attachment models of the self and others

Attachment models of the self and others:

Relations with self-esteem, humanity-esteem, and

parental treatment

MICHELLEA. LUKE,a,* GREGORYR.MAIO,a AND KATHERINE B. CARNELLEYb

aCardiff University and bUniversity of Southampton

AbstractThe present research tested the extent to which perceptions of early childhood experiences with parents predicted

general views of the self (i.e., self-esteem) and others (i.e., humanity-esteem), and whether attachment self- and

other-models mediated these links. Two studies used a new measure of humanity-esteem (Luke & Maio, 2004) to

achieve these ends. As expected, indices that tapped a positive model of the self in relationships were associated

with high self-esteem and indices that tapped a positive model of others in relationships were associated with high

humanity-esteem. Also, early attachment experiences with fathers and mothers predicted self-esteem and

humanity-esteem, respectively, and these direct relations were mediated by the attachment models. The studies,

therefore, provide direct evidence that attachment measures predict general favorability toward the self and others,

while revealing novel differences in the roles of childhood experiences with fathers and mothers.

No variables, it is held, have more far-reaching effects on personalitydevelopment than a child’s experiences within the family: for, starting during hisfirst months in his relation with his mother figure, and extending through theyears of childhood and adolescence in his relation to both parents, he builds upworking models of how attachment figures are likely to behave towards him inany of a variety of situations; and on those models are based all his expectations,and therefore all his plans, for the rest of his life.

Bowlby (1973; p. 369)

There is an enduring fascination with thenotion that children’s early experienceswith their parents leave a permanent stampon their lives. This idea is the centerpiece ofBowlby’s (1969, 1973) seminal theory aboutattachment processes. As indicated in theabove quote, Bowlby suggested that earlyattachment experiences (e.g., parental care)should influence the development of internalworking models, which are crucial variablesin subsequent social behavior throughoutthe lifespan. These internal working modelsare complex schemas that include feelingsof self-worth and self-acceptance (model ofself) and beliefs about the availability and

*Present address: Centre for Research on Self andIdentity, School of Psychology, University ofSouthampton, Southampton, UK.This work was part of a doctoral dissertation by

Michelle A. Luke under the supervision of Gregory R.Maio. Portions of this research were presented at theannual meetings of the American Psychological Society,Toronto, Canada, June 2001 and the British PsychologicalSociety, Glasgow, Scotland, March 2001. We thankEmma Dowden, Laura Keyse, and Maria Lima for theirassistance in data collection. We also thank GordonHarold and Constantine Sedikides for their helpfulsuggestions concerning this research and Emma Dowden,John-Mark Frost, KatyGreenland, Geoff Haddock, TomHuskinson, Maria Lima, Kerry Rees, Susan Sprecher,Ulrich von Hecker, and six anonymous reviewers fortheir comments on an earlier version of this paper.Correspondence should be addressed to Michelle

A. Luke, Centre for Research on Self and Identity,School of Psychology, University of Southampton,Highfield Campus, Southampton SO17 1BJ, UK;e-mail: [email protected].

Personal Relationships, 11 (2004), 281–303. Printed in the United States of America.Copyright # 2004 IARR. 1350-4126/02

281

responsiveness of attachment figures, such asprimary caregivers and significant others(model of others; Bowlby, 1969, 1973). Ingeneral, if a child receives sensitive, respon-sive care, then he/she will develop a view ofthe self as worthy and a view of others asreliable and predictable. However, if a childreceives inconsistent care, rejection, orneglect, he/she may develop a view of the selfas unworthy andaviewof others as unreliable.These models of early attachment experienceswill be carried forward in life and influenceexpectations for other attachment figuresand for people in the world in general.Because internal working models are pre-

sumed to subsume many beliefs and feelingsabout the self and others, efforts to validatemeasures of these internal working modelshave focused on their relations to evalua-tions and expectations of the self and others(e.g., Bartholomew & Horowitz, 1991;Cozzarelli, Hoekstra, & Bylsma, 2000;Griffin&Bartholomew,1994a, 1994b;Gurung,Sarason, & Sarason, 2001). The presentresearch re-examines the relations betweeninternal working models of attachment,general evaluations of the self and others,and perceptions of parental treatment earlyin life. To foreshadow, we argue that priorresearch that has examined these linkshas not adequately assessed evaluations ofothers in general at the ultimate level (i.e.,humanity-esteem), beyond the contexts ofrelationships. Furthermore, we test whethera direct measure of humanity-esteem yieldsrelations with parental treatment and inter-nal working models of the self and othersthat are consistent with Bowlby’s theory.

Attachment and Esteem for the Self and

Others

Internal working models of the self andothers can be tapped by measures that havebeen developed to assess attachment stylesin infants (Ainsworth, Blehar, Waters, &Wall, 1978), adults (George, Kaplan, &Main, 1984; Main & Goldwyn, 1994), andadult romantic relationships (Hazan &Shaver, 1987). For example, Bartholomewand Horowitz (1991) developed measures

(interview and self-report) that can be usedto assess four attachment styles in adultrelationships: secure, preoccupied, fearful,and dismissing. Securely attached peoplehave a positive model of both the self andothers in their relationships. People witha preoccupied attachment have a negativemodel of self and a positive model ofothers in their relationships. Fearfullyattached people have a negative model ofboth self and others in their relationships.People with a dismissive attachment have apositive model of self and a negative modelof others in their relationships. Throughoutthis paper, we refer to models derived fromadult relationship attachment measures asthe attachment self-model and the attach-ment other-model, in order to distinguishthem from variables that tap global viewsof the self and others.Researchers have subsequently determined

that dimensions similar to self- and other-models, attachment anxiety and attachmentavoidance, respectively, underlie all attach-ment measures (e.g., Bartholomew &Shaver, 1998; Brennan, Clark, & Shaver,1998; Fraley, Waller, & Brennan, 2000).Regardless of how researchers examinethe attachment dimensions, results havesupported many of Bowlby’s (1969, 1973)predictions regarding the development ofinternal working models. In support ofBowlby’s (1973) contention that the internalworking models of the self and others arelogically independent, Bartholomew andHorowitz (1991) found that, although themajority of their participants had models ofself and others that were congruent invalence (i.e., secure and fearful attachmentstyles), other participants demonstrateddifferent valences in their models of selfand others (i.e., preoccupied and dismissingattachment styles). Their research showsthat models of self and others can varyindependently and that each is an importantpart of a person’s attachment orientation.For the present investigation, it is impor-

tant that these models should predict gen-eral beliefs and feelings about the self andothers that do not necessarily arise from thepersonal relationships in which attachment

282 M. A. Luke, G. R. Maio, and K. B. Carnelley

models are based. For example, accordingto Bowlby (1969, 1973), a positive internalworking model of the self provides a securebase to explore and pursue other endeavors.This hypothesis that attachment modelsinfluence more global evaluations of theself has been supported by findings thatpeople who possess a positive attachmentself-model indicate higher self-esteem andself-acceptance (e.g., Bartholomew &Horowitz, 1991; Collins & Read, 1990;Griffin & Bartholomew, 1994a, 1994b;Murray, Holmes, Griffin, Bellavia, & Rose,2001). Other research (Carnelley, Pietromo-naco, & Jaffe, 1994; Griffin & Bartholomew,1994a, 1994b) has found that a positiveattachment self-model is also negativelyassociated with psychological distress.

Similarly, a positive-attachment modelof others should predict beliefs and feelingsabout others in general (e.g., colleagues andstrangers), if, as Bowlby (1969, 1973) pre-dicted, our relationships provide the mostpowerful basis for forming internal workingmodels of others. Consistent with thisview, findings of past research indicatethat people with a positive attachment other-model expect positive intentions in others(Feeney & Noller, 1990; Hazan & Shaver,1987) and possess positive beliefs (e.g., trust)about others in general (Collins & Read,1990; Cozzarelli et al., 2000; Mikulincer,1998).

Nonetheless, positive expectations andbeliefs about others are only a sample ofimportant associates of a positive attach-ment model of others. Another importantpotential associate is humanity-esteem or aglobal evaluation of others in general. Glo-bal evaluations of others may be an impor-tant associate of a positive-attachmentmodel of others, because, as describedbelow, they should summarize more specificfacets (just as attitudes in general sum-marize more specific cognitions, feelings,and behaviors; Eagly & Chaiken, 1993).That is, a global evaluation of others shouldreflect specific beliefs, feelings, and pastbehaviors that are associated with others.Yet, no research has examined how theattachment other-model relates to a global

evaluation of others. This evaluative com-ponent was the focus of our studies.

Because none of the past associates of theattachment other-model directly tap aglobal evaluation of others, the attachmentother-model has not been examined in thesame way as the attachment self-model (i.e.,with regard to its ability to predict an over-all attitude toward others outside the con-text of relationships). If models of the selfand others were both examined as globalevaluations, researchers could test thedistinctiveness of the attachment models.That is, does the attachment model of selfaccount for unique variance in self-esteemafter controlling for the attachment modelof others? Similarly, does the attachmentmodel of others account for unique variancein humanity-esteem after controlling for theattachment model of self? If these questionsare answered in the affirmative, the resultswould indicate that the attachment modelshave distinct relations to global evaluationsof the self and others.

Griffin and Bartholomew (1994a, 1994b)have previously indicated the importanceof using such evidence to establish the con-struct validity of the attachment measures.Their research found support for Bowlby’s(1973) hypothesis regarding the extent towhich the internal working models reflectseparate evaluations of the self and others.In particular, they found that the attach-ment self-model uniquely predicted apositive self-concept (i.e., high self-esteemand self-acceptance and low levels of dis-tress) and that the attachment other-modeluniquely predicted positive interpersonalfunctioning (i.e., sociability and warmth).Their measure of self-esteem directly tappedglobal evaluations of the self, whereas theirmeasures of interpersonal functioningtapped specific orientations toward peoplein the participants’ environment. Becausethey did not assess global evaluations ofpeople in general, they were not able toexamine whether there are distinct relationsbetween the models of self and others andglobal evaluations of the self and others. Weextend this and other research (e.g., Collins& Read, 1990; Cozzarelli et al., 2000) by

Attachment models 283

examining fully the unique relations betweenattachment models of self and others andglobal evaluations of the self and others.

Assessing Humanity-Esteem

Perhaps one reason why past research hasnot examined the relation between attach-ment models of others and humanity-esteem is that the latter measure has notbeen developed and validated. The needfor a measure of humanity-esteem was madeevident by self-categorization theory (e.g.,Turner, 1985; Turner, Hogg, Oakes,Reicher, & Wetherell, 1987; Turner, Oakes,Haslam, & McGarty, 1994). Self-categoriza-tion theory suggests that people can definethemselves as unique individuals (personalidentity), as members of a particular group(social identity), or as members of hum-anity (human identity). Furthermore, self-categorization theory postulates that allthree levels of categorization are importantand interrelated, such that the personalidentity is included in the social and humanidentity and the social identity is includedin the human identity.

Researchers have conceptualized evalua-tions of the personal identity and socialidentity in terms of overall favorabilitytoward the self and one’s social groupusing established measures of self-esteem(e.g., Fleming & Courtney, 1984; Fleming &Watts, 1980;Robins,Hendin,&Trzesniewski,2001; Rosenberg, 1989) and collective self-esteem (e.g., Luhtanen & Crocker, 1992),respectively. However, until now, a measureof overall favorability toward humanityhas not existed. In the past, researchershave conceptualized evaluations of thehuman identity in terms of specific beliefs(e.g., trust) about human beings (e.g.,Rosenberg, 1956; Wrightsman, 1992). Infact, Wrightsman (1991) acknowledged thefact that the existing measures of specificbeliefs toward people all tap beliefs aboutthe trustworthiness of people in some way.General positive or negative evaluationshave not been assessed.

There is empirical precedent for believingthat a general attitude toward humanity

should differ from specific beliefs abouthuman nature. For example, past research(e.g., Marsh, 1986; Rosenberg, 1979) hasfound that specific and global self-esteemare distinct. (A similar model also existsfor specific and general values; see Cantril,1932.) Also, researchers have suggested thatattitudes, in general, consist of emotionsand past behavior, in addition to beliefs(e.g., Katz & Stotland, 1959; Rosenberg &Hovland, 1960; Triandis, 1971; Zanna &Rempel, 1988). Therefore, specific beliefsabout human nature should explain onlysome of the variance in humanity-esteem.Part of the remaining unexplained variancein humanity-esteem should be explainedby past behavior and emotions.

An interesting implication of theself-categorization perspective is thathumanity-esteem should reflect not onlyevaluations of the self, but also evaluationsof individuals outside the self, includingoutgroup members. As a result, self-esteemand humanity-esteem should be related,but not highly. There should be a part ofhumanity-esteem that is based on thoughtsand feelings about others, such as thosesubsumed within the attachment model ofothers. These hypotheses can be tested usingthe Humanity-Esteem Scale, which werecently constructed (Luke & Maio, 2004).Unlike previous measures that assess specificbeliefs about people (Rosenberg, 1956;Wrightsman, 1992), this new measure focuseson general evaluations of humanity.

We have shown that the global humanity-esteem construct is unidimensional and thatthe Humanity-Esteem Scale possesses goodconstruct validity (Luke & Maio, 2004;Luke,Maio,&Carnelley, 2002). For example,consistent with self-categorization theory(e.g., Turner, 1985; Turner et al., 1987, 1994),people who have high humanity-esteempossess high self-esteem and high collectiveself-esteem. In addition, and consistent withthe contention that people affiliate withothers whom they like, people who havehigh humanity-esteem possess high levelsof extraversion. Supporting the measure’sdiscriminant validity, humanity-esteem isuncorrelated with cognitive motivation and


cognitive style. Perhaps most importantly,the measure of humanity-esteem is stronglyrelated to perceptions of the extent towhich people in general possess a varietyof traits and values and with emotionsfelt toward people in general (Luke, 2003).Thus, humanity-esteem is more than justspecific beliefs about people. Just as attitudesin general summarize specific beliefs andfeelings, humanity-esteem summarizes beliefsand (especially) feelings about people ingeneral.

Potential antecedents and consequencesof humanity-esteem have been examined.We found that media images of peopleupholding social values increase humanity-esteem, whereas media images of peoplethreatening social values decrease humanity-esteem (Luke &Maio, 2004). With regard toconsequences, both manipulated and meas-ured humanity-esteem predict intergroupbias (Luke & Maio, 2004), such that peoplewith high humanity-esteem exhibit lessintergroup bias than people with lowhumanity-esteem. Therefore, the Humanity-Esteem Scale possesses good construct validityand predicts important social variables.

Another potential antecedent of humanity-esteem should be attachment experiences.Our prediction is consistent with Bowlby’s(1973) hypothesis that one’s experiences andexpectations for an attachment figure’sreliability will influence one’s expectationsfor attachment figures and the benevolenceof others in general because humanity-esteemsummarizes a variety of specific beliefs aboutand feelings toward others (Luke, 2003).In otherwords internalmental representationsof early attachment experiences represent ourfirst attempt to make sense of and predictothers’ behavior. These working models arecarried forward and influence how we judgehumanity.

Attachment experiences as an antecedentto humanity-esteem has not yet been exam-ined and is the focus of the present research.Study 1 addresses this issue by examiningthe ability of the attachment self-model andthe attachment other-model to predictglobal evaluations of the self and others.

Study 2 extends the analysis to include therole of early attachment experiences indetermining models of self and others.

Study 1

By including extant measures of self-esteem(Fleming & Courtney, 1984; Rosenberg,1989) and the Humanity-Esteem Scale(Luke & Maio, 2004) in the presentresearch, we were able to examine the abilityof the attachment self-model and the attach-ment other-model to predict global evalua-tions of the self and others. We anticipateda small but significant relation betweenself-esteem and humanity-esteem becauseself-categorization theory (e.g., Turner, 1985;Turner et al., 1987, 1990) predicts that thedifferent levels of the self are intertwined. Inaddition, drawing from attachment theoryand research (e.g., Bowlby, 1969, 1973;Cozzarelli et al., 2000) and self-categorizationtheory (e.g., Turner, 1985; Turner et al., 1987,1994), we expected that the attachmentmodel of the self would uniquely predict self-esteem and that the attachment model ofothers would uniquely predict humanity-esteem.

Method

Participants and procedure

Participants were 140 undergraduates (117women and 23 men) who received coursecredit or were paid £3.00 for participation.Five additional participants were omittedfrom the analyses because of failure tofollow instructions (n¼ 3), or because theyanticipated the nature of the relationsbetween all of the variables examined inthis study when probed for suspicion (n¼ 2).

The study utilized a correlational design toexamine relations between (a) attachmentmodels of self and others and (b) self-esteem and humanity-esteem. Participantstook part individually and were asked tocomplete Bartholomew and Horowitz’s(1991) Relationship Questionnaire, Luke


and Maio’s (2004) Humanity-Esteem Scale,and other measures that were not relevantto the study. In addition, one subset ofparticipants (n¼ 59) completed Rosenberg’s(1989) Self-Esteem Scale. In another subsetof participants (n¼ 81), Fleming and Court-ney’s (1984) Self-Rating Revised Scale wasused to assess self-esteem, rather than theRosenberg Self-Esteem Scale. These partici-pants (n¼ 81) also completed Fraley et al.’s(2000) Experiences in Close RelationshipsRevised Questionnaire, as an additionalmeasure of adult attachment. In the secondsample, the attachment measures weregrouped together and presented in randomorder. In both samples, the esteem measureswere also grouped together and presentedin random order. The attachment andesteem measures were presented in counter-balanced order across participants.

Measures

General attachment styles. As we notedabove, we used two questionnaires tomeasure the two attachment dimensions ofthe self and others. The RelationshipQuestionnaire (Bartholomew & Horowitz,1991) contains one-paragraph descriptionsof four attachment styles: secure, preoccu-pied, fearful, and dismissing. Participantsindicated the extent to which each descrip-tion reflects them, using a seven-point scalefrom �3 (‘‘not at all like me’’) to þ3 (‘‘verymuch like me’’). Following the proceduresoutlined by Griffin and Bartholomew(1994a), we calculated the model of selfby subtracting the sum of the endorsementsof the two attachment styles that subsume anegative self-model (preoccupied and fear-ful) from the sum of the ratings of the twoattachment styles that subsume a positiveself-model (secure and dismissing). Themodel of others was calculated by subtract-ing the sum of the endorsements of thetwo attachment styles that subsume thenegative other-model (dismissing and fearful)from the sum of the ratings of the twoattachment styles that subsume the positiveother-model (secure and preoccupied). Inour sample, this procedure revealed a mildly

positive self-model (M¼ 1.28, SD¼ 4.55)and a mildly positive other-model(M¼ 0.94, SD¼ 4.34).

In addition, 81 participants completeda modified version of the Experiences inClose Relationships Revised Questionnaire(Fraley et al., 2000). To modify the scale toreflect attachment relationships with peoplein general, the word ‘‘others’’ was substi-tuted for the word ‘‘partner.’’ This ques-tionnaire consists of two 18-item subscales,reflecting the anxiety and avoidance dimen-sions of attachment. High scores on theanxiety dimension may reflect a negativeself-model, whereas high scores on theavoidance dimension may reflect a negativeother-model (see Bartholomew & Shaver,1998; Griffin & Bartholomew, 1994b).Items from the anxiety subscale include ‘‘Ioften worry that others don’t really loveme’’ and ‘‘I don’t often worry about beingabandoned’’ (reverse-scored). Items fromthe avoidance subscale include ‘‘I find itdifficult to allow myself to depend onothers’’ and ‘‘I usually discuss my problemsand concerns with others’’ (reverse-scored).Participants responded to each item using aseven-point scale ranging from 1 (‘‘disagreestrongly’’) to 7 (‘‘agree strongly’’). Afterreverse scoring the negatively worded items,total scores for each subscale were derivedby averaging the responses across all theitems from the subscale. Internal consis-tency was high for both the anxiety subscale(a¼ 0.94) and avoidance subscale(a¼ 0.93). The subscales revealed moderatelevels of attachment anxiety (M¼ 3.14,SD¼ 1.09) and attachment avoidance(M¼ 3.05, SD¼ 1.00).

Self-esteem. The Rosenberg Self-EsteemScale (Rosenberg, 1989) contains 10 itemsthat tap aspects of respondents’ evaluationsof themselves. Example items include‘‘On the whole, I am satisfied with myself’’and ‘‘I wish I could have more respectfor myself’’ (reverse-scored). Participantsresponded to each item using a seven-pointscale ranging from �3 (‘‘strongly disagree’’)to þ3 (‘‘strongly agree’’). After reversescoring the negatively worded items, total


scores were derived by averaging theresponses across all the items. The internalconsistency of this scale was high (a¼ 0.90),and the scale revealed that participants’mean self-esteem was somewhat positive(M¼ 1.25, SD¼ 1.03).

The 36-item Self-Rating Revised Scale(Fleming & Courtney, 1984) contains fivesubscales, which assess (a) self-regard,(b) social confidence, (c) school abilities,(d) physical appearance, and (e) physicalabilities. Participants completed all fivesubscales, but our focus was on the self-regard subscale, which contains seven itemsthat reflect self-acceptance and globalself-esteem (Fleming & Courtney, 1984).Examples of the seven items include ‘‘Howconfident do you feel that someday thepeople you know will look up to you andrespect you?’’ and ‘‘Do you ever think thatyou are a worthless individual?’’ (reverse-scored). Participants responded to eachitem using a seven-point scale rangingfrom 1 (‘‘strongly disagree’’) to 7 (‘‘stronglyagree’’). After reverse scoring the negativelyworded items from the self-regard subscale,the items were averaged. The internal con-sistency of the self-regard subscale washigh (a¼ 0.87), and the subscale revealedthat participants possessed somewhat posi-tive self-esteem (M¼ 4.76, SD¼ 1.11).

Fleming and Courtney (1984) found evi-dence that the Rosenberg Self-Esteem Scaleand the self-regard subscale measure thesame construct. In addition, our principalanalyses revealed similar results for bothmeasures. Therefore, to simplify the presen-tation of our findings, we converted scoreson the Rosenberg Self-Esteem Scale and theself-regard subscale to z-scores, enabling usto use the measures interchangeably in onelarge sample.

Humanity-esteem. TheHumanity-EsteemScale (Luke & Maio, 2004) has twoportions.1 The first portion contains 10items and taps specific aspects of people’s

evaluations of humanity by using items thatare similar to those of the Rosenberg’s(1989) Self-Esteem Scale. The principaldifference between the first portion of theHumanity-Esteem Scale and the RosenbergSelf-Esteem Scale is that the former tapsaspects of evaluations of humanity (e.g.,satisfaction with humanity and itsevolution), rather than aspects of self-esteem (Gurung et al., 2001, have used asimilar approach to operationalize esteemfor significant others). Examples include,‘‘On the whole, I am satisfied with the evo-lution of humanity,’’ ‘‘I feel human beingshave a number of very good qualities,’’ ‘‘Allin all, I am inclined to regard the humanspecies as a failure’’ (reverse-scored), and ‘‘Iwish I could have more respect for humanityin general’’ (reverse-scored). Participantsresponded to these items using a seven-point scale ranging from �3 (‘‘stronglydisagree’’) to þ3 (‘‘strongly agree’’).

The second portion of the Humanity-Esteem Scale is a single-item measure thatdirectly asks participants to report theirglobal evaluations of humanity. Pastresearch has revealed that a single-item‘‘attitude thermometer’’ is well suited forassessing global evaluations of groups(e.g., Esses, Haddock, & Zanna, 1993;Esses & Maio, 2002) and that a comparablesingle-item measure is well suited for asses-sing global evaluations of the self (e.g.,Robins et al., 2001). Consequently, weasked participants to indicate their overallfavorability versus unfavorability towardhumanity using a nine-point scale rangingfrom �4 (‘‘extremely unfavorable’’) to þ4(‘‘extremely favorable’’).

To compute humanity-esteem scores forthe first portion of the scale, we reverse-scored the negatively worded items and thencomputed the mean response across theitems. The interrelations among the itemswere high (a¼ 0.80), and they revealed thatparticipants possessed somewhat positivehumanity-esteem (M¼ 1.22, SD¼ 0.84).

Participants also indicated positivehumanity-esteem on the second portion ofthe scale (M¼ 2.22, SD¼ 1.36). The scoreson both portions were strongly correlated

1. Copies of the Humanity-Esteem Scale are availablefrom Michelle A. Luke upon request.


[r(138)¼ 0.73, p< 0.001], and we obtainedsimilar results when we examined the 10-item and single item measures separately.To simplify our analyses and to be certainthat we assessed global humanity-esteem, wecreated an index that gave equal weight tothe single global item and the mean of theother 10 items. That is, we computed parti-cipants’ standardized score (z-score) for eachportion and then averaged the z-scores.

Results

Possible sex differences were examined foreach measure and were found only forattachment avoidance. Men indicated higher

levels of attachment avoidance (M¼ 3.64,SD¼ 1.00) than women (M¼ 2.95, SD¼0.98) [t(79)¼�2.25, p< 0.05]. Nevertheless,sex was not included as a variable in theanalyses below because there were fewmen in the sample (n¼ 23) compared towomen (n¼ 117), and there were no sexdifferences on any of the other measures.2

Correlations

The upper panel of Table 1 summarizes thecorrelations among all of the variablesincluded in Study 1. Consistent with self-categorization theory (e.g., Turner, 1985;Turner et al., 1987, 1994), participants whopossessed high self-esteem reported highhumanity-esteem. In addition, a positiveattachment self-model was associated withlow levels of attachment anxiety. Similarly,having a positive attachment other-modelwas associated with low levels of attachment

2. In both studies, similar results emerged when menwere excluded from the analyses. However, theseresults should be interpreted with caution becausethere were few men compared to women in bothstudies. This issue is discussed further in the GeneralDiscussion.

Table 1. Correlations Between the Parental Treatment, Attachment Models, and EsteemMeasures

SE HE MS MO AN AV FT MT

Study 1SE — 0.38** 0.53** 0.10 �0.60** �0.19yHE — 0.27** 0.28** �0.08 �0.25*MS — 0.09 �0.56** �0.27*MO — 0.15 �0.57**AN — 0.22*AV —

Study 2SE — 0.40** 0.37** 0.16* �0.54** �0.29** 0.17* 0.09HE — 0.12 0.27** �0.29** �0.33** 0.13y 0.13yMS — 0.07 �0.69** �0.25** 0.17* 0.12MO — �0.10 �0.64** 0.12 0.16*AN — 0.28** �0.20** �0.07AV — �0.16* �0.21**FT — 0.39**MT —

Note. SE¼Self-Esteem; HE¼Humanity-Esteem; MS¼Model of Self; MO¼Model of Other; AN¼Anxiety

Dimension; AV¼Avoidance Dimension; FT¼Father Treatment; MT¼Mother Treatment. In Study 1, for all of

the correlations involving AN and AV, N¼ 81. N¼ 140 for the remaining correlations. In Study 2, N¼ 179–183

because not all of the participants completed all of the measures.

yp< 0.10. *p< 0.05. **p< 0.01.


avoidance. Also, consistent with our hypo-theses, participants who indicated a positiveattachment self-model possessed high self-esteem, and participants who indicatedlow levels of attachment anxiety possessedhigh self-esteem. In addition, participantswho indicated a positive attachment other-model possessed high humanity-esteem, andparticipants who indicated low levels ofattachment avoidance possessed highhumanity-esteem.

Partial correlations

We also hypothesized that the attachmentself-model (or anxiety dimension) andattachment other-model (or avoidancedimension) would independently predictself-esteem and humanity-esteem, respec-tively. To test these hypotheses, we com-puted partial correlation coefficients thatcontrolled for either the attachment self-model or the attachment other-model. Whencontrolling for the attachment other-model,the correlation between the attachment self-model and self-esteem remained the same[r(137)¼ 0.53, p< 0.001]. Similarly, whencontrolling for the avoidance dimension, thecorrelation between the anxiety dimensionand self-esteem remained significant [r(78)¼�0.59, p< 0.001]. In addition, when control-ling for the attachment self-model, the corre-lation between the attachment other-modeland humanity-esteem remained significant[r(137)¼ 0.26, p< 0.01]. Likewise, when con-trolling for the anxiety dimension, the correla-tion between the avoidance dimension andhumanity-esteem remained significant[r(78)¼ �0.24, p< 0.05].

Discussion

Our principal novel finding was thathumanity-esteem was correlated with theattachment other-model and attachmentavoidance. In the past, research has exam-ined the relation between the attachmentother-model and interpersonal orientation(e.g., interpersonal warmth, sociability;Bartholomew & Horowitz, 1991; Griffin &Bartholomew, 1994a, 1994b) or a specifictype of evaluation, such as trust (Collins &

Read, 1990; Cozzarelli et al., 2000). How-ever, the present study examined the relationbetween the attachment other-model andoverall favorability or unfavorability towardothers. The results indicated that theattachment self-model (attachment anxiety)uniquely predicted self-esteem over andabove the attachment other-model (attach-ment avoidance) and that the attachmentother-model (attachment avoidance) uniquelypredicted humanity-esteem over and abovethe attachment self-model (attachmentanxiety). These findings are consistent withBartholomew andHorowitz’s (1991) proposalthat the two internal working models aredistinct and with Bowlby’s (1973) statementthat they are logically independent. Theyare also consistent with Bowlby’s (1973)contention that internal working models ofattachment are used as a basis for expecta-tions about the self and others in general.

Study 2

In Study 2,we examinedBowlby’s (1969,1973)theory regarding the development of beliefsand feelings about the self and others.Specifically, we were interested in whetherpositive perceptions of parental treatmentwould predict self-esteem and humanity-esteem. Past research has found that peoplewho feel accepted by their parents possesspositive evaluations of the self and positivebeliefs about others (e.g., Coopersmith,1981; Gecas & Schwalbe, 1986; Rosenberg,1989; Stack, 1972, as cited in Wrightsman,1992; Tafarodi & Swann, 1995). Therefore,we anticipated that positive parental treat-ment would predict high self-esteem andhumanity-esteem.

More important, we expected that theattachment model of the self would mediatethe relation between parental treatment andself-esteem and that the attachment modelof others wouldmediate the relation betweenparental treatment and humanity-esteem(Figure 1a). We predicted that the directrelations between parental treatment andself-esteem (Figure 1a; paths c, g) andhumanity-esteem (Figure 1a; paths e, j)would become nonsignificant after the


j

c

b

Other-modelAvoidancedimension

Portion 2

Mothertreatment

Other-model ofattachment

v1 v2 v1 v2

v1 v2

d g

h i

a

e

Self-modelAnxiety

dimension

Portion 1

f

Fathertreatment

Self-model ofattachment

Self-esteem

v1 v2

Humanity-esteem

a

b

0.45* –0.16 0.09

–0.16 1.12*

0.46*

–0.01

0.18 0.61*

0.28

Self-modelAnxiety

dimension

Other-modelAvoidancedimension

Portion 1 Portion 2

–0.04

Fathertreatment

Self-model ofattachment Self-esteem

Mothertreatment

Other-model ofattachment

Humanity-esteem

v1 v2 v1 v2

v1 v2

Figure 1. (a) Pictorial representation of the full structural model. The paths from fathertreatment (mother treatment, self-esteem) to v1 and v2, respectively, represent the sum ofthe first and second half of the items included in the father treatment subscale (Tafarodi &Swann, 1995; mother treatment subscale, Tafarodi & Swann, 1995; self-regard subscale,Fleming & Courtney, 1984). (b) Full structural model (N¼ 173, *p< 0.05). The paths fromfather treatment (mother treatment, self-esteem) to v1 and v2, respectively, represent thesum of the first and second half of the items included in the father treatment subscale(Tafarodi & Swann, 1995; mother treatment subscale, Tafarodi & Swann, 1995; self-regardsubscale, Fleming & Courtney, 1984). The loadings were set to the value of 1 for oneindicator of every scale, but both loadings were strong (i.e., > 0.50) and significant ineach and every case (ps< 0.05). Goodness-of-fit indices: w2(43)¼ 563.96; standardizedroot-mean-square residual (SRMR)¼ 2.81; root-mean-square error of approximation(RMSEA)¼ 0.27.


relations between parental treatment and theattachment models of self and others and therelations between the attachment models ofself and others and self-esteem and humanity-esteem are controlled (Figure 1a; paths a, b, d,f, h, i). This last result should occur if, asreflected in our opening quote, attachmentmodels are the vehicles through which earlyrelationships influence later judgments andbehavior (Bowlby, 1973). Experiences withparents are thought to influence beliefsregarding our own worthiness of care andexpectations for responsiveness from care-givers, and these expectations should, inturn, predict global self-views and views ofothers.

Prior evidence supports three essentialcomponents of this prediction. First, paren-tal treatment predicts the development ofattachment models (e.g., Carnelley &Janoff-Bulman, 1992; Carnelley et al.,1994; Cassidy, Kirsh, Scolton, & Parke,1996; Hazan & Shaver, 1990). Second, con-sistent with theories that relationships withothers affect people’s attitudes towardthemselves and others (e.g., Bowlby, 1969,1973; Erikson, 1995; Wrightsman, 1992; seeCoopersmith, 1981 for a review), parentaltreatment predicts self-esteem (e.g., Cooper-smith, 1981; Gecas & Schwalbe, 1986;Rosenberg, 1989; Tafarodi & Swann, 1995)and trust in others (Stack, 1972). Third,Study 1 revealed that the attachment modelspredict self-esteem and humanity-esteem.What remains to be shown is that (a) parentaltreatment predicts self-esteem and humanity-esteem and (b) the relations between parentaltreatment and self-esteem and humanity-esteem become nonsignificant when therelations between parental treatment and theattachment models and the relations betweenthe attachment models and the target esteemvariable (i.e., self-esteem or humanity-esteem)are controlled. These results would indicatethat the attachment models mediate therelations between parental treatment andself-esteem and humanity-esteem (Baron &Kenny, 1986).

An interesting issue was whether fathertreatment and mother treatment wouldexhibit different relations with the attach-

ment models, self-esteem, and humanity-esteem. Some evidence indicates that fathertreatment plays a more significant role thanmother treatment in the development of asecure attachment style (Levy, Blatt, &Shaver, 1998; Maio, Fincham, & Lycett,2000), positive friendships (Youngblade &Belsky, 1992), and social competence (Rice,Cunningham, & Mitchell, 1997). On theother hand, researchers have also foundthat one’s mother plays a more significantrole than one’s father in predicting adultattachment styles (Carnelley & Janoff-Bulman, 1992), positive adult romanticrelationship functioning (Carnelley et al.,1994), positive friendships in preschoolchildren (Park & Waters, 1989), and trustin young children (Stack, 1972). Giventhe conflicting evidence, our analysis offather and mother treatment was explora-tory, and we formed no predictions aboutdifferences in their relations with the attach-ment models, self-esteem, and humanity-esteem.

Method

Participants and procedure

Participants were 91 female undergraduateswho received course credit and 92 high-school students (46 women, 44 men, and 2who did not indicate their gender) whovolunteered as part of a classroom demon-stration of social psychology research. Allparticipants completed the same scales asthose in Study 1, except that the self-regardsubscale of the Self-Rating Revised Scale(Fleming & Courtney, 1984) was used asthe only measure of self-esteem. All ofthese measures were reliable (as> 0.80). Inaddition, participants completed the Paren-tal Treatment Questionnaire (Tafarodi &Swann, 1995) which assesses perceptions ofparental support and acceptance. Sevenadditional participants were omitted fromthe analyses because they anticipated thenature of the relations between all of thevariables examined in this study whenprobed for suspicion.


Measures

Parental treatment. The Parental Treat-ment Questionnaire (Tafarodi & Swann,1995) contains 12 items that tap respon-dents’ perceptions of parental support andacceptance during childhood. The question-naire contains two 6-item subscales. Onesubscale reflects perceptions of how respon-dents’ mothers treated them during child-hood, whereas the other subscale assessesrespondents’ perceptions of how theirfathers treated them during childhood.Example items include ‘‘When I experi-enced difficulties as a child, my mother(father) was very supportive’’ and ‘‘Mymother (father) never seemed to care muchhow I was doing in my endeavors’’ (reverse-scored). Participants responded to eachitem using a five-point scale ranging from�2 (‘‘strongly disagree’’) to þ2 (‘‘stronglyagree’’). After reverse scoring the negativelyworded items, total scores for each subscalewere derived by averaging the responsesacross all of the items in the subscale. Theinternal consistency was high for the fathertreatment subscale (a¼ 0.76) and mothertreatment subscale (a¼ 0.85), and the sub-scales revealed that participants feltaccepted and supported by their fathers(M¼ 0.92, SD¼ 0.95) and mothers(M¼ 1.14, SD¼ 0.85).3

Results

A 2 (sex of participant: women and men)�2 (sample type: university students andhigh-school students) analysis of variancecomputed on all of the measures revealedno main effects or interaction involving

sex of the participant or sample type.Consequently, sex and sample type werenot included as factors in the analyses below.

Correlations and partial correlations

The lower panel of Table 1 shows that, con-sistent with Study 1, high humanity-esteem,a positive attachment self-model, and lowlevels of attachment anxiety were associatedwith high self-esteem. Also, a positive attach-ment other-model and low levels of attach-ment avoidance were associated with highhumanity-esteem. In addition, the attach-ment self-model was significantly correlatedwith attachment anxiety, and the attachmentother-model was significantly correlatedwith attachment avoidance.

Also, consistent with Study 1, the corre-lation between the attachment self-modeland self-esteem remained significant whenwe controlled for the attachment other-model [r(178)¼ 0.36, p< 0.001]. Similarly,the correlation between the anxiety dimen-sion and self-esteem remained significantwhen we controlled for the avoidance dimen-sion [r(180)¼�0.50, p< 0.001]. In addition,the correlation between the attachmentother-model and humanity-esteem remainedsignificant when we controlled for theattachment self-model [r(178)¼ 0.26, p<0.001]. Likewise, the correlation betweenthe avoidance dimension and humanity-esteem remained significant when we con-trolled for the anxiety dimension [r(180)¼�0.27, p< 0.001].4

3. A subset of participants (n¼ 95) also completedmeasures of parental acceptance andencouragement of independence (Epstein, 1960).In general, Epstein’s (1960) scales yielded resultssimilar to those achieved using Tafarodi andSwann’s (1995) scales. If we included Epstein’smeasure in the structural equation models, wewould have had very low power to compute theseanalyses. Therefore, we only discuss Tafarodi andSwann’s scales in the results.

4. Supplementary analyses examined the extent towhich self-esteem accounts for unique varianceindependent of humanity-esteem in the attachmentself-model and the extent to which humanity-esteemaccounts for unique variance independent of self-esteem in the attachment model of others. Most ofthese partial correlations were significant(rs<�0.60, p< 0.001 or rs> 0.22, p< 0.01). Thesefindings indicate that the correlations between self-esteem and the attachment self-model (anxietydimension) are not confounded with humanity-esteem and that the correlations between humanity-esteem and the attachment other-model (avoidancedimension) are not confounded with self-esteem.


Finally, the correlations with percep-tions of parental treatment supported thecontention that parental treatment predictsattachment models of the self and othersand favorability toward the self and othersin general. Specifically, participants whoreported positive father treatment indicatedhigh self-esteem, a positive attachment self-model, low levels of attachment anxiety,and low levels of attachment avoidance[all rs(179)> 0.16, ps< 0.05]. On the otherhand, participants who reported positivemother treatment indicated high humanity-esteem, a positive attachment other-model,and low levels of attachment avoidance (allrs(179)> 0.12, ps< 0.05 using the lopsidedtest of significance [Abelson, 1995]).

Mediation

Given that perceptions of parental treat-ment were correlated with the attachmentmodels, self-esteem, and humanity-esteemand that the attachment models were corre-lated with self-esteem and humanity-esteem, we tested our hypothesis that theattachment models of self and others med-iate the relationship between early attach-ment experiences and evaluations of the selfand others. Because we included two differ-ent measures of the attachment models ofself and others, we used structural equationmodeling to examine the direct and indirectrelations between parental treatment andself-esteem and humanity-esteem. In ourstructural equation models, we divided theitems from the father treatment subscaleof the Parental Treatment Questionnaire(Tafarodi & Swann, 1995), mother treat-ment subscale of the Parental TreatmentQuestionnaire (Tafarodi & Swann, 1995),and self-regard subscale of the Self-RatingRevised Scale (Fleming & Courtney, 1984)into two equal groups. The items in thesegroups were then summed to produce twoobserved variables that loaded onto latentfactors representing father treatment, mothertreatment, and self-esteem, respectively (seeFletcher, Simpson, Thomas, & Giles, 1999for a similar procedure). The attachmentself-model and the anxiety dimension were

used as joint indicators of the latent self-model. The attachment other-model and theavoidance dimension were used as jointindicators of the latent other-model. Thetwo portions of the Humanity-Esteem Scaleserved as indicators of humanity-esteem.

We initially examined the full structuralmodel: father treatment and mother treat-ment served as the independent variables,the attachment self-model and attachmentother-model served as the mediating vari-ables, and self-esteem and humanity-esteemserved as the dependent variables. However,as shown in Figure 1b, this model did notyield a good fit [�2(43, N¼ 173)¼ 562.96,p< 0.001, standardized root-mean-squareresidual (SRMR)¼ 2.81, root-mean-squareerror of approximation (RMSEA)¼ 0.27].According to Hu and Bentler (1999), fit isgood when the SRMR is near 0.08 and theRMSEA is near 0.06. This lack of fit mayhave occurred because the full model con-tained too many latent variables for our sizeof sample and number of indicators. Con-sequently, we used two separate structuralequation models, one for each of the attach-ment models and the corresponding esteemvariables.

First, we examined the structural equa-tion model for the attachment self-modeland self-esteem. A model that includedseparate latent factors to represent fathertreatment and mother treatment yieldedgood fit on the SRMR index [�2(14,N¼ 173)¼ 39.40, p< 0.001, SRMR¼ 0.00],although the value of RMSEA (0.10) did notreach 0.06. In addition, mother treatmentwas not significantly related to the attach-ment self-model (�¼�0.06, ns). Therefore,we computed a separate structural equationmodel that included only father treatmentas a predictor of the attachment self-modeland self-esteem. As shown in Figure 2, thismodel yielded optimal fit [�2(6, N¼ 173)¼3.25, ns, SRMR¼ 0.00, RMSEA¼ 0.00].Positive father treatment was associatedwith a more positive attachment model ofthe self, which predicted high self-esteem,and the direct relation between father treat-ment and self-esteem became marginallysignificant when the self-model served as a


mediator. Furthermore, when we removedthe path from the attachment self-model toself-esteem and compared it to the fit of themodel with this path, the mediation modelwas significantly better [w2D(1)¼ 48.66,p< 0.001]. In addition, consistent withBaron and Kenny’s (1986) criteria formediation, the path coefficient from fathertreatment to self-esteem was substantiallyreduced (from 0.29 to 0.09) when the attach-ment self-model was entered as a mediatingvariable (z¼ 1.92, p< 0.06).

Next, we examined the structural equa-tion model for the attachment other-modeland humanity-esteem. The model thatincluded two separate latent factors torepresent father treatment and mothertreatment again yielded good fit on theSRMR index [�2(14, N¼ 173)¼ 45.81,p< 0.001, SRMR¼ 0.01], although thevalue of RMSEA (0.12) did not reach 0.06.

In addition, father treatment was not signif-icantly related to the attachment other-model (�¼ 0.08 ns). Therefore, we com-puted a separate structural equation modelin which mother treatment served as theonly predictor of the attachment model ofother and humanity-esteem. As shown inFigure 3, this model yielded excellent fit[�2(6, N¼ 173)¼ 5.66, ns, SRMR¼ 0.01,RMSEA¼ 0.00]. Positive mother treatmentwas associated with a positive attachmentmodel of others, which predicted higherhumanity-esteem, and the direct relationbetween mother treatment and humanity-esteem was nonsignificant when the other-model served as a mediator. Furthermore,when we removed the path from theattachment other-model to humanity-esteem and compared it to the fit of themodel with this path, the mediation modelwas significantly better, [w2D(1)¼ 18.36,

0.24* 0.60*

0.09†

Father treatmentSelf-model ofattachment

Self-esteem

Self-model Anxiety dimension

v1 v2v1 v2

Figure 2. Structural model for the mediating role of the self-model of attachment on therelation between father treatment and self-esteem (N¼ 173, yp< 0.10, *p< 0.05). Thepaths from father treatment (self-esteem) to v1 and v2, respectively, represent the sum ofthe first and second half of the items included in the father treatment subscale (Tafarodi &Swann, 1995; self-regard subscale, Fleming & Courtney, 1984). The loadings were set tothe value of 1 for one indicator of every scale, but both loadings were strong (i.e., > 0.50)and significant in each and every case (ps< 0.05). Goodness-of-fit indices: w2(6)¼ 3.25;standardized root-mean-square residual (SRMR)¼ 0.00; root-mean-square error ofapproximation (RMSEA)¼ 0.00.


p< 0.001]. In addition, consistent withBaron and Kenny’s (1986) criteria for med-iation, the path coefficient from mothertreatment to humanity-esteem was substan-tially reduced (from 0.27 to 0.05) when theattachment other-model was entered as amediating variable (z¼ 2.08, p< 0.05).5

Discussion

As in Study 1, the attachment model of self(and the anxiety dimension) uniquelypredicted self-esteem, and the attachment

model of others (and the avoidance dimen-sion) uniquely predicted humanity-esteem.However, the principal aims of Study 2were to examine the relations betweenearly attachment experiences, attachmentmodels, self-esteem, and humanity-esteemand to test whether the attachment modelsof self and others mediate the relationsbetween early attachment experiences andesteem for the self and others. The struc-tural equation models supported our pre-diction that the relations between parentaltreatment and esteem for the self and othersare mediated by the attachment models ofself and others. It is interesting, however,that the treatment from one’s father wasthe primary predictor of the self-modeland self-esteem, whereas treatment fromone’s mother was the primary predictor ofthe other-model and humanity-esteem.These findings are important because theysuggest that each parent has a differentimpact on beliefs about the self and others,

5. Two additional structural models were examined.The models were similar to the ones presented inFigures 2 and 3, except that both attachmentmodels were used as simultaneous mediators ofthe effects of father treatment and mothertreatment on self-esteem and humanity-esteem.Regardless of whether father treatment or mothertreatment was the sole exogenous variable, thesemodels yielded weaker fit than the original modelspresented in Figures 2 and 3.

0.27* 0.40*

0.05

Mother treatmentOther-model of

attachmentHumanity-esteem

Other-model Avoidance dimension

v1 v2Portion 1 Portion 2

Figure 3. Structural model for the mediating role of the other-model of attachment onthe relation between mother treatment and humanity-esteem (N¼ 173, *p< 0.05). Thepaths from mother treatment to v1 and v2, respectively, represent the sum of the firstand second half of the items included in the mother treatment subscale (Tafarodi & Swann,1995). The loadings were set to the value of 1 for one indicator of every scale, but bothloadings were strong (i.e., > 0.50) and significant in each and every case (ps< 0.05).Goodness-of-fit indices: w2(6)¼ 5.66; standardized root-mean-square residual (SRMR)¼0.01; root-mean-square error of approximation (RMSEA)¼ 0.00.


and this possibility is discussed in furtherdetail in the General Discussion.

General Discussion

Across two studies, we replicate pastresearch by demonstrating that the attach-ment self-model reliably predicts a globalevaluation of the self (self-esteem; Bartho-lomew & Horowitz, 1991; Collins & Read,1990; Griffin & Bartholomew, 1994a, 1994b;Murray et al., 2001). In addition, we extendpast research by demonstrating that theattachment other-model reliably predicts aglobal evaluation of people in general.

The data in both studies enabled us toexamine the relation between the attach-ment self-model and self-esteem and therelation between the attachment other-model and humanity-esteem, while control-ling for the attachment self-model. Theseresults are best summarized using a singlestructural equation model for the partici-pants who completed the self-regard sub-scale of the Self-Rating Revised Scale(Fleming & Courtney, 1984) across bothstudies. This model includes a path fromthe attachment self-model to self-esteemand a path from the attachment other-model to humanity-esteem. In addition, themodel includes a path from the attachmentself-model to humanity-esteem, becauseself-categorization theory (e.g., Turner, 1985;Turner et al., 1987, 1994) predicts that anevaluation of humanity includes an evaluationof the self (see Figure 4). This model yielded anear-adequate fit [w2(14, N¼ 254)¼ 36.49,p< 0.001, SRMR¼ 0.02, RMSEA¼ 0.08](Hu and Bentler, 1999, recommend thatRMSEA be ‘‘near’’ 0.06.) More important,the paths from (a) the attachment self-modelto self-esteem, (b) the attachment other-modelto humanity-esteem, and (c) the attachmentself-model to humanity-esteem were signifi-cant (see Figure 4).

Also, consistent with past research (e.g.,Coopersmith, 1981; Gecas & Schwalbe,1986; Rosenberg, 1989; Tafarodi & Swann,1995), we found that perceptions of fathertreatment predict the attachment self-modeland self-esteem. A novel finding was that

perceptions of mother treatment predictthe attachment other-model and humanity-esteem. It is interesting that father treat-ment and mother treatment were differentlyrelated to the attachment models, self-esteem,and humanity-esteem. In the past, researchershave found that fathers are influential in thedevelopment of a secure attachment style(Maio et al., 2000), positive friendships(Youngblade & Belsky, 1992), and socialcompetence (Rice et al., 1997). However,other researchers have found that mothersplay a more significant role in the predictionof attachment styles, relationship functioning,and positive friendships (Carnelley & Janoff-Bulman, 1992; Carnelley et al., 1994; Park &Waters, 1989).

It is possible that our findings occurredbecause parental treatment may reflect theparents’ own beliefs about themselves andothers. Past research has shown that mentraditionally have more positive specificbeliefs about themselves than do women(e.g., Beyer, 1990; Marsh, 1990) and thatwomen have more positive beliefs aboutothers than do men (e.g., Luke, & Maio,2004; Wrightsman, 1992). These differentialbeliefs are consistent with the finding thatmen see themselves as separate from others,whereas women see themselves as connectedto others (see Cross & Madson, 1997 for areview). There is also evidence that thisdifferential emphasis on the individualversus community emerges partly from tra-ditional sex roles (Eagly, 1987). As a resultof their sex-role learning, fathers’ treatmentof their children may reflexively highlightthe importance of the self within the contextof a relationship, whereas mothers’ treat-ment of their children may highlight theimportance of others within the context ofa relationship. For example, fathers mighttend to ask their children how theyresponded to bullying at school (e.g., ‘‘Didyou stick up for yourself?’’), whereasmothers might tend to ask their childrenabout how others responded to bullying(e.g., ‘‘Did the teacher intervene?’’). Thesequestions may shape the attachment modelsof the self and others and evaluations ofoneself and others outside the relationship


context. These post-hoc explanations for ourfindingsare speculative. Itwill be important forfuture research to replicate these findings andexamine more closely the different roles thatmothers and fathers may play in the develop-ment of self and other models of attachment.

Another possibility is that the difference inthe availability of fathers and mothers isresponsible for the separate effects of fatherandmother treatment.Mothers may form thebasis for the model of others because they aretypically the most important ‘‘other’’ frombirth onward. In fact, past research hasshown that, on average fathers spend lesstime with their offspring than mothers(Lamb & Oppenheim, 1989). Furthermore,

when fathers are available, they are morelikely to play with their offspring than engagein caretaking functions (Lamb, 1977, 1981;Yogman, 1982). Because fathers in generalare less likely to be consistently available, chil-dren may believe that the availability of theirfather may have something to do with theirown behavior and use this to draw inferencesabout the self.6 Future research shouldexamine this possibility.

There are a few methodological issuesthat should be considered when interpretingthese results. First, although we obtained no

6. We thank an anonymous reviewer for mentioningthis possibility.

–0.31*

–0.21*

Avoidancedimension

Other-model

0.26*

0.67*Self-model ofattachment Self-esteem

Anxietydimension

v1 v2

Other-modelof attachment

Humanity-esteem

Portion 1

Self-model

Portion 2

Figure 4. Structural model for the self-model of attachment as a predictor of self-esteemand both attachment models as simultaneous predictors of humanity-esteem (N¼ 254,*p< 0.05). The paths from self-esteem to v1 and v2, respectively, represent the sum of thefirst and second half of the items included in self-regard subscale (Fleming & Courtney,1984). The variance for each factor was set to the value of 1, but both loadings were strong(i.e., >0.50) and significant in each and every case (ps< 0.05). In this analysis, high scoresfor the other-model and humanity-esteem represent low levels of each construct (i.e.,scores computed from the Bartholomew and Horowitz, 1991, measure loaded negatively,whereas scores computed from the Fraley et al. (2000) measure loaded positively on theother-model latent variable and two portions of the Humanity-Esteem Scale loadednegatively on the humanity-esteem latent variable). Goodness-of-fit indices:w2(14)¼ 36.49; standardized root-mean-square residual (SRMR)¼ 0.02; root-mean-square error of approximation (RMSEA)¼ 0.08.


significant differences between men andwomen across both studies, it is difficult todetermine whether our findings are limitedto women only because the majority of par-ticipants were women. Despite similar pat-terns of correlations for men and women(see Tables 2 and 3), it may be possible thatthe relations between the variables exam-ined in these studies are different for menthan women. In fact, because past researchhas found that people describe an opposite-sex parent more favorably than a same sexparent (Hazan & Shaver, 1987) and that theopposite-sex parent has more of aninfluence on self-esteem than the same sexparent (Winefield, Goldney, Tiggeman, &Winefield, 1990), it is possible that fathertreatment and mother treatment are differ-entially related to the attachment models,self-esteem, and humanity-esteem for theseparate sexes. Future research shouldexamine this possibility by using samples

that are more equal in their distribution ofmen and women.

Second, although our results are consis-tent with Bowlby’s (1973) theory regardingthe influence of early attachment experi-ences, the data are correlational and donot establish causality. Related to thisissue, the measure of parental treatmentis a retrospective recollection of feelingaccepted and supported by parents, whichmay or may not reflect actual parentalbehavior experienced in childhood (Tafar-odi & Swann, 1995). Furthermore, we con-ducted additional analyses testing twomodels that were similar to the models pre-sented in Figures 2 and 3, except that theesteem measures served as the mediator(rather than the outcome variable) and theattachment measures served as the criterion(rather than the mediator). These modelsexhibited the same degree of fit as the origi-nal models, which were more strongly based

Table 2. Correlations Between the Parental Treatment, Attachment Models, and EsteemMeasures (Women Only)


Study 1SE — 0.40** 0.55** 0.13 �0.63** �0.20yHE — 0.25* 0.23** 0.01 �0.06MS — 0.06 �0.55** �0.20yMO — 0.14 �0.61**AN — 0.11AV —

Study 2SE — 0.33** 0.39** 0.15y �0.55** �0.27** 0.16y 0.14HE — 0.19* 0.30** �0.32** �0.33** 0.20* 0.18*MS — 0.12 �0.75** �0.37** 0.22* 0.22*MO — �0.17y �0.62** 0.18* 0.21*AN — 0.35** �0.20* �0.15yAV — �0.23* �0.28**FT — 0.44**MT —



the correlations involving AN and AV, N¼ 69. N¼ 117 for the remaining correlations. In Study 2, N¼ 137.

yp< 0.10. *p< 0.05. **p< 0.01.


on extant theory (see Introduction). Despitethe theoretical basis for accepting the for-mer models, the success of the alternativemodels suggests that different relationsamong these variables may operate simulta-neously. For example, it is quite possiblethat a positive attachment model of othersfacilitates the affective, cognitive, and beha-vioral experience of high humanity-esteem,which in turn sets the stage for more posi-tive attachments to others in relationships.To disentangle the causal relationships,researchers could manipulate each of thesevariables to examine these effects by prim-ing positive or negative parental treatment,attachment models of the self and others,and/or global views of the self and others(see Rowe & Carnelley, 2003 for a discus-sion of this procedure).

It is interesting that the associationsbetween self-esteem and the attachmentself-model and the anxiety dimension weregenerally stronger than the associationsbetween the humanity-esteem and the

attachment other-model and the avoidancedimension. These results are consistentwith prior suggestions that it is easier tooperationalize the attachment self-modelthan the attachment other-model (Griffin &Bartholomew, 1994a; Klohnen & John,1998). Fraley et al. (2000) have taken a steptoward adequately assessing the other-model, because they computed an item-response analysis on all available attachmentquestionnaires (Brennan et al., 1998; Collins& Read, 1990; Griffin & Bartholomew,1994a; Simpson, 1990) to devise their scale.However, even this approach is limited bythe conceptualization of the model of others.The biggest impediment to the conceptuali-zation of the attachment model of others andglobal evaluations of others is that the‘‘others’’ may vary from close romanticpartners or relatives to strangers, and thisvariation in targets is greater than whatoccurs when focusing on the self as a target.Thus, these models may inherently be moredifficult to assess.

Table 3. Correlations Between the Parental Treatment, Attachment Models, and EsteemMeasures (Men Only)


Study 1SE — 0.45* 0.59** �0.01 �0.73** �0.59*HE — 0.32 0.43** �0.22 �0.65*MS — 0.22 �0.58* �0.51yMO — 0.17 �0.46AN — 0.66*AV —

Study 2SE — 0.59** 0.31* 0.24 �0.55** �0.42** 0.20 �0.12HE — �0.05 0.17 �0.21 �0.29y �0.05 0.03MS — �0.09 �0.54** �0.06 �0.02 �0.25yMO — �0.12 �0.70** �0.16 0.00AN — 0.08 �0.15 0.22AV — 0.08 �0.08FT — 0.11MT —



the correlations involving AN and AV, N¼ 12. N¼ 23 for the remaining correlations. In Study 2, N ¼ 44.

yp< 0.10. *p< 0.05. **p< 0.01.


Regardless of the difficulties capturing theattachment models of self and others, ourdata provide evidence that these modelsmediate the relations between parental treat-ment and self-esteem and humanity-esteem.These analyses are consistent with the pro-posal that attachment models are vital con-structs through which early relationshipsaffect our global views of the self, others,and the world (Bowlby, 1969,1973). Forthis reason, it is important to explore otherpossible antecedents of the attachmentmodels of self and others, self-esteem, andhumanity-esteem, such as the effects of peertreatment. In a review of past literature,Harris (1995) discusses the importance ofpeer groups and downplays the importanceof parents in the development and modifica-tion of a child’s personality. Furthermore,Fraley and Davis (1997) found that peersserved as important attachment figuresamong undergraduates. If this hypothesis iscorrect, peer treatment may be anotherimportant predictor of attachment models,self-esteem, and humanity-esteem.

Future research should also consider thepossibility that humanity-esteem mediatesthe relations between attachment styleand other important social variables. Forexample, past research has shown thatpeople with a secure attachment styleexhibit low amounts of antisocial behavior(e.g., Bowlby, 1944; Rosenstein & Horowitz,1996), and we suggest that humanity-esteemmay mediate this relationship. Indeed, otherresearch has shown that people who commitextreme acts of violence against others tendto be manipulative, show a lack of remorse,and feel loosely connected to others (Hare,1996). Similarly, McHoskey, Worzel, andSzyarto (1998) found that psychopathy was

related to Machiavellianism and narcissismin a nonclinical sample. These findings gohand-in-hand with our findings that peoplewho possess low humanity-esteem feel lessinterconnected with others and are moremanipulative. People low in humanity-esteem are also more willing to discriminateagainst people from other social groups,including groups based on gender, ethnicity,and nationality (Luke &Maio, 2004). There-fore, the links between attachment patternsand antisocial behavior might be partlyattributable to low humanity-esteem, whichstems, in part, from the quality of relation-ships with one’s parents while growing up.Indeed, it is possible that links between manyvariables (e.g., exposure to violence) andantisocial behavior are mediated by theeffects of these variables on humanity-esteem, which is consistent with theoriesthat such behaviors often result from adehumanization of targets (Leyens et al.,2000, 2001). These possibilities underscorethe potential theoretical importance of thehumanity-esteem construct.

The present studies yield valuable knowl-edge about the content of the internal work-ing models by finding that attachmentmodels of the self predict self-esteem inde-pendently of attachment models of othersand that attachment models of others pre-dict humanity-esteem independently ofattachment models of the self. Moreover,the attachment models of self and othersmediate the relations between early attach-ment experiences with parents and self-esteem and humanity-esteem. Together,these results provide new evidence thatattachment experiences and models of selfand others predict how we view ourselvesand human beings as a group.

References

Abelson, R. P. (1995). Statistics as principledargument. Hillsdale, NJ: Erlbaum.

Ainsworth, M. D. S., Blehar, M., Waters, E., &Wall, S. (1978). Patterns of attachment: A psycholo-gical study of the strange situation. Hillsdale,NJ: Erlbaum.

Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psy-chological research: Conceptual, strategic, and

statistical considerations. Journal of Personalityand Social Psychology, 51, 1173–1182.

Bartholomew, K., & Horowitz, L. M. (1991). Attach-ment styles among young adults: A test of a fourcategory model. Journal of Personality and SocialPsychology, 61, 226–244.

Bartholomew, K., & Shaver, P. R. (1998). Methods ofassessing adult attachment: Do they converge? InJ. A. Simpson, & W. S. Rholes (Eds.), Attachment


theory and close relationships (pp. 25–45). London:Guilford Press.

Beyer, S. (1990). Gender differences in the accuracy ofself-evaluations of performance. Journal of Person-ality and Social Psychology, 59, 960–970.

Bowlby, J. (1944). Forty-four juvenile thieves: Theircharacters and homelife. International Journal ofPsychoanalysis, 25, 19–53.

Bowlby, J. (1969). Attachment and loss (Vol. 1):Attachment. New York: Basic Books.

Bowlby, J. (1973). Attachment and loss (Vol. 2):Separation. New York: Basic Books.

Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998).Self-report measurement of adult attachment:An integrative overview. In J. A. Simpson, &W. S. Rholes (Eds.), Attachment theory and closerelationships (pp. 46–76). London: Guilford Press.

Cantril, H. (1932). General and specific attitudes.Psychological Monographs, 42 (5, Whole No. 192).

Carnelley, K. B., & Janoff-Bulman, R. (1992). Opti-mism about love relationships: General vs. specificlessons from one’s personal experiences. Journal ofSocial and Personal Relationships, 9, 5–20.

Carnelley, K. B., Pietromonaco, P. R., & Jaffe, K.(1994). Depression, working models of others,and relationship functioning. Journal of Personalityand Social Psychology, 66, 127–140.

Cassidy, J., Kirsh, S. J., Scolton, K. L., & Parke, R. D.(1996). Attachment and representations of peer rela-tionships. Developmental Psychology, 32, 892–904.

Collins, N. L., & Read, S. J. (1990). Adult attachment,working models, and relationship quality in datingcouples. Journal of Personality and Social Psychol-ogy, 58, 644–663.

Coopersmith, S. (1981). Antecedents of self-esteem.Palo Alto, CA: Consulting Psychologists Press.(Original work published 1967)

Cozzarelli, C., Hoekstra, S. J., & Bylsma, W. H.(2000). General versus specific mental models ofattachment: Are they associated with different out-comes? Personality and Social Psychology Bulletin,26, 605–618.

Cross, S. E., & Madson, L. (1997). Models of the self:Self-construal and gender. Psychological Bulletin,122, 5–37.

Eagly, A. H. (1987). Sex differences in social beha-vior: A social role interpretation. Hillsdale, NJ:Erlbaum.

Eagly, A. H., & Chaiken, S. (1993). The psychology ofattitudes. London: Harcourt Brace Jovanovich.

Epstein, S. (1960). The Mother-Father-Peer Scale.Unpublished manuscript. University of Massachu-setts at Amherst.

Erikson, E. H. (1995). Childhood and society (2nd ed.).London: Vintage. (Original work published 1963)

Esses, V. M., Haddock, G., & Zanna, M. P. (1993).Values, stereotypes, and emotions, as determinantsof intergroup attitudes. In D. M. Mackie, &D. L. Hamilton (Eds.), Affect, cognition, andstereotyping: Interactive processes in group perception(pp. 137–166). New York: Academic Press.

Esses, V. M., & Maio, G. R. (2002). Expanding theassessment of attitude components and structure:The benefits of open-ended measures. InW. Stroebe, & M. Hewstone (Eds.), Europeanreview of social psychology (Vol. 12, pp. 71–102).Chichester, UK: Wiley.

Feeney, J. A., & Noller, P. (1990). Attachment styleas a predictor of adult romantic relationships.Journal of Personality and Social Psychology, 58,281–291.

Fleming, J. S., & Courtney, B. E. (1984). The dimen-sionality of self-esteem. II. Hierarchical facetmodel for revised measurement scales. Journal ofPersonality and Social Psychology, 46, 404–421.

Fleming, J. S., & Watts, W. A. (1980). The dimension-ality of self-esteem: Some results from a collegesample. Journal of Personality and Social Psychol-ogy, 39, 921–929.

Fletcher, G. J. O., Simpson, J. A., Thomas, G., &Giles, L. (1999). Ideals in intimate relationships.Journal of Personality and Social Psychology, 76,72–89.

Fraley, R. C., & Davis, K. E. (1997). Attachmentformation and transfer in young adults’ closefriendships and romantic relationships. PersonalRelationships, 4, 131–144.

Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000).An item response theory analysis of self-reportmeasures of adult attachment. Journal of Person-ality and Social Psychology, 78, 350–365.

Gecas, V., & Schwalbe, M. L. (1986). Parentalbehavior and adolescent self-esteem. Journal ofMarriage and the Family, 48, 37–46.

George, C., Kaplan, N., & Main, M. (1984). Adultattachment interview. Unpublished manuscript.University of California at Berkeley.

Griffin, D. W., & Bartholomew, K. (1994a). Themetaphysics of measurement: The case of adultattachment. In K. Bartholomew, & D. Perlman(Eds.), Advances in personal relationships (Vol. 5,pp. 17–52). London: Jessica Kingsley Publishers.

Griffin, D. W., & Bartholomew, K. (1994b). Models ofself and other: Fundamental dimensions underly-ing measures of adult attachment. Journal ofPersonality and Social Psychology, 67, 430–445.

Gurung, R. A., Sarason, B. R., & Sarason, I. G.(2001). Predicting relationship quality and emo-tional reactions to stress from significant-other-concept clarity. Personality and Social PsychologyBulletin, 27, 1267–1276.

Hare, R. D. (1996). Psychopathology: A clinicalconstruct whose time has come. Criminal Justiceand Behavior, 23, 25–54.

Harris, J. R. (1995). Where is the child’s environment?A group socialization theory of development.Psychological Review, 102, 458–489.

Hazan, C., & Shaver, P. R. (1987). Romantic love asconceptualized as an attachment process. Journalof Personality and Social Psychology, 52, 511–524.

Hazan, C., & Shaver, P. R. (1990). Love and work:An attachment–theoretical perspective. Journal ofPersonality and Social Psychology, 52, 511–524.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fitindexes in covariance structure analysis: Conven-tional criteria versus new alternatives. StructuralEquation Modeling, 6, 1–55.

Katz, D., & Stotland, E. (1959). A preliminary state-ment to a theory of attitude structure and change.In S. Koch (Ed.), Psychology: A study of science(Vol. 3, pp. 423–475). New York: McGraw-Hill.

Klohnen, E. C., & John, O. P. (1998). Working modelsof attachment: A theory-based prototype approach.In J. A. Simpson, & W. S. Rholes (Eds.),


Attachment theory and close relationships (pp. 115–140). London: Guilford Press.

Lamb, M. E. (1977). Father–infant and mother–infantinteraction in the first year of life. Child Develop-ment, 48, 167–181.

Lamb, M. E. (1981). The development of infant–fatherrelationships. In M. E. Lamb (Ed.), The role of thefather in child development (pp. 458–488). NewYork: Wiley.

Lamb, M. E., & Oppenheim, D. (1989). Fatherhood andfather–child relationships: Five years of research. InS. H. Cath, A. R. Gurwitt, & J. M. Ross (Eds.),Fathers and their families: Developmental Perspec-tives (pp. 11–26). Hillsdale, NJ: Erlbaum.

Levy, K. N., Blatt, S. J., & Shaver, P. R. (1998). Attach-ment styles and parental representations. Journal ofPersonality and Social Psychology, 74, 407–419.

Leyens, J.-P., Paladino, M.-P., Rodriguez-Torres, R.,Vaes, J., Demoulin, S., Rodriguez-Perez, A., &Gaunt, R. (2000). The emotional side of prejudice:The role of secondary emotions. Personality andSocial Psychology Bulletin, 4, 186–197.

Leyens, J.-P., Rodriguez-Perez, A., Rodriguez-Torres,R., Gaunt, R., Paladino, M. P., Vaes, J., &Demoulin, S. (2001). Psychological essentialismand the differential attribution of uniquely humanemotions to ingroups and outgroups. EuropeanJournal of Social Psychology, 31, 395–411.

Luhtanen,R.,&Crocker, J. (1992).Acollective self-esteemscale: Self-evaluation of one’s social identity. Person-ality and Social Psychology Bulletin, 18, 302–318.

Luke, M. A. (2003). Oh the humanity! Correlates,antecedents, and consequences of humanity-esteem. Unpublished doctoral dissertation, CardiffUniversity, Cardiff, UK.

Luke, M. A., & Maio, G. R. (2004). Oh the humanity!The measurement, manipulation, and social import-ance of humanity-esteem. Manuscript submitted forpublication.

Luke, M. A., Maio, G. R., & Carnelley, K. B. (2002,February). O’ the humanity! The development of theHumanity-Esteem Scale. Poster presented at the 3rdannual meeting of the Society for Personality andSocial Psychology, Savannah, GA.

Main, M., & Goldwyn, R. (1994). Adult attachmentrating and classification system: Manual in draft(Version 6). Unpublished manuscript. Universityof California at Berkeley.

Maio, G. R., Fincham, F. D., & Lycett, E. J. (2000).Attitudinal ambivalence toward parents andattachment style. Personality and Social PsychologyBulletin, 36, 1451–1464.

Marsh, H. W. (1986). Global self-esteem: Its relationto specific facets of the self-concept and theirimportance. Journal of Personality and SocialPsychology, 51, 1224–1236.

Marsh, H. W. (1990). A multidimensional, hierarchicalmodel of self-concept: Theoretical and empiricaljustification. Educational Psychology Review, 81,77–172.

McHoskey, J. W., Worzel, W., & Szyarto, C. (1998).Machiavellianism and psychopathy. Journal ofPersonality and Social Psychology, 74, 192–210.

Mikulincer, M. (1998). Attachment working modelsand sense of trust: An exploration of interactiongoals and affect regulation. Journal of Personalityand Social Psychology, 74, 1209–1224.

Murray, S.L.,Holmes, J.G.,Griffin,D.W.,Bellavia,G.,& Rose, P. (2001). The mismeasure of love: Howself-doubt contaminates relationship beliefs. Person-ality and Social Psychology Bulletin, 27, 423–436.

Park, K. A., & Waters, E. (1989). Security of attach-ment and preschool friendships. Child Develop-ment, 60, 1076–1081.

Rice, K. G., Cunningham, T. J., & Young, M. B.(1997). Attachment to parents, social competence,and emotional well-being: A comparison of blackand white late adolescents. Journal of CounselingPsychology, 44, 89–101.

Robins, R. W., Hendin, H. M., & Trzesniewski, K. H.(2001). Measuring global self-esteem: Constructvalidation of a single-item measure and the Rosen-berg Self-Esteem Scale. Personality and SocialPsychology Bulletin, 27, 151–161.

Rosenberg, M. (1956). Misanthropy and political ideol-ogy. American Sociological Review, 21, 690–695.

Rosenberg, M. (1979). Conceiving the self. Malabar,FL: Krieger.

Rosenberg, M. (1989). Society and the adolescent self-image (Rev. ed.). Middletown, CN: WesleyanUniversity Press.

Rosenberg, M. J., & Hovland, C. I. (1960). Cognitive,affective, and behavioral components of attitudes.In C. I. Hovland, & M. J. Rosenberg (Eds.),Attitude organization and change: An analysis ofconsistency among attitude components (pp. 1–14).New Haven, CT: Yale University Press.

Rosenstein, D. S., & Horowitz, H. A. (1996). Adoles-cent attachment and psychopathology. Journal ofConsulting and Clinical Psychology, 64, 244–253.

Rowe, A., & Carnelley, K. B. (2003). Attachment styledifferences in the processing of attachment-relevantinformation: Primed-style effects on recall,interpersonal expectations and affect. PersonalRelationships, 10, 59–75.

Simpson, J. (1990). The influence of attachment styleon romantic relationships. Journal of Personalityand Social Psychology, 59, 51–64.

Tafarodi, R. W., & Swann, W. B. (1995). Self-Likingand self-competence as dimensions of global self-esteem: Initial validation of a measure. Journal ofPersonality Assessment, 65, 322–342.

Triandis, H. C. (1971). Attitude and attitude change.New York: Wiley.

Turner, J. C. (1985). Social categorization and the self-concept: A social cognitive theory of group behaviour.In E. J. Lawler (Ed.), Advances in group processes(Vol. 2, pp. 77–122). Greenwich, CT: JAI Press.

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D.,& Wetherell, M. S. (1987). Rediscovering the socialgroup: A self-categorization theory. Oxford: BasilBlackwell.

Turner, J. C., Oakes, P. J., Haslam, S. A., &McGarty, C. (1994). Self and collective: Cognitionand social context. Personality and Social Psycho-logy Bulletin, 20, 454–463.

Winefield, H. R., Goldney, R. D., Tiggeman, M., &Winefield, A. H. (1990). Parental rearing beha-viors: Stability of reports over time and relationto adult interpersonal skills. Journal of GeneticPsychology, 15, 211–219.

Wrightsman, L. S. (1991). Interpersonal trust and atti-tudes toward human nature. In J. P. Robinson,P. R. Shaver, & L. S. Wrightsman (Eds.),Measures


of personality and social psychological attitudes.London: Academic Press.

Wrightsman, L. S. (1992). Assumptions about humannature: Implications for researchers and practi-tioners. London: Sage.

Yogman, M. W. (1982). Observations on the father–infant relationship. In S. H. Cath, A. R. Gurwitt, &J. M. Ross (Eds.), Father and child: DevelopmentalPerspectives (pp. 101–122). Boston, MA: Little, Brown.

Youngblade, L. M., & Belsky, J. (1992). Parent–childantecedents of 5-year olds’ close friendships: Alongitudinal analysis. Developmental Psychology,28, 700–713.

Zanna, M. P., & Rempel, J. K. (1988). Attitudes: Anew look at an old concept. In D. Bar-Tal, &A. W. Kruglanski (Eds.), The social psychology ofknowledge (pp. 315–334). Cambridge, UK:Cambridge University Press.


Copyright of Personal Relationships is the property of Wiley-Blackwell and its content may not be copied or

emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.

However, users may print, download, or email articles for individual use.

attachment models of the self and others

Documents