Age and Sex Differences in Perceptions ofNetworks of Personal Relationships

Wyndol FurmanUniversity of Denver

Dviane BuhrmesterUniversity of Texas at Dallas

FuBMAN, WYNDOL, and BUHRMESTER, DUANE. Age and Sex Differences in Perceptions of Networksof Personal Relationships. CHILD DEVELOPMENT, 1992, 63, 103-115. In this study, 549 youths inthe fourth grade, seventh grade, tenth grade, and college completed Network of RelationshipInventories assessing their perceptions oftheir relationships with significant others. The findingswere largely consistent with 7 propositions derived from major theories of the developmentalcourses of personal relationships. In particular, mothers and fathers were seen as the most fre-quent providers of support in the fourth grade. Same-sex friends were perceived to be as support-ive as parents in the seventh grade, and were the most frequent providers of support in the tenthgrade. Romantic partners moved up in rank with age nntil college, where they, along with friendsand mothers, received the highest ratings for support. Age differences were also observed inperceptions of relationships with grandparents, teachers, and siblings. Finally, age differencesin perceived conflict, punishment, and relative power suggested that there was a peak in tensionin parent-child relationships in early and middle adolescence. Discnssion centers around therole various relationships are perceived as playing at different points in development.

Throughout the course of development. Heeding this call, a number of investi-individuals have personal relationships with gators have begun to examine children's anda number of people in their networks, in- adolescents' social networks (e.g., Blyth,eluding parents, siblings, relatives, teachers, 1982; Bryant, 1985; Cauce, Felner, & Prima-friends, and romantic partners. Although vera, 1982; Coates, 1987; Reid, Landesman,research has examined the developmental Treder, & Jaccard, 1989). Recently, Furmancourse of many of these relationships, most and Buhrmester (1985a) adopted such a net-investigations have considered each type of work perspective and assessed elementaryrelationship in isolation. The few studies ex- school children's perceptions of conflict, rel-amining more than one relationship have ative power, and six types of support in dy-usually been limited to comparisons of adic relationships with numerous types ofparent-child relationships and friendships network members. These data allowed them(e.g.. Hunter, 1985; Hnnter & Youniss, 1982) to compare systematically the similaritiesor to comparisons of relationships in yoiing and differences in relationships with eachchildren's networks (Lewis, Feiring, & Kot- network member, thereby yielding informa-sonis, 1984). Systems theorists, however, tion about the perceived contributions thathave called for studies of networks of rela- each relationship makes to children's socialtionships and their interrelations so that an lives. The present cross-sectional study ex-integrated account of children and adoles- tended this work by examining differencescents' social lives can be obtained (Bronfen- in perceptions of relationships among fourbrenner, 1979; Hartnp, 1979; Lewis & Feir- developmental periods—preadolescenceing, 1979). (grade 4), early adolescence (grade 7), mid-

This research was supported by grant no. 1R01HD16142 from the National Institute of ChildHealth and Human Development (Wyndol Eurman, Principal Investigator), and the preparationof the manuscript was supported hy a William T. Grant Faculty Scholar award to Wyndol Furman.,Appreciation is expressed to Terry Adler, Richard Lanthier, and Elizabeth Wehner for theirassistance in the data collection and analyses. We are also indehted to the students of GherryGreek Schools, Jefferson County Schools, Denver Christian Schools, and the University of Den-ver. Reprints of this article can he obtained from Wyndol Furman, Department of Psychology,University of Denver, Denver, CO 80208.

[Child Development, 1992,63,103-115. © 1992 by the Society for Research in Child Development, Inc.All rights reserved. 0009-3920/92/6301-0012$01.00]

104 Child Development

die adolescence (grade 10), and late adoles-cence (college).

Onr goal was to create a broad descrip-tive account of age differences in percep-tions of the central relationships in youths'networks. The value of such a broad pictureis that it can capture general patterns thatare missed in more fine-grained studies ofspecific relationships. That is, a network per-spective allows one to evaluate and interpretage differences in perceptions of one rela-tionship in comparison to those in other re-lationships. Thus, it allows one to evaluatedevelopmental theories and facilitate the in-terpretation and integration of findings frommore focused studies.

We thought that the youths' perceptionswould provide unique and valuable informa-tion about their impressions and evaluationsoftheir relationship experiences. The youthis in a unique position of being the only "in-sider" privy to the history of interactions thatoccur across a wide variety of public and pri-vate contexts (Furman, 1984; Olson, 1977).His or her perceptions are certainly notidentical with an outside observer's percep-tions of the interactions in a relationship.They are likely to reflect the relationship'shistory and the youth's subjective interpre-tation of interactions as well as the actualinteractions themselves. As such, they arelikely to affect their interactions with peoplein their networks. Moreover, they are inter-esting phenomena in their own right, assome research suggests that such subjectiveevaluations of relationships may have astronger impact on adjustment than objec-tive indices (Barrera, 1981). These ratingsalso provide one of the few common metricsavailable for comparing the diverse set of re-lationships and age groups examined here.

As a means of organizing our investiga-tion, we reviewed major theories of socialdevelopment and compiled a list of propo-sitions about age changes in relationships.Sullivan's (1953) theory of socioemotionaldevelopment, Blos's (1967) conceptualiza-tion of the development of autonomy, andYouniss and Smollar's (1985) integration ofPiaget and Sullivan's insights proved partic-ularly valuable. We examined past researcbon age differences in observed interactions,as well as perceptions of relationships, as webelieved changes in interactions may lead toand reflect changes in perceptions. By inte-grating the insights of these theorists andprevious research, we derived seven propo-sitions that could be evaluated empirically.

1. Marked age differences occur in thenetwork members that children and adoles-cents report relying upon the most assources of support. Specifically, preadoles-cent children report that they depend moston parents, early and middle adolescents re-port that they turn most often to friends,and late adolescents report that they de-pend most on romantic partners. Severallines of reasoning converge to create thisbroad picture. Bios (1967) contended thatadolescents' struggle to establish autonomyleads to decreased emotional dependencyon parents, but they simultaneously come torely more on peers for support and guidancebecause they cannot function completely in-dei)endently. Similarly, Sullivan (1953) pro-posed a series of significant developmentalshifts in whom children relied upon mostoften to fulfill particular social needs (Buhr-mester & Furman, 1986). Specifically, he ar-gued that young adolescents establish inti-mate chumships with peers in an effort tosatisfy a need for an egalitarian "collabora-tive" relationship. In so doing, adolescentscome to rely less and less on parents to meetsocial needs. Furthermore, he suggestedthat as sexuality needs intensify after pu-berty, their fulfillment becomes integratedwith the fulfillment of intimacy needsthrough romantic relationships. These trans-formations in the network entail changes ina number of specific relationships.

2. Parent-child relationships during ad-olescence are seen as less supportive, moreconflicted, and more egalitarian in naturethan during preadolescence. Bios (1967)conceptualized this transformation in termsof separation and individuation. Duringearly adolescence, parent-child conflict es-calates as youths struggle to renegotiate de-pendency bonds and authority relationships.Eventually, during late adolescence, conflictsubsides as a rapprochement is eventuallyreached in a new form of parent-offspringrelationship. Youniss and Smollar (1985)also believed that parent-child relationshipsbecome less asymmetrical in adolescence, inpart because of the experience of havingegalitarian experiences with peers.

3. Sibling relationships are seen asgradually becoming less intense and moreegalitarian with age. Although there hasbeen relatively little theorizing about thedevelopmental course of sibling relation-ships, Bossard (1948) suggested that adoles-cents may move away from the family as awhole as they individuate from their par-ents. It is not clear, however, whether the

separation from siblings is accompanied byelevated or decreased conflict. If the strivingfor autonomy extends to relationships withsiblings, then a rise in sibling confiict mayoccur during early adolescence. However, ifautonomy issues are not played out in sib-ling relationships, a general distancing ofsiblings may occur, leading to fewer occa-sions for confiict.

4. Other significant adults, includinggrandparents and teachers, are seen as lessfrequent sources of support as childrengrow older. During early childhood, grand-parents and teachers often serve as second-ary attachment figures, helping to fulfill chil-dren's needs for nurturance and assistance(Tinsley & Parke, 1984). The significanceof personal relationships with these adultstypically diminishes with age, because theneeds older children experience are bettermet by others.

5. Same-sex friendships are perceivedas increasingly supportive during early ado-lescence. Sullivan (1953) argued that duringpreadolescence and early adolescence theneed for intimacy and desire for "consensualvalidation" increase, which are both bestsatisfied in same-sex friendships. Bios (1967)also described such changes, but in his viewthe changes result from unmet dependencyneeds emanating from the struggle to indi-viduate from parents.

6. Romantic relationships are thoughtto become increasingly supportive duringlate adolescence. As noted previously, Sulli-van (1953) contended that the integration ofintimacy and sexual needs occurs in lateadolescents' and young adults' romantic re-lationships. From a different perspective,Erikson (1950) argued that intimacy in ro-mantic relationships occurs after the resolu-tion of the identity crisis.

7. Girls' same-sex friendships are per-ceived as more supportive than boys' friend-ships. Previous research has consistentlyfound that girls' interactions with friends arelikely to be perceived as more intimate thanboys' (Buhrmester & Furman, 1987; Furman& Buhrmester, 1985a; Hunter & Youniss,1982; Sharabany, Gershoni, & Hoffman,

Fnrman and Buhrmester 105

1981). It is not clear, however, if these differ-ences apply to other relationships.

Method

ParticipantsThe participants were 107 fourth-grade

(53 males, 54 females; M age = 9 years, 4months old), 119 seventh-grade (67 males,67 females; M age = 12-5), 112 tenth-grade(51 males, 65 females; M age = 15-6), and216 college students (123 males, 98 females;M age = 19-3). Most were Caucasian chil-dren from middle- to upper-middle-classfamilies. Approximately 90% were from in-tact families, whereas the remaining 10%were from stepfamilies or one-parent fami-lies. The college students were attending amedium-size private university in a largecity; almost all lived in dormitories awayfrom their family's home. The younger stu-dents attended suburban public schools inthe same metropolitan area.

Procedure and MeasureRevised versions of Furman and Buhr-

mester's (1985a) Network of RelationshipsInventory (NRl) were administered to col-lege students individually and to classroomsof fourth- seventh-, and tenth-grade stu-dents. The college students received coursecredit for participating.

The present version of the NRI assesses10 relationship qualities, which includeseven provisions of support derived fromWeiss's (1974) tiieory: {a) reliable alli-ance—a lasting dependable bond, (fo) en-hancement of worth, (c) affection, {d) com-panionship, (e) instmmental help, (/)intimacy, and (g) nurturance of the other.Additionally, three other characteristicswere rated: (a) confiict, (fo) punishment, and(c) relative power of the child and otherperson.^

Subjects rated each of these qualities intheir relationships with each of the follow-ing people: (c) mother or stepmother, (fo) fa-ther or stepfather, (c) older brother, {d)younger brother, (e) older sister, (/) youngersister, (g) grandparent, {h) closest same-sexfriend,^ (i) romantic partner. These peoplewere selected on the basis of Furman and

1 Subjects also rated relationsbips on four other dimensions—satisfaction importancelarity, and inclusion. These results are not reported because they fell outside of̂ Oie ;

p b H n ? .2 The college students were asked to describe their closest local same-sex friend and their

closest same-sex firiend living elsewhere. In this paper, results for tbe one designated as mostimportant are presented. Supplementary analyses of scores for each of the two friendship^revealed, however, the same pattem of results. menasmps re

106 Child Development

Buhrmester's (1985a) finding that most chil-dren had spontaneously named them as im-portant people in their networks; althoughother people may be important for somechildren, the majority had not named anyoneelse. If subjects knew more than one personin a category, they were instructed to ratethe relationship with the one which wasmost important to them. If they rated morethan one sibling, scores for the one rated asmost important were included in the pri-mary analyses. We chose to do this ratherthan select one randomly to insure compara-bility with ratings of best friends and roman-tic partners, who were implicitly the sub-jects' most important peer relationships. Ifthere was no one who fit in a particular cate-gory, the subject was omitted from analysesof that particular relationship.

The subjects were asked to rate howmuch a relationship quality occurred in eachrelationship (e.g., "How much free time doyou spend with each of these persons?")Questions were phrased in concrete termsto insure that children of all grades wouldunderstand them and interpret them simi-larly (e.g., for relative power, "Who tells theother person what to do more often, you ortiiis person?"). Ratings were done on stan-dard five-point Likert scales. In the versionadministered to the fourth-, seventh-, andtenth-grade students, each relationship qual-ity was assessed by three items. In the ver-sion administered to college students, eachquality was assessed by six items.'' Only thetiiree items on each scale that were identicalin the two versions were used to derive the10 scale scores for each of the relationships.The college version also did not includequestions about punishment or teachers.

The seven different provisions werecombined into one dimension because theywere highly intercorrelated and becauseWeiss (1974) had thought of them all as so-cial provisions. The internal consistency co-efficients for the support composite and theconfiict, punishment, and relative powerscales for the different relationships were

calculated for each grade level and found tobe satisfactory (M alpha = .81). Alphas didnot vary as a function of grade. The fourscales were relatively independent of oneanother, - .30 < r's < .30, except for the con-flict and punishment scales, which weremoderately correlated, M r = .43.

Results and Discussion

As an initial screening test, a multivari-ate repeated-measures analysis of variancewas conducted in which grade and sex wereindependent variables, relationship typewas a repeated variable, and relationshipqualities (support, conflict, and power) weredependent variables.* Significant effectswere found for grade, sex, relationship,grade x relationship, and sex x relation-ship, all Wilks's lambda's < .01. To examinethe specific propositions previously deline-ated, ANOVAs were conducted comparingrankings of relationships at each grade level(Proposition 1) and age and sex differencesin each relationship (Propositions 2-7).

Comparisons among RelationshipsProposition 1.—Marked age differences

were expected to occur in the network mem-bers that children and adolescents report re-lying upon the most as sources of support.To compare the relative ranking of relation-ships at each grade level, separate ANOVAswere conducted on the support, confiict,punishment, and relative power scores forthe set of relationships at each grade level.Relationship type was a within-subject vari-able, whereas sex was a between-subjectsvariable. Follow-up analyses of significanteffects were conducted using Newman-Keuls tests. Mean scores at each grade levelare presented in Table 1.

For support scores, significant main ef-fects of relationship type were found at eachof the four grade levels, all F's > 6.50, p's< .05. Additionally, analyses of the collegestudents' scores revealed a significant inter-action between relationship type and sex,F(5,500) = 2.84, p < .05. Consistent with

^ Copies of the two versions of the NRI are available from tbe first autbor.'' Ratings of relationsbips witb romantic partners were not included in tbese analyses be-

cause a significant proportion of the cbildren (48%) did not have tbese relationsbips; ratings forteachers were also excluded because college students bad not been asked to describe theserelationships. Supplementary repeated-measures analyses of variance were conducted to com-pare the various ratings of those youths who did have one of these relationships. References intbe text to differences between romantic partners or teachers and tbe other relationships arebased on these analyses. Similarly, ratings of punishment were not included because collegestudents had not completed those ratings, but supplementary analyses of the punishment, sup-port, conflict, and relative power measures for the fourth-, seventh-, and tenth-grade samplesrevealed the same signiflcant multivariate effects.

Furman and Buhrmester 107

TABLE 1

MEANS AND STANDARD DEVIATIONS OF RATINGS BY GRADE AND RELATIONSHIP TYPE

GRADE

10 13 +

Support:Mother 3.90(.61)'''' 3.51(.76)''''Fatber 3.89(.74)'''' 3.39(.9O)'''Sibling 3.43(.85) '̂̂ ''' 2.99(1.00)̂ -'̂ () ()Grandparent 3.64(.70)2'» 3.01(.77)^'' 2.75(.76)^'^'' 2.68(.69) '̂>Teacher 2.61(.85)5''' 1.86(.62)3''' 1.93(.6O) '̂'Same-sex friend 3.41(.85)3*'' 3.61(.87) '̂'' 3.57(.81)'''' 3.37(.62)**Romantic friend 2.77(1.22)̂ '"= 3.08(1.07)2''''= 3.19(.93)2'''''' 3.50(.90)''^

Conflict:Motber 1.98(.75) '̂'' 2.37(.99) '̂'' 2.30(.98)^''' 2.07(.99)''''Fatber 1.94(.74) '̂'' 2.26(.99) '̂'' 2.30(1.00)2''' 1.98(.81)'*Sibling 3.01(1.13)'''' 3.36(1.18)'''' 2.69(1.08)'° 1.94(.81)WGrandparent 1.59(.69) '̂' 1.38(.6O)̂ ''' 1.20(.47)''-'' 1.30(.64)̂ '>Teacher 1.64(.73)̂ ''» 1.81(.91)3''' 1.36(.56)*'''Same-sex friend 2.01(.97)2''' 1.94(.83)̂ "'' 1.66(.7O)̂ ''' 1.54(.63)2'''Romantic friend 1.68(.78)'*'' 1.47(.64)5A'' 1.74(.77)'A'' 1.88(.88)i'''

Punishment:Motber 2.28(.99)'''' 2.83(1.08)''' 2.60(1.01)'-''Fatber 2.25(1.01)'''> 2.84(1.17)''' 2.66(1.12)''>'Sibling 1.97(1.10)2''' 2.15(1.24)2'» 1.55(.82)2'''Grandparent 1.71(.75)2'''-'> 1.82(.91)3''' 1.51(.61)2''Teacher 1.85(.84)^'' 2.09(.92)2''' 1.64(.65)̂ '<̂Same-sex friend 1.39(.64) '̂' 1.51(.68)3-'' 1.32(.58)2-''Romantic friend 1.35(.59) '̂'' 1.30(.57)'''' 1.5l(.7O) '̂''

Relative power:Motber 2.19(.98)2''' 1.91(.81)2-'' 1.84(.66) '̂'' 2.30(.75)*>''Father 2.18(.98)^-'' 1.97(.83)2'''''' 1.73(.65)*''' 2.04(.78)5'''Sibling 2.93(1.14)'''' 2.98(1.19)'''' 3.34(.93)'''' 3.24(.76)''''Grandparent 2.33(.93)2''' 2.28(.75)2» 2.26(.71)3» 2.50(.70)3'''Teacher 2.07(.88)'''' 1.72(.73) '̂'' 1.89(.6O)'''''Same-sex friend 2.84(.82)'''' 2.95(.49)''''''' 3.04(.30) '̂'' 3.09(.39)2'''Romantic friend 2.97(.95)'''' 2.85(.62)'''' 2.85(.57)2''' 3.05(.55)2'''

NOTE.—Higher scores indicate perceptions of more of a quality. Higher scores on relative power indicate greaterperceptions of power by the child. The numhers in superscripts indicate the rank order of the means across relation-ships within each grade. Means with different number ranks in the same column are significantly different from eachother. The letters in superscripts indicate the rank order of means across grade levels within each type of relationship.Means with different letters in the same row are significantly different from each other. Sex effects and interactionswith sex are described in the text. The numbers in parentheses are standard deviations.

our first proposition, fourth graders ratedtheir mothers and fathers as the most fre-quent providers of support. In the seventhgrade, they shared this position with same-sex friends, and they fell to second andfourth place in the tenth grade. Same-sexfriends were perceived to be relatively lowin the hierarchy in the fourth grade (tied forfourtli with siblings) but were at the top ofthe hierarchy in the seventh and tenthgrades. Romantic friends were in the sixthposition in the fourth grade, tied for third inthe seventh grade, and tied for second in thetenth grade. In college, males rated roman-tic relationships as the most supportive,whereas females saw relationships with

mothers, friends, siblings, and romantic part-ners as the most supportive.

These findings are largely consistentwith those from past studies that examinedperceptions of a limited number of relation-ships or relationship features (Buhrmester &Furman, 1987; Hunter & Youniss, 1982;Sharabany et al., 1981). The strength of thepresent network approach is that it enablesus to map the age differences in perceptionsof a wide spectrum of relationships simulta-neously and formally compare each relation-ship's relative position in the hierarchy. Al-though our findings are consistent with bothSullivan's (1953) and Blos's (1967) theories

108 Child Development

of development, the explanation of theseage-related differences need not be limitedto a single theoretical perspective. Rather, anumber of different factors probably contrib-ute to these differences. For example, thereported greater reliance on friends ratherthan parents for support during early andmiddle adolescence may refiect: (a) ad-vances in cognitive abilities that foster theexamination of one's self, which in turn mo-tivates adolescents to seek same-age con-fidants who share their desires for self-exploration (Gottman & Mettetal, 1986) andconsensual validation (Sullivan, 1953); (fo)pubertal changes, which awaken adolescentconcerns about sexuality and physical ap-pearance, which are most comfortably dis-cussed with peers; (c) the development ofinterests and concerns outside the familycontext; and {d) the emerging autonomyfrom the family, which leads adolescents totum to friends for emotional support in deal-ing with family tensions and to fill the voidleft by their attempts to demonstrate emo-tional independence from parents (Bios,1967). Similarly, the ascendance of romanticpartners as perceived sources of support dur-ing late adolescence is probably not onlylinked to the drive for sexual intimacy (Sulli-van, 1953) but also to culturally derived ex-pectations to begin the search for romanticpartners with whom they may want to havelasting relationships.

For the conflict, punishment, and rela-tive power scores, significant differenceswere found among the relationships at allgrade levels, all F's > 11.00, p's < .01, buttheir ranking was relatively similar acrossthe four grade levels. For example, in grades4, 7, and 10, siblings received the highestratings of conflict, and mothers and fathersreceived the second highest ratings. In col-lege, the ratings for all relationships tendedto be relatively low, resulting in a tie amongthese three and romantic relationships. Forpunishment, mothers and fathers were al-ways ranked highest, whereas same-sexfriends and romantic friends were rankedlowest.

Many theorists have emphasized thecontributions of peer conflict to develop-ment (see Shantz, 1987), but these resultsindicate that the bulk of confiict is thoughtto occur with family members, particularlysiblings. Although conflict with family mem-bers may differ in many respects from con-flict with peers, it should not be overlookedas an important contributor to the reduction

of egocentrism and the growth of interper-sonal understanding.

Finally, relationships with the variousadults (parents, teachers, and grandparents)were ranked as the most asymmetrical,whereas those with other children (siblings,same-sex friends, and romantic friends) wereranked as the most egalitarian. The analysesof the confiict, punishment, and relativepower scores did not reveal any significantsex effects or interactions with sex exceptthat seventh-grade girls reported having lesspower in their relationships than did boys,F(l,37) = 5.48, p < .05.

Proposition 2.—Parent-child relation-ships were expected to be seen as less sup-portive, more confiicted, and more egali-tarian in nature during adolescence thanduring preadolescence. To examine this andsubsequent propositions, separate grade Xsex ANOVAs were conducted on each of thefour scores for each of the seven types ofrelationships to identify the specific effectsof grade, sex, and their interaction. New-man-Keuls follow-up tests were used toidentify the bases of significant effects. Sig-nificant F values are presented in Table 2,and results of the post hoc comparisons arecontained in Table 1. A brief summary ispresented in the following sections.

The findings provide consistent supportfor Proposition 2. Perceptions of support forboth parents were greater in the fourth gradethan in the seventh and tenth grade. Collegewomen's ratings of support from motherswere greater than those in seventh and tenthgrade, whereas college men's scores weresimilar to tenth graders (girls, fourth M =3.90, seventh M = 3.48, tenth M = 3.43,college M = 3.61; boys, fourtii M = 3.91,seventh M = 3.53, tenth M = 3.19, collegeM = 3.26). Men's and women's scores dif-fered significantiy in the tenth grade and col-lege. Both males' and females' ratings ofsupport from fathers tended to be greater incollege than in the tenth grade, p < .10. Boysalso tended to perceive their relationshipswith fathers to be more supportive than didgirls, F(l,519) = 2.98, p < .10. These sexdifferences are consistent with prior worksuggesting that parents are more responsiveto same-sex children (Baumrind, 1971; Mar-golin & Patterson, 1975; Noller, 1980).

Perceptions of conflict with both parentswere greater in the seventh and tenth gradethan in the fourth grade or college. Similarly,ratings of punishment from parents were

Fnrman and Buhrmester 109

TABLE 2

SIGNIFICANT GRADE AND SEX EFFECTS ON RELATIONSHIPQUALITIES

Mother:SuDDortConflictPunishmentRelative power

FatherSupportConfiictPunishmentRelative power

Sibling:

ConflictPunishmentRelative power

Grandparent:SupportConflictPunishmentRelative power

Teacher:SupportGonflictPunishmentRelative power

Same-sex friend:

GonHictPunishmentRelative power

Romantic friend:SuDportGonflictPunishmentRelative power

+ p < .10.*p< .05.** p < .01.

Grade

14.92**4.64**7.64**

10.62**

26.03**5 14**7.56**5.66**

3 78*37 63**6.57**3.51*

39.65**7 52**4.36*2.66*

37.61**10.21**8.79**6.21**

3 27*12.01**2.68+6.21**

6 60**3.08*

Sex

4.30*

9.64**

2.98"^

4.89*

5 76*

8.27**

4.14*9.94**

10.44**

18 21**

6 60*

Grade X Sex

2.70*

3.50*

2.78+

3.60**

greater in the seventh and tenth grade thanin the fourth grade. Perceptions of theyouths' power in their relationships witheach parent decreased between the fourthand tenth grade and the increased again incollege. Boys thought they had more powerin these relationships than girls thought theyhad, but the absolute levels of scores forboth sexes fell below the scale's midpoint,indicating that both parents were thought tohave more power than either their sons ordaughters had.

The multifeature assessment allowed usto create a portrait of perceptions of parent-child relationships that revealed age and sexdifferences on a number of dimensions. In

general, the results are consistent withProposition 2 and the view that during ado-lescence a transformation occurs in relation-ships with parents (Grotevant & Cooper,1986; Youniss & Smollar, 1985). The courseof this transformation is best seen in Fig-ure 1.

Perceived support is lower in early andmiddle adolescence, but this age differencejhould not be overstated. The network ap-proach makes the scope of the age differ-ences particularly clear by permitting com-parisons with other relationships in theadolescents' lives. Even when perceivedsupport from parents is at its lowest in mid-dle adolescence, the ratings of these rela-

110 Child Development

4.0 n

3.5-

^ 3 . 0 .

crz

2 .0 -

_ - • SUPPORT

PUNISHMENT

a RELATIVE POWER

/••^" "« CONFLICT

10 13+GRADE

FIG. 1.—Age differences in characteristics of parent-child relationship

tionships are still exceeded only by those forfriends.

At the same time, perceived conflict andpunishment with parents is greater in earlyand middle adolescence, as others havefound (see Steinberg, 1989). Again, though,the difference is modest, amounting to abouta third of a standard deviation. The findingsconcerning perceptions of relative power,however, were not consistent with the viewthat the power structure of parent-child re-lationships becomes gradually less asym-metrical over the course of adolescence.Perceived power actually decreases duringmiddle adolescence. This reduced sense ofpower may be more perceived than real, asobservational studies have actually foundadolescents to have more influence in familydecisions than preadolescents (Jacob, 1974).Adolescents' striving for autonomy may in-creasingly expand the desired boundariesfor self-determination, thereby creatinga situation where desired levels of self-determination may outrace parents' willing-ness to grant that self-determination. Thus,adolescents may feel that parents givethem less power than they "ought" to have,even though their actual levels of self-determination and influence with parentsmay be greater in adolescence. Such argu-ments over autonomy issues may also be one

of the reasons why youths perceived conflictto be more frequent dnring this period.

Onr findings also indicate that somerapprochement is seen as occurring in lateadolescence. Perceived conflict was less fre-qnent, and the adolescent's sense of powerwas greater. Perceptions of support alsotended to be greater, despite the fact thatmost of the sample of college studentslived away from home. This reappraisal hasbeen reported by other investigators aswell (Pipp, Shaver, Jennings, Lamborn, &Fischer, 1985; Shaver, Furman, & Bnhr-mester, 1985). These relationships may beappreciated more when there are few oc-casions for power struggles over the manage-ment of their daily lives, the most commonbasis of conflicts (Montemayor, 1983). Itwonld be interesting to determine if parents'perceptions are also different when theirchildren are in high school and in college.

Proposition 3.—Sibling relationsbipswere expected to he seen as being less in-tense and more egalitarian with age. Con-sistent with this prediction, grade x sexANOVAs and follow-up tests revealed thatperceptions of support were highest in thefourth grade and lowest in the seventhgrade, with scores for tenth graders and col-lege students falling in the middle. Ratingsof conflict were substantially lower in the

Furman and Bnhrmester 111

two oldest grades. Similar findings werefound for punishment. Finally, emalyses ofrelative power scores revealed that individu-als perceived having more power in thetentliL grade and college than in the fourth orseventh grade.

These findings suggest that early in ado-lescence, youth are less reliant on these rela-tionships or want to see themselves as lessreliant, but they are still faced with the fre-quent clashes almost inherent in these rela-tionships. The perceptions of lower conflictin middle and late adolescence may reflectthe lower rates of interaction among siblingsat this age (Buhrmester & Furman, 1990;Raffaelli & Larson, 1987). This differencemay also reflect adolescents' tendency todistance themselves from family and investmore in peer relationships. Finally, the per-ceptions of greater support in late adoles-cence suggest that some rapprochement oc-curs in these relationships as well as thosewith parents.

Ratings also varied as a function of thesex of the subject. Girls saw their sibling re-lationships to be more supportive than didboys. On the other hand, girls thought thatthey had less power in their sibling relation-ships than boys indicated they had in theirs.Similar differences in perceptions of powerwith parents were found. It may be that par-ents and siblings foster traditional sex-rolestereotypes.

The age and sex differences, however,may raask individual differences in the rela-tionships. When we have interviewed chil-dren about their sibling relationships (e.g.,Furman & Buhrmester, 1985b), it appearedto us that the majority of these relationshipswere perceived as somewhat distant in ado-lescence, but some of them were importantsources of social support. An important taskfor future work is to study the nature andbases of such individual differences in thedevelopmental course of sibling relation-ships or others in the social network.

Proposition 4.—Grandparents andteachers were expected to be seen as lessfrequent sources of support as children growolder. As expected, a grade x sex ANOVArevealed that perceptions of support fromboth grandparents and teachers were high-est in the fourth grade. Grandparents andteachers are supplemental attachment fig-ures during early childhood, but adolescentsmay believe that they need them less often.In the case of teachers, the change from hav-ing one or two teachers in elementary school

to having many teachers in secondary schoolmay also reduce the likelihood of formingpersonal ties.

For grandparents, ratings of conflict andpunishment were also higher in the fourthgrade than in the tenth grade, suggestingthat the relationship becomes more distantwith age. There were also no grade differ-ences in the perceived power balance in re-lationships with grandparents, suggestingthat adolescent autonomy struggles may nottypically extend to relationships with grand-parents. Grandparents are generally not re-sponsible for the day-to-day management ofadolescents' lives, a frequent source offriction in adult-adolescent relationships. Bycontrast, and consistent with previous re-search (Epstein & McPartland, 1976; Hirsch& Rapkin, 1987), friction with teachers waselevated during early adolescence. Ratingsof punishment were higher and feelings ofpower were lower in the seventh grade thaneither the fourth or tenth grade. A similarpattern was found for conflict, except tbatthe difference between grades 4 and 7 didnot achieve significance.

Several sex differences were also foundin relationships with teachers. Girls thoughtthey received more support from teachers,but examination of the significant sex Xgrade interaction revealed that this differ-ence held only for fourth graders (girls'fourth M = 2.79, seventh M = 1.80, tenth M= 2.03, boys' fourth M = 2.42, seventh M =1.92, tenth M = 1.83). Boys' perceptions ofconflict and punishment were greater thangirls' (conflict M's 1.75 vs. 1.48; punishmentM's 2.01 vs. 1.72). Interestingly, relation-ships with teachers were the only ones inwhich sex differences in conflict and punish-ment were observed, which contrasts withthe well-documented sex difference in so-cial interactions with groups of peers (seeMaccoby, 1989). Apparently, snch differ-ences may not apply to perceptions of mostclose dyadic relationships.

Proposition 5.—The proposition that re-lationships with same-sex friends would beperceived as more supportive during adoles-cence received some support. A significantgrade effect for tbe composite support scoreswas found, but supplementary analyses re-vealed that the nature of the grade differ-ences varied substantially from provision toprovision. (In contrast, analyses of specificprovisions for other relationships revealedconsistent patterns across provisions.) Inparticular, intimacy and affection scores

112 Child Development

were significantly greater in the seventh andtenth grades than the fourth grade, whereascorresponding differences were not foundfor the other provisions. Perceptions of com-panionship and nurturance of the friendwere lower in college than in the tenthgrade, whereas no differences were ob-served on the other provisions or, in thecase of reliable alliance, scores were greaterin college. Numerous investigators, usingquestionnaires, interview, and observationaltechniques, have reported similar age differ-ences in intimacy (see Bnhrmester & Fur-man, 1987); the current study extends ourunderstanding by demonstrating that suchdifferences do not occur on all provisions ofsupport. The specific nature of these differ-ences is quite consistent with Sullivan's(1953) theory of the development of chum-ships. It is also noteworthy that perceptionsof intimacy and affection are greater duringearly and middle adolescence, the period ofgreatest family tension. This parallel lendscredence to the view that perception ofgreater dependence on friends is a conse-quence of the process of developing auton-omy from parents (Bios, 1967).

Conflict with friends was seen as lessfreqnent in the tenth grade and college thanin the fourth or seventh grade, whereas per-ceptions of power in the relationship weregreater in the tenth grade and college thanin the fourth grade.

Proposition 6.—As expected, relation-ships with romantic partners were seen asmore supportive by older youth. These find-ings are consistent with Snllivan's (1953)and Erikson's (1950) idea that the desires forsexuality and intimate exchange converge inromantic relationships during late adoles-cence. Analyses of conflict scores also re-vealed a significant grade effect, with scoresin college greater than those in the tenthgrade. For relative power, there was a sig-nificant interaction between grade and sex.Examination of cell means revealed thatboys saw an increase in power in the rela-tionship as they got older, whereas girls sawa decrease (boys' fourth M = 2.81, seventhM = 2.86, tenth M = 3.09, college M = 3.19,girls' fourth M = 3.10, seventh M = 2.84,tenth M = 2.70, college M = 2.85).

In other research we have conducted(Gavin & Furman, 1989), age differences inchildren's perceptions of their peer groupswere examined. The general peer group wasseen as less supportive and more confliotual

in early adolescence (grades 7-8) than inpreadolescence (grades 5-6), whereas herefriends and romantic partners were seen asmore supportive. Apparently, close dyadicrelationships are seen as sources of supportduring a difficult period when youths maybe vying for status in the general peer gronp.The contrasting findings illustrate the im-portance of distinguishing between thefunctions served by friendships and peergronps (Furman & Robbins, 1985).

Yonng men felt there was more supportfrom romantic partners than did youngwomen, M = 3.40 versus 3.22. Wheeler,Reis, and Nezlak (1983) found that collegestudents of both sexes were more satisfiedand more willing to disclose personal infor-mation when interacting with women thanmen. The present findings suggest this dif-ference may emerge by preadolescence.

Proposition 7.—Consistent with expec-tations and prior research (see Berndt, 1988),the grade x sex ANOVA of support scoresin same-sex friendships revealed that girls'ratings of support were significantly higherthan boys', M's = 3.62 versus 3.31. As notedpreviously, similar sex differences also ap-peared in perceptions of relationships withsiblings and mothers. On the other hand,boys saw their relationships with fathers andromantic partners to be more supportivethan did girls. Thus, the present network ap-proach makes it clear that sex differences insupport completely depend upon the spe-cific relationship.

Limitations and Future DirectionsThe present sample consisted primar-

ily of Caucasian children from middle- toupper-middle-class families. It is unclear ifa similar pattem of differences would befound in samples from different ethnic orsocial class backgrounds. Investigators havediscnssed ethnic differences in children'ssocial relationships, particularly familialones (see Powell, 1984), but as yet few stud-ies examining age differences in perceptionsof social networks in different subcultureshave been conducted.

A longitudinal stndy wonld be anotherimportant extension, as it would provide in-formation about change and continuity ofperceptions of different relationships andwould control for any cohort effects. In an-other sample, 45 fifth- and sixth-grade par-ticipants wbo had completed NRIs on thesame set of relationships except romanticpartners were readministered the measure

Furman and Buhrmester 113

3 years later. The results concerning meandifferences paralleled those presented here.Additionally, the amount of perceived sup-port in the various relationships was foundto be moderately stable (mean r = .39). Con-flict was less stable in most relationships, al-though it was remarkably stable in siblingrelationships (r = .54). The present resultsare also consistent with longitudinal studiesthat have examined changes in networkswith the transition to college (e.g.. Shaver etal., 1985). Such consistencies help rule outthe possibility tbat the observed differencesbetween high school and college studentsreflect differences between the general pop-ulation of high school students and that sub-set who attend college.

The results described here are based onchildren's self-reports about their relation-ships. The questions were phrased in con-crete terms to insure that all of the childreninterpreted them in a similar manner. Weavoided using abstract terms and conceptsbecause we did not want any observed dif-ferences to reflect developmental differ-ences in children's comprehension of theconcepts. If anything, then, the present re-sults may underestimate the size of age dif-ferences. For example, adolescents mayhave reported receiving even greater levelsof support from friends if they had beenasked about abstract types of support snch aspersonal validation. At the same time, theirreports are not objective accounts of theirrelationships, even thongh snch perceptionshave been found to be significantly corre-lated with other peoples' perceptions (Fur-man, Jones, Bnhrmester, & Adler, 1988). Infact, it seems likely that some perceptionsare not based on actual interactions (e.g., col-lege students reporting frequent support inparent-child relationships). An importanttask for subsequent research will be to de-velop paradigms for observing youths' in-teractions with various members of theirnetwork.

In general, future investigators shouldconsider simnltaneously examining chil-dren's relationships with various membersof their social network. By taking a networkperspective, we can move toward a compre-hensive picture of the development of socialrelationships.

ReferencesBarrera, N. (1981). Social support in the adjust-

ment of pregnant adolescents: Assessment is-

sues. In B. H. Gottlieb (Ed.), Social networksand social support (pp. 69—96). Beverly Hills,GA: Sage.

Banmrind, D. (1971). Gurrent patterns of parentalauthority. Developmental Psychology Mono-graph, 4(1), (Whole).

Berndt, T. J. (1988). The nature and significanceof children's friendships. In R. Vasta (Ed.),Annals of child development (Vol. 3, pp.155-186). Greenwich, GT: JAI.

Bios, P. (1967). The second individuation processof adolescence. Psychoanalytic Study of theChild, 12, 162.

Blyth, D. A. (1982). Mapping the social world ofadolescents: Issues, techniques, and prob-lems. In F. G. Serafica (Ed.), Social-cognitivedevelopment in context (pp. 240-272). NewYork: Guilford.

Bossard, J. H. S. (1948). The sociology of childdevelopment. New York: Harper & Brothers.

Bronfenbrenner, U. (1979). The ecology of humandevelopment. Gambridge, MA: Harvard Uni-versity Press.

Bryant, B. (1985). The neighborhood walk: Astudy of sources of support in middle child-hood from the child's perspective. Mono-graphs of the Society for Research in ChildDevelopment, 50(3, Serial No. 210).

Bnhrmester, D., & Fnrman, W. (1986). The chang-ing functions of friends in childhood: A neo-Sullivanian perspective. In V. J. Derlega &B. A. Winstead (Eds.), Friendship and socialinteraction (pp. 41-62). New York: Springer-Verlag.

Bnhrmester, D., & Furman, W. (1987). The devel-opment of companionship and intimacy.Child Development, 58, 1101-1103.

Buhrmester, D., & Furman, W. (1990). Age difFer-ences in perceptions of sibling relationshipsin childhood and adolescence. Child Devel-opment, 61, 1387-1398,

Gauce, A. M., Felner, R. D., & Primavera, J.(1982). Social support in high risk adoles-cents: Structural components and adaptiveimpact. American Joumal of Community Psy-chology, 10, 417-428.

Goates, D. L. (1987). Gender differences in thestructure and support characteristics of blackadolescents' social networks. Sex Roles, 17,667-687.

Epstein, J., & McPartland, J. (1976). The conceptand measurement of the quality of school life.American Educational Research Journal, 50,13-30.

Erikson, E. (1950). Childhood and society. NewYork: Norton,

Furman, W. (1984). Some observations on thestudy of personal relationships. In J. G. Mas-ters & K. Yarkin-Levin (Eds.), Interfaces be-

114 Child Development

tween developmental and social psychology(pp. 15-42). New York: Academic Press.

Fnrman, W. (1989). Developmental changes inchildren's social networks. In D. Belle (Ed.),Children's social networks and social sup-ports. New York: Wiley.

Furman, W., & Bnhrmester, D. (1985a). Ghil-dren's perceptions of the personal relation-ships in their social networks. DevelopmentalPsychology, 21, 1016-1024.

Furman, W., & Buhrmester, D. (1985b). Ghil-dren's perceptions oF the qualities oF sihlingrelationships. Child Developnient, 56, 448—461.

Furman, W., Jones, L., Buhrmester, D., & Adler,T. (1988). Ghildren's parents', and observers"perspective on sibling relationships. In P. G.Zukow (Ed.), Sibling interaction across cul-ture (pp. 165-183). New York: Springer-Verlag.

Fnrman, W., & Robbins, P. (1985). What's thepoint? Issues in the selection oFtreatment ob-jectives. In B. Schneider, K. Rubin, & J. Led-dingham (Eds.), Children's relations: Issuesin assessment and intervention (pp. 41—54).New York: Springer-Verlag.

Gavin, L., & Furman, W. (1989). The develop-ment oF cliques in adolescence. Develop-mental Psychology, 25, 827-834.

Gottman, J., & Mettetal, G. (1986). Speculationsabout social and aFFective development:Friendship and acquaintanceship through ad-olescence. In J. M. Gottman & J. G. Parker(Eds.), Conversations of friends: Specula-tions on affective development (pp. 192—240).Gambridge: Gambridge University Press.

Grotevant, H., & Gooper, G. (1986). Individuationin Family relationships: A perspective on in-dividual diFFerences in the development oFidentity and role-taking skill in adolescence.Human Development, 29, 82-100.

Hartup, W. W. (1979). The social worlds oF child-hood. Am,erican Psychologist, 34, 944-950.

Hirsch, B. J., & Rapkin, B. D. (1987). The transi-tion to junior high school: A longitudinalstudy oF self-esteem, psychological symptom-atology, school liFe, and social support. ChildDevelopment, 58, 1235-1243.

Hunter, F. T. (1985). Adolescents' perceptions oFdiscussions with parents and friends. Devel-opmental Psychology, 21, 433-440.

Hunter, F. T., & Youniss, J. (1982). Ghanges inFunctions oi three relationships during ado-lescence. Developmental Psychology, 18,806-811.

Jacob, T. (1974). Patterns oF Family conflict anddominance as a Function oF child age andsocial class. Developmental Psychology, 10,1-12.

Lewis, M., & Feiring, G. (1979). The child's so-cial network: Social object, social Functions,and their relationship. In M. Lewis & L. A.Rosenblum (Eds.), The child and its family(pp. 9-27). New York: Plenum.

Lewis, M., Feiring, G., & Kotsonis, M. (1984). Thesocial network oF the young child: A devel-opmental perspective. In M. Lewis (Ed.),Beyond the dyad (pp. 129-160). New York:Plenum.

Maccoby, E. E. (1990). Gender and relationships:A developmental account. American Psychol-ogist, 45, 513-520.

Margolin, G., & Patterson, G. R. (1975). Differen-tial consequences provided by mothers andFathers For their sons and daughters. Develop-mental Psychology, 11, 537-538.

Montemayor, R. (1983). Parents and adolescentsin conflict: All Families some oF the time andand some Families most of the time. Journalof Early Adolescence, 3, 83-103.

Noller, P. (1980). Gross-gender efFects in two-child Families. Developmental Psychology,16, 159-160.

Olson, P. H. (1977). Insiders' and outsiders" viewsoFrelationships: Research studies. In G. Lev-inger & H. L. Rausch (Eds.), Close relation-ships: Perspectives on the meaning of inti-macy. Amherst: University oF MassachusettsPress.

Pipp, S., Shaver, P., Jennings, S., Lamborn, S., &Fischer, K. W. (1985). Adolescents' theoriesabout the development oF their relationshipswith parents. Joumal of Personality and So-cial Psychology, 48, 991-1001.

Powell, G. J. (Ed.). (1984). The psychosocial de-velopment of minority children. New York:Brunner/Mazel.

Raffaelli, M., & Larson, R. W. (1987). Sibling in-teractions in late childhood and early adoles-cence. Paper presented at the meetings oFtheSociety for Research in Ghild Development,Baltimore, MD.

Reid, M., Landesman, S., Treder, R., & Jaccard, J.(1989). "My Family and Friends": Six- totwelve-year-old children's perceptions oF so-cial support. Child Development, 60,896-910.

Shantz, G. U. (1987). Gonflicts between children.Child Development, 58, 283-305.

Sharabany, R., Gershoni, R., & Hoffinan, J. E.(1981). GirlFriend, boyfriend: Age and sex diF-Ferences in intimate Friendship. Develop-mental Psychology, 17, 800-808.

Shaver, P., Furman, W., & Buhrmester, D. (1985).Aspects oFa liFe transition: Network changes,social skills and loneliness. In S. Duck & D.Perlman (Eds.), The Sage series in personalrelationships (Vol. 1, pp. 193-220). London:Sage.

Furman and Buhrmester 115

Steinberg, L. (1989). Adolescence. New York:KnopF.

Sullivan, H. S. (1953). The interpersonal theory ofpsychiatry. New York: W. W. Norton.

Tinsley, B. R., & Parke, R. D. (1984). Grandpar-ents as support and socialization agents. In M.Lewis (Ed.), Beyond the dyad (pp. 161-194).New York: Plenum.

Weiss, R. S. (1974). The provisions of social rela-

tionships. In Z. Rubin (Ed.), Doing unto oth-ers. Englewood Gliffs, NJ: Prentice-Hall.

Wheeler, L., Reis, H. T., & Nezlak, J. (1983).Loneliness, social interaction, and sex roles.Journal of Personality and Social Psychology,45, 945-953.

Youniss, J., & Smollar, J. (1985). Adolescent rela-tions with mothers, fathers, and friends. Ghi-cago: University of Ghicago Press.