The relation between identity status and romantic attachment style in middle and late adolescence

ARTICLE IN PRESS

Journal ofAdolescence

Journal of Adolescence 29 (2006) 737–748

0140-1971/$30

rights reserved

doi:10.1016/j.

�CorresponE-mail add

(E.T. Rodrigu

www.elsevier.com/locate/jado

The relation between identity status and romantic attachmentstyle in middle and late adolescence

Steven L. Bermana,�, Carl F. Weemsb, Eileen T. Rodriguezc, Irving J. Zamorad

aDepartment of Psychology, University of Central Florida, 1200 W. International Speedway Blvd.,

Daytona Beach, FL 32114, USAbDepartment of Psychology, University of New Orleans, New Orleans, LA 70148, USA

cDepartment of Psychology, New York University, 6 Washington Place, New York, NY 10003, USAdCollege of Medicine, University of Florida, Gainesville, FL 32610, USA

Abstract

This study examined the linkages between identity formation and romantic attachment in an ethnicallydiverse sample of high school (n ¼ 189) and college students (n ¼ 324). Individuals in the foreclosedidentity status group had significantly lower relationship avoidance scores than the diffused identity statusgroup, and the foreclosed group had significantly lower relationship anxiety scores than both the achievedidentity and moratorium groups. Identity status and romantic attachment style were significantly relatedamong the college sample, but not among the high school sample. Some ethnic differences in attachmentstyle were noted. Developmental implications are discussed.r 2005 The Association for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rightsreserved.

Keywords: Identity; Attachment

.00 r 2005 The Association for Professionals in Services for Adolescents. Published by Elsevier Ltd. All

.

adolescence.2005.11.004

ding author. Tel.: +1 386 506 4049; fax: +1 386 506 4098.

resses: [email protected] (S.L. Berman), [email protected] (C.F. Weems), [email protected]

ez), [email protected] (I.J. Zamora).

ARTICLE IN PRESS

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748738

Introduction

The conceptual link between attachment theory, rooted in the work of Bowlby (1982, 1988),and Erikson’s psychosocial development theory can be seen in several of Erikson’s stages (e.g.trust, generativity). One of the most salient links is between the concepts of identity and romanticattachment. This link is inferred in Erikson’s writings on the stages of adolescence and youngadulthood, when the crisis of identity is followed by the crisis of intimacy. ‘‘It is only whenidentity formation is well on its way that true intimacy—which is really a counterpointing as wellas a fusing of identities—is possible’’(Erikson, 1968, p. 135). In other words, those who have notresolved their crisis of identity confusion are likely to either isolate and avoid intimacy altogetheror else make futile, desperate and frantic attempts at intimacy, often with improbable orinappropriate partners. According to Erikson’s theorizing identity is crucial for adult romanticattachment. Thus, the purpose of this study was to empirically examine the link between identitystatus and romantic attachment style by integrating the adult attachment constructs ofBartholomew (1990) with the identity constructs of Marcia (1966).

Identity status

Marcia (1966) has operationalized Erikson’s concept of identity formation as involving twobasic dimensions, exploration and commitment. Exploration is the process by which the individualactively searches for a resolution to the issues of choosing the goals, roles, and beliefs about theworld that provide the individual’s life with direction and purpose. Commitment represents theactual resolution of identity issues such as the selection of an occupation, relationship, groupmembership, religion, etc. Marcia’s four identity statuses are: diffusion, foreclosure, moratorium,and achievement.The diffusion status (low in exploration and commitment) is characterized by individuals who

are not committed to any particular goals, roles, or beliefs about the world and are not activelysearching for ones either. The foreclosure status (low in exploration, but high in commitment) isdescriptive of individuals who lack a period of exploration of alternatives but are neverthelesscommitted. Generally, these commitments represent those goals, roles, and beliefs about theworld suggested by others, often parental figures, and are assumed without being questioned orexamined, and therefore are attained more from a process of modelling rather than through self-reflection. The moratorium status (high in exploration, low in commitment) precedes identityachievement. The individuals in this status experience a ‘‘crisis’’ due to their active exploration ofdifferent options but have not yet chosen from the alternatives. Finally, those individuals who areable to move beyond the moratorium status and choose their goals, roles, and beliefs about theworld are said to be in the achievement status (high in exploration and commitment).

Romantic attachment style

Bartholomew (1990) developed a four-category model of adult romantic attachment style basedon two underlying continuums—relationship anxiety and relationship avoidance. The ‘‘secure’’style is low in both anxiety and avoidance. The ‘‘preoccupied’’ style is high in anxiety and low inavoidance, presenting as overly dependant, jealous, clingy, and insecure. The ‘‘dismissing’’ style,

ARTICLE IN PRESS

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748 739

being high in avoidance, but low in anxiety, seems cold, aloof, and detached from others. Finally,the ‘‘fearful’’ style, high in both anxiety and avoidance, although desiring close relationships,none-the-less avoids them for fear of rejection. The relationship anxiety and relationshipavoidance continuums have also been labelled ‘‘model of self’’ and ‘‘model of other’’, respectively,as relationship anxiety seems to be related to low valuing of the self, and relationship avoidanceseems to be related to low valuing of others (Bartholomew, 1990; Bartholomew & Horowitz, 1991;Brennan, Clark, & Shaver, 1998).

Identity and attachment research

A number of studies have found that individuals who are securely attached to their parents tendto have higher levels of identity development (e.g. Benson, Harris, & Rogers, 1992; Kroger, 1985;Kroger & Haslett, 1988; Lapsley, Rice, & Fitzgerald, 1990; Zimmermann & Becker-Stoll, 2002).However, there has been far less research on whether identity is related to romantic attachment.We are aware of only four studies that speak to this association; however each have limitationsthat preclude firm conclusions. For example, Reich and Siegel (2002) surveyed 161 universitystudents and found that high ego development (as measured across the first six stages of Erikson’stheory) was associated with attachment security. The Eriksonian measure of ego developmentused in this study did not use Marcia’s identity status paradigm, so direct links between identitystatus and romantic attachment could not be inferred. Similarly, in a study of 91 college students,Vogensen (2003) hypothesized that securely attached participants were more likely to be identityachieved. However, only 5 participants scored as identity achieved, but all of them did reporthaving a secure attachment style.Hoegh and Bourgeois (2002) studied 79 undergraduates and did find that identity achieved

individuals showed higher levels of secure attachment and diffused individuals showed higher levels offearful attachment. Individuals in moratorium also scored highest in secure attachment. Foreclosedindividuals tended to have higher scores on either secure or dismissive attachment styles. Specifically,while these two attachment styles differ in relationship avoidance, what they share in common is alack of relationship anxiety, which is typical of the fearful and preoccupied attachment styles.Finally, Kennedy (1999) surveyed 225 college freshmen and found that individuals identified as

preoccupied in attachment style had higher identity diffusion and moratorium scores thandid secure individuals. Fearful individuals also had higher diffusion scores than the secureindividuals. Secure individuals had higher achievement scores than the fearful individuals. Noneof the attachment styles had significantly different foreclosure scores. As we attempt to comparethe results of Kennedy (1999) with those of Hoegh and Bourgeois (2002), we are caught with theproblem that the former used a categorical attachment style variable with continuous identitystatus scores, while the later used a categorical identity status variable with continuous attachmentstyle scores. However, taken together, the results point to linkages between secure attachment andachieved identity, and between fearful attachment and diffused identity.

The present study

Erikson (1968) suggested that one must have a strong sense of identity before one isable to achieve a deeper sense of intimacy. Thus, we might expect to find a relationship

ARTICLE IN PRESS

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748740

between identity status and romantic attachment variables. The four studies cited aboveprovide some empirical support for such a relationship, but contain a number of methodo-logical limitations. This study seeks to expand on these beginning efforts by first determiningif these limited findings can be replicated and second, to expand upon these studies byinvestigating the underlying attachment dimensions of relationship anxiety and relationshipavoidance.Specifically, in this study we predicted that identity achieved and foreclosed individuals (the two

committed statuses) would score significantly lower in relationship avoidance than those inmoratorium and diffusion. Because achieved and foreclosed both have a firmly committed senseof identity, we expected them to feel ready to explore the stage of intimacy, and thus would likelybe pursuing romantic relationships.Relationship anxiety, on the other hand, might be more strongly related to identity

exploration, rather than commitment. This is because those who are more thoughtfulabout identity issues and what they want to achieve out of life are also likely to be morethoughtful about intimacy issues and what they want to achieve out of a relationship.Uncertainty often brings anxiety. We predicted that those in the achieved and moratoriumidentity status would have significantly higher relationship anxiety scores than the diffused andforeclosed.In a direct comparison of identity status with attachment style, we predicted that

identity achieved individuals would be over-represented in the secure or preoccupiedattachment style and under-represented in the fearful and dismissive styles. Althoughprevious studies have found a link between achieved identity and secure attachment, theyounger age of this sample makes it more likely that the stage of intimacy has not beenmastered and some relationship anxiety may still exist. Moratorium tends to be a time ofconfusion and anxiety, so we predicted that these individuals would be over-representedin the preoccupied or fearful attachment style and under-represented in the secure anddismissive styles. Based on previous literature, we expected that many of the identitydiffused individuals would be over-represented in the fearful attachment style, however,based on our theorizing above, we predicted that the diffused would also be over-represented in the dismissive style, and under-represented in the secure and preoccupiedstyles. Due to their high level of identity commitment without having engaged in much deepand meaningful exploration, we predicted that individuals in the foreclosed identity statuswere more likely to be low in both relationship avoidance and relationship anxiety—inother words, over-represented in the secure attachment style and under-represented in the otherthree styles.Erikson (1963, 1968) states that resolving the psychosocial crisis of identity is usually the major

developmental task of adolescence and that resolving the psychosocial crisis of intimacy is usuallythe major developmental task of young adulthood. He further states that difficulty in resolvingidentity issues is likely to interfere with successful resolution of intimacy issues. We thereforepredicted that the relation between identity status and romantic attachment style would bestronger among a college sample than among a high school sample, because the older samplewould be more likely to be grappling with intimacy issues. Although we made no specifichypotheses in regard to gender and ethnicity, these demographic variables were also examined forexploratory purposes.

ARTICLE IN PRESS

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748 741

Method

Participants

Middle adolescent participants in this study were high school students recruited from twopublic high schools in southeast Florida (n ¼ 189; mean age 15.7 years) and late adolescentparticipants were college undergraduate students recruited from psychology classes at FloridaInternational University (n ¼ 324; mean age 21.7 years). The total sample was comprised of 365women and 147 men (one individual failed to report their gender). In terms of ethnicity, 9% ofparticipants were Euro-American, 63% were Hispanic-American, 18% were African-American,3% were Asian, 5% were of other ethnic backgrounds and 2% chose not to report their ethnicbackground.

Measures

Attachment styles were assessed via the Experiences in Close Relationships (ECR; Brennanet al., 1998). The ECR is a 36-item self-report measure of attachment that uses a 7-point Likertscale ranging from disagree strongly to agree strongly. The items were derived from a factoranalysis of several existing self-report measures of adult attachment. The ECR has two 18question subscales labelled ‘‘Model of Self’’ and ‘‘Model of Others’’ also called ‘‘RelationshipAnxiety’’ and ‘‘Relationship Avoidance’’, respectively. For the Model of Self higher scoresindicate more anxiety about rejection by others and feelings of personal unworthiness regardinginterpersonal relationships. For the Model of Others higher scores indicate more interpersonaldistrust and avoidance of closeness with others. The Relationship Anxiety subscale contains itemssuch as the following: ‘‘I worry about being abandoned’’. The Relationship Avoidance subscalecontains such items as ‘‘I try to avoid getting too close to my partner’’. Individuals are assigned toone of the attachment categories based on their scores from the two subscales using classificationcoefficients derived from Fischer’s linear discriminant functions in a normative sample ofN ¼ 1082. Complete scoring and classification procedures can be found in Brennan et al. (1998)or on the web at (http://psychology.ucdavis.edu/Shaver/brennan.html#scoring). Put simply,individuals with high scores on both the anxiety and avoidance subscales are classified as fearful,individuals with low scores on the anxiety subscale and high scores on the avoidance subscaleare classified as dismissive, individuals with high scores on the anxiety subscale and low scores onthe avoidance subscale are classified as preoccupied, and individuals with low scores on bothsubscales are classified as secure.The ECR has good reliability and validity estimates. For example, internal consistency and

test–retest reliability for its two subscales have been reported at .94 and .90 for avoidance and .91and .91 for anxiety, respectively (Brennan et al., 1998; Fraley, Waller, & Brennan, 2000). Inaddition, Brennan et al. assigned individuals to the attachment style categories and found thatthey predicted test variables better than previous measures of attachment. The typology equateswell to Ainsworth’s theory for infants/young children (Ainsworth, Blehar, Waters, & Wall, 1978)and has excellent correspondence to Bartholomew’s adult classification scheme (see Brennanet al.). The ECR was used to group participants into the four adult attachment styles and the twosubscales were also used as continuous variables for Pearson correlation analyses.

ARTICLE IN PRESS

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748742

The Ego Identity Process Questionnaire (EIPQ; Balistreri, Busch-Rossnagel, & Geisinger, 1995)was used to identify participants’ identity status. The EIPQ has two subscales, identityexploration and identity commitment. Cronbach’s alpha for the exploration subscale has beenreported to be .86 with test–retest reliability of .76. Cronbach’s alpha for the commitment subscalehas been reported to be .80 with test–retest reliability of .90 (Balistreri et al., 1995). Balistreri et al.(1995) used median splits on the two subscales to assign participants into one of four identitystatuses as defined by Marcia (1966). For this study, new median splits were calculated based onthe current sample, and the same median scores were used with both the high school and collegesamples. Participants with low scores on exploration and commitment are classified as diffused,low in exploration but high in commitment is classified as foreclosed, high in exploration but lowin commitment classified as moratorium, and high in both exploration and commitment isclassified as achieved.

Procedure

College participants were recruited from upper and lower division psychology classes at a largeurban state university in South East Florida. Individuals who agreed to take part in the study forextra credit in the class were given a packet that included informed consent, a demographicsquestionnaire, and the measures. Participants were told the nature of the study was to surveybeliefs and feelings associated with interpersonal relationships. Participants were giveninstructions on how to complete the measures. High school participants were recruited fromtwo Miami-Dade county schools (also South East Florida). High school participants wereprovided with an informed consent sheet approximately 1 week before the assessment day.Participants were again told the nature of the study was to survey students’ beliefs and feelingsabout interpersonal relationships. Students with consent forms signed by a parent or legalguardian and returned by the assessment date participated in the study for extra class free time.Students completed the measures in a group classroom setting and were assisted as necessary bythe authors or trained research assistants.

Results

Preliminary analyses

Identity status distribution by age group and gender can be seen in Table 1. A w2 analysisrevealed some differences in status distribution between high school and university students(w2ð3Þ ¼ 15:99, p ¼ :001). High school students had higher rates of diffusion, while universitystudents had higher rates of moratorium and achievement. Rates of foreclosure were almostequal.There were also some differences in distribution by gender (w2ð3Þ ¼ 10:6, p ¼ :014), with males

having higher rates of foreclosure and diffusion, while females had higher rates of moratoriumand achievement. Ethnic distribution revealed no statistically significant differences in identitystatus distribution.

ARTICLE IN PRESS

Table 2

Percentages for attachment styles by age group and gender

Secure (%) Preoccupied (%) Dismissive (%) Fearful (%)

College 33.3 31.8 9.6 25.3

High school 17.9 14.5 30.7 36.9

Males 27.1 22.9 13.9 36.1

Females 28.2 26.8 18.4 26.5

Total sample 27.8 25.6 17.1 29.4

Table 1

Percentages for identity statuses by age group and gender

Diffused (%) Foreclosed (%) Moratorium (%) Achieved (%)

College 13.1 28.7 34.3 24.0

High school 26.5 28.6 28.1 16.8

Males 25.3 30.1 29.5 15.1

Females 15.0 27.9 33.1 24.0

Total sample 18.0 28.7 32.0 21.3

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748 743

Attachment style distribution by age group and gender can be seen in Table 2. A w2 analysisrevealed some differences in attachment style distribution between high school and universitystudents (w2ð3Þ ¼ 58:7, po:001). High school students had higher rates of fearful and dismissivestyles, while university students had higher rates of secure and preoccupied styles.There were no statistically significant differences in attachment style distribution by gender, but

there was a difference by ethnicity (w2ð12Þ ¼ 33:4, po:001). Most noticeably, Whites had a muchhigher rate of secure styles (52.1%) versus 24.4% for Blacks, 27.7% for Hispanics, 15.4% forAsians, and 7.7% for other.

How are identity and relationship variables related?

Analyses were done to examine if identity status was related to relationship anxiety andrelationship avoidance, the two underlying dimensions of attachment style. Two 2� 2� 4 (agegroup� sex� identity status) ANOVAs were conducted; one with relationship avoidance as thedependant measure, and the other with relationship anxiety as the dependant measure. TheANOVA for relationship avoidance revealed significant main effects for age group (college vs.high school; Fð1; 480Þ ¼ 14:77, po:001) and identity status (Fð3; 480Þ ¼ 2:67, p ¼ :047), but notfor sex. There was, however, a significant interaction between sex and age group (F ð1; 480Þ ¼ 7:92,p ¼ :015). As can be seen in Fig. 1, high school females had more relationship avoidance than highschool males, and both males and females in high school had a lower mean score than males andfemales in college, but in the college sample, the females had a lower mean avoidance score thanthe males. Scheffe post hoc analyses on identity status revealed that the foreclosed group had asignificantly lower mean score on relationship avoidance than the diffused group (p ¼ :046). Themeans are presented in Table 3.

ARTICLE IN PRESS

3.6o

3.4

3.2

o = Females

x

x = Males

3.0x

2.8 o

Age Group

High School College

2.6

Fig. 1. Mean avoidance scores by age group and sex.

Table 3

Means and standard deviations for relationship avoidance as a function of identity status, gender, and age group

Diffused Foreclosed Moratorium Achieved Total

College Male 3.11(1.21) 2.77(1.17) 3.08(.93) 2.81(1.02) 2.95(1.07)

Female 3.16(1.10) 2.57(1.18) 2.88(.97) 2.67(1.02) 2.77(1.07)

Total 3.14(1.13) 2.62(1.18 2.93(.96) 2.70(1.02) 2.81(1.07)

High school Male 3.16(.94) 2.95(1.01) 3.48(.96) 2.84(.83) 3.14(.96)

Female 3.65(1.04) 3.51(1.11) 3.63(1.32) 3.44(1.34) 3.57(1.20)

Total 3.44(1.02) 3.30(1.10) 3.59(1.22) 3.30(1.25) 3.42(1.14)

Total Male 3.14(1.05) 2.84(1.10) 3.22(.95) 2.82(.94) 3.03(1.02)

Female 3.40(1.09) 2.87(1.24) 3.11(1.14) 2.87(1.16) 3.03(1.17)

Total 3.29(1.08) 2.86(1.19) 3.14(1.09) 2.86(1.11) 3.03(1.13)

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748744

The ANOVA for relationship anxiety yielded a significant main effect only for identity status(F ð3; 480Þ ¼ 6:09, po:001) and no significant interactions. Scheffe post hoc analyses revealed thatthe foreclosed group scored significantly lower in anxiety than the achieved (p ¼ :005) andmoratorium (p ¼ :002) groups (see Table 4).

Identity status and attachment style

Identity status by attachment style distributions were also examined using multinomial logisticregression. Identity status, sex, and age group was used as independent variables with attachmentstyle as the dependant variable, yielding a statistically significant model (w2ð15Þ ¼ 82:89, po:001).Age group (w2ð3Þ ¼ 49:97, po:001) and identity status (w2ð9Þ ¼ 22:52, p ¼ :007) were significantfactors, but sex was not significant. Results by age group and sex revealed a statistically significantrelationship between identity status and attachment style for both males (w2ð9Þ ¼ 20:78, p ¼ :014)and females (w2ð9Þ ¼ 18:51, p ¼ :030), as well as for college students (w2ð9Þ ¼ 20:04, p ¼ :018), butthey were not significantly related for high school students.Some interesting observations also emerged from these analyses. The identity status by

attachment style distribution can be seen in Table 5. The distribution is presented separately for

ARTICLE IN PRESS

Table 4

Means and standard deviations for relationship anxiety as a function of identity status, gender, and age group

Diffused Foreclosed Moratorium Achieved Total

College Male 3.39(1.22) 3.59(1.17) 3.72(1.08) 4.33(1.00) 3.72(1.14)

Female 3.54(1.07) 3.42(1.13) 4.09(1.24) 3.92(1.32) 3.80(1.24)

Total 3.48(1.12) 3.47(1.14) 3.99(1.21) 4.00(1.27) 3.78(1.22)

High school Male 3.60(.95) 3.08(.83) 3.94(.86) 4.43(1.08) 3.63(.99)

Female 3.50(1.16) 3.28(.87) 3.51(1.10) 3.38(.96) 3.42(1.02)

Total 3.55(1.06) 3.20(.85) 3.64(1.04) 3.63(1.07) 3.49(1.01)

Total Male 3.51(1.07) 3.37(1.06) 3.80(1.00) 4.37(1.00) 3.68(1.08)

Female 3.52(1.11) 3.38(1.05) 3.91(1.22) 3.78(1.25) 3.68(1.19)

Total 3.52(1.08) 3.37(1.05) 3.88(1.17) 3.90(1.23) 3.68(1.15)

Table 5

Frequencies of attachment style within identity statuses for college and high school samples

Secure Preoccupied Dismissive Fearful

Obs Exp Obs Exp Obs Exp Obs Exp

College Diffused 14 13.9 8 13.3 8 4.1 12 10.7

Foreclosed 41 30.4 21 29.2 9 8.9 21 23.5

Moratorium 31 36.3 39 35.0 9 10.6 31 28.1

Achieved 20 25.4 34 24.5 5 7.4 18 19.7

High school Diffused 9 8.4 8 6.8 15 13.9 14 17.0

Foreclosed 12 9.1 3 7.4 15 15.1 20 18.5

Moratorium 5 9.3 9 7.5 16 15.4 21 18.8

Achieved 6 5.3 6 4.3 7 8.7 10 10.7

Obs, observed; Exp, expected.

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748 745

the college and high school sample because age grouping was a significant factor in the regressionanalysis. Many of our predictions in regard to the distribution between identity and attachmentwere confirmed within the college sample where the w2 analysis was significant, but not within thehigh school sample where the w2 analysis revealed no significant difference, and thus, for thefollowing section, we will consider only the college sample. Specifically, we predicted that identityachieved individuals would be over-represented in the secure and preoccupied attachment stylesand under-represented in the other two styles. As compared to expectations calculated by w2

analysis, our predictions were true for all but the secure group, who were below expectation(20 vs. 25.4). For those in moratorium, we predicted that they would be over-represented in thepreoccupied and fearful styles, and under-represented in the secure and dismissive styles. Resultswere consistent with these predictions. We predicted that the diffused would be over-representedin the fearful and dismissive styles and under-represented in the secure and preoccupied styles.This was true for all but the secure group who were at expectation (14 vs. 13.9). Our prediction in

ARTICLE IN PRESS

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748746

regard to foreclosure, that it would be over-represented in the secure style and under-representedin the other three styles, held for all but the dismissive group, who were at expectation (9 vs. 8.9).

Discussion

This study adds to the small handful of studies that have attempted to investigate therelationship between identity status and romantic attachment. It is the first study to lookspecifically at relationship avoidance and relationship anxiety, and also the first to include a highschool sample. By comparing the high school and college sample, we found some expecteddevelopmental differences, consistent with previous literature, such as a greater percentage ofmoratorium and achievement in the older group, as well as a greater percentage of secureattachment style among the college students. In terms of attachment, we observed an expecteddecrease in relationship avoidance from high school to college, but relationship anxiety was notsignificantly different between the two groups. An interaction was also found between age and sexin that high school females were highest in avoidance while college females were lowest. Althoughavoidance was lower in college for both males and females, there was a much larger drop inrelationship avoidance for females than there was for males between high school and collegegroups. However, it is important to remember that the age groups were drawn from differentpopulations and therefore, these samples may differ on other variables besides age, so anyinterpretation about age differences should be tentative.Our ANOVAs partially confirmed our hypotheses. We predicted that the achieved and

foreclosed would score significantly lower in avoidance than the diffused and moratorium, but theonly difference that was significant was between the foreclosed and diffused respondents. In termsof relationship anxiety, we predicted that the foreclosed and diffused would score significantlylower than those in moratorium and achieved. Again we were half right. The foreclosed scoredsignificantly lower than both the achieved and those in moratorium, but the diffused did not scoresignificantly different from the other identity statuses.When examining the relation between identity status and attachment style, we predicted that

they would be more closely related in our college sample where developmentally, we theorizedthat these individuals would be more likely to be grappling with intimacy and identity issues, asoppose to our high school sample, where we theorized that they would be more concerned withidentity issues and much less so with intimacy issues. This did seem to be the case in that theattachment by identity distribution was significant for the college sample, but not for the highschool sample. Not only were attachment style and identity status significantly related within thecollege sample, but much (although not all) of the specific relationships were similar to what wehad predicted. The strongest findings in the college sample that were consistent with ourpredictions was that those in the identity achieved group would be over-represented in thepreoccupied attachment style (34 observed vs. 24.5 expected) and those in the foreclosed identitystatus group would be over-represented in the secure style (41 vs. 30.4). Although scant, previousliterature has suggested a link between achieved identity and secure attachment, and betweendiffused identity and fearful attachment. Contrary to those findings, in our college sample,the achieved group were under-represented in the secure style (20 vs. 25.4), and although thediffused group were over-represented in the fearful style, it was not much different than

ARTICLE IN PRESS

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748 747

expectation (12 vs. 10.7). Although we predicted that those in the achieved status were more likelyto be secure or preoccupied, it did come as somewhat of a surprise that they were considerablymore likely to be preoccupied than secure. One possible reason for the discrepancy may have to dowith ethnicity. Our sample was much more diverse than previous studies. We also found asignificantly lower rate of secure attachment style among our ethnic minority members thanamong our white participants. Future studies might want to take a closer look at ethnicdifferences and take them into consideration in their theorizing on the relationship betweenidentity status and attachment style.In summary, it would appear that the relationship between identity status and romantic

attachment style is a complex one. From Table 3 it became clear that status does not dictatestyle—a person in any identity status can have any of the attachment styles. Yet the distributionwas non-random and some trends did emerge. The relationship between identity and attachmentdoes seem to warrant further study. Looking for mediators and moderators of the relationshipmight further help to clarify the trends and lead to deeper understanding of their significance.Some limitations of the current study should also be noted. The sample is cross-sectional so causalinferences as to whether identity influences attachment or vice versa cannot be made. Althoughthe ethnically diverse sample has brought to light some interesting findings, these need to bereplicated and studied further in a more balanced sample. Finally one must keep in mind thatthe measures used in this study are self-report and the accuracy of individual reporters cannot beassured. However, this study does contribute to what will hopefully be a growing literature onidentity and attachment, and future studies could certainly address these limitations.

References

Ainsworth, M. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the

strange situation. Oxford, UK: Lawrence Erlbaum.

Balistreri, E., Busch-Rossnagel, N. A., & Geisinger, K. F. (1995). Development and validation of the Ego Identity

Process Questionnaire. Journal of Adolescence, 18, 79–192.

Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspective. 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.

Benson, M. J., Harris, P. B., & Rogers, C. S. (1992). Identity consequences of attachment to mothers and fathers among

late adolescents. Journal of Research on Adolescents, 2, 187–204.

Bowlby, J. (1982). Attachment and loss: Retrospect and prospect. American Journal of Orthopsychiatry, 52, 664–678.

Bowlby, J. (1988). A secure base: Parent–child attachment and healthy human development. 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 close relationships (pp. 46–76). New York,

NY: Guilford Press.

Erikson, E. H. (1963). Childhood and society (2nd ed.). New York: Norton.

Erikson, E. H. (1968). Identity: Youth and crisis. New York: Norton & Co.

Fraley, R. C., Waller, N. J., & Brennan, K. A. (2000). An item response theory analysis of self-report measures of adult

attachment. Journal of Personality and Social Psychology, 78, 350–365.

Hoegh, D. G., & Bourgeois, M. J. (2002). Prelude and postlude to the self: Correlations of achieved identity. Youth &

Society, 33, 573–594.

ARTICLE IN PRESS

S.L. Berman et al. / Journal of Adolescence 29 (2006) 737–748748

Kennedy, J. H. (1999). Romantic attachment and ego identity, attributional style, and family of origin in first-year

college students. College Student Journal, 33, 171–180.

Kroger, J. (1985). Separation-individuation and ego identity status in New Zealand university students. Journal of

Youth and Adolescence, 14, 133–147.

Kroger, J., & Haslett, S. J. (1988). Separation-individuation and ego identity status in late adolescence: A two-year

longitudinal study. Journal of Youth and Adolescence, 17, 59–79.

Lapsley, D. K., Rice, K. G., & Fitzgerald, D. P. (1990). Adolescent attachment, identity, and adjustment to college:

Implications for the continuity of adaptation hypothesis. Journal of Counseling and Development, 68, 561–565.

Marcia, J. E. (1966). Development and validation of ego-identity status. Journal of Personality and Social Psychology, 3,

551–558.

Reich, W. A., & Siegel, H. I. (2002). Attachment, ego-identity development and exploratory interest in university

students. Asian Journal of Social Psychology, 5, 125–134.

Vogensen, A. C. (2003). Finding themselves: Identity formation and romantic attachment styles in late adolescence.

Dissertation Abstracts International, Section B: The Sciences & Engineering, 64(3-B), 1529.

Zimmermann, P., & Becker-Stoll, F. (2002). Stability of attachment representations during adolescence: The influence

of ego-identity status. Journal of Adolescence, 25, 107–124.