satisfaction in close relationships. Journal of Personality and Social Psychology, 70, 79-98. Murray, S. L., Holmes, J. G., & Griffin, D. W. (1996b). The self-fulfilling nature of positive illusions in romantic relationships: Love is not blind, but prescient. Journal of Personality and Social Psychology, 71, 1155-1180. Murray, S. L., Holmes, J. G., & Griffin, D. W. (2000). Self-esteem and the quest for felt security: How perceived regard regulates attach- ment processes. Journal of Personality and Social Psychology, 78, 478-498. Murray, S. L., Holmes, J. G., MacDonald, G., & Ellsworth, P. (1998). Through the looking glass darkly? When self-doubts turn into rela- tionship insecurities. Journal of Personality and Social Psychology, 75, 1459-1480. Nezlek, J. B., Kowalski, R. M., Leary, M. R., Blevins, T., & Holgate, S. (1997). Personality moderators of reactions to interpersonal rejec- tion: Depression and trait self-esteem. Personality and Social Psychol- ogy Bulletin, 23, 1235-1244. Reis, H. T., & Shaver, P. (1988). Intimacy as an interpersonal process. In S. W. Duck (Ed.), Handbook of personal relationships (pp. 367-389). New York: John Wiley. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Shrauger, J. S., & Schoeneman, T. J. (1979). Symbolic interactionist view of the self-concept: Through the looking glass darkly. Psycholog- ical Bulletin, 86, 549-573. Swann, W. B., Hixon, J. G., & De La Ronde, C. (1992). Embracing the bitter “truth”: Negative self-concepts and marital commitment. Psy- chological Science, 3, 118-121. Wieselquist, J., Rusbult, C. E., Foster, C. A., & Agnew, C. R. (1999). Commitment, pro-relationship behavior, and trust in close rela- tionships. Journal of Personality and Social Psychology, 77, 942-966. Received June 21, 1999 Revision accepted March 21, 2000 436 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN at University of Newcastle on September 26, 2014psp.sagepub.comDownloaded from http://psp.sagepub.com/