Intimacy of Friendship, Interpersonal Competence, and A4justment during Preadolescence and Adolescence

BUHRMESTER, DUANE. Intimacy of Friendship, Interpersonal Competence, and Adjustment durinf,

hypotheses that (a) intimacy offriendshipis more integral to socioemotional adjustment during adolescence than preadolescence, and (b) that competence in close relationship skills is more

ered using a 2-step procedure ensuring that students rated only reciprocated friendships. Self- anc

among preadolescents. Significant age differences in coefficients were predominantly found foi

,s, intimacy has most often been equated

nature and significance of friendship during Qourard, 1979), although Sullivan (1953), and

early adolescence (Bemdt, 1982; Buhrmester & Furman, 1986; Gottman & Mettetal, 1987; SuIIivan, 1953). Friendships among preschool and elementary school-aged children revolve primarily around playmate activities and group acceptance, whereas adolescent friend-

more recently Reis and Shaver (1988), indicated that the core process of intimate interactions is not disclosure, per se, butratherthe experiences of feeling understood, validated, and cared for that accompany self-disclosure, As a feature of relationships, intimacy usually

J functioning (Buhrmester & furman, iy?H>; Sullivan, 1953), few studies have empirically examined diese implications. The present study investigated whether there are age differences between preadolescence and adolescence in how important intimacy of friendship is to adjustment and growth of interpersonal competencies. The construct of intimacy has been used

tion." Although exactly what features comprise intimate collaborative friendships is open to debate, at a minimum they involve engaging in mutual activities, self-disclosure, and reciprocal feelings of satisfaction with the relationship (Furman & Robins, 1985; Mannarino, 1976; Sullivan, 1953). The current study is primarily concerned with intimacy as ^ feature of relationships.

Los Angeles Unified Schools for participating in this study. I we

Development, the University of Texas at Dallas, Richardson, T>

1102

Child Development
(roughly 8 to 12 years) and adolescent (older than about 13 years) samples to determine whether there is an increase with age in the more broadly defined (e.g., anxiety, depression, hostility, sociabilily, and self-esteem).

ies have documented diat, during early adolescence, children's fnendships become more intimate as indicated by morefrequentcompanionate exchanges, personal disclosure, and provision of emotional support (Buhrmes-

strength of the association between friendship intimacy and psychosocial adjustment

ter & Furman, 1987; Gottman & Mettetal, SmoUar, 1985; see Steinberg,

^^j^^^^ jg^gj j ^ ^^,^t^ adolescent friend^^ demands greaterfecUityin a number of ^ relattomhip competencies (here re-

tic relationships and only partially overlap WiA Ae playmate skUls demanded in early childhood. lie association beAccording to Ais reasoning, adolescents ind a^ustment. A who lack Aese specific relationship com^le links between petencies are more likely to have difficulty self-esteem and friendship intimacy. Self- achieving intimacy in Aeirfriendships.They reported closeness offriendshiphas been are likely to have fewerfriendshipsAat are found to be positively related to self-esteem more superficial in nature, and Aus Aese among pieadolescents (Bukowski & Hosa, youAs are also likely to be at greater risk for 1989; Cauce, 1986; Maimarino, 1978), and actfustment problems. At the same time, besimilarfindingshave beenreportedin some cause Aey are unable to estabUsh intimate studies of adolescents (O'DonneU, 1976). friend^ps, Aey probably miss out on experiOAer studies, however, haverevealedweak ences Aat couldfinAerpromote Ae growA correlations betweenfriendshipintimacy and of Aese competencies (Hartup & Sancilio, self-esteem (Bemdt, 1987; BlyA & Traeger, 1986). That is, many of Ae subtleties concem1987; Jourard, 1979). Almost no one, how- ing when and what to self-disclose and how
ever, has directly compared preadolescent to provide emotional support are probably

erac, ,,ould expect uie assuciauon between interpersonal competence and intimacy in friendship to be stronger during adolescence than preadolescence because it is during adolescence that
' lteractional processes that demand and lescents.

issues concerning: (a) the extent of convergence among self- and friend's ratings of fnendship quality and interpersonal competence, and (b) the validity ofthe Interpersonal Competence Questionnaire (Buhrmester et al., 1988) for use with preadolescents and ado-

io studies have attempted to investigate

r1i/]opmental change directly. Man-

Subjects

The initial preadolescent sample con-

sisted of 133 fifth and sixth graders (68 girls), ages 10-13 years (M = 11.3), whereas the initial adolescent group consisted of 100 eighth and ninth graders (44 girls), ages 13-16 years ._.^ ^ , re likely to (M = 14.4), fi-om ethnically and racially dibe involved in stable close friendships than verse metropolitan Los Angeles schools. The children who scored low on these measures, preadolescents represented 85% of the chilThese investigators, however, studied only a dren from seven mixed-aged classrooms in w age range of children (10-12 years) two elementary schools, whereas the adolesid consequently .1.. J. did J not . _ii attempt iI to l,,J. evaluate ^^^t.^ .*:*,,J. cents constituted ] AAOf^ .,f *!> 44% J . 1 ofthe f youths : from nine .'hether there were developmental changes classrooms in one junior high school. The competencies. Fur- lower p percentage of adolescents taking part in in the importance of these competences thermore, they did not assess altruism and the study w o perspective taking as it was manifested parental consent forms. After the friendship
specifically in the context of intimate friendpairing procedure (explained below), the

6, 1979) and McGuue and Weisz o scored high on "

ship (see Price & Ladd, 1986). This is problematic because we know that children benave aiirerentiy towara n^ienas ana
nonfKends (Hartup, 1983). One general obstacle to researcb on adoavailability of adequate measures. Althougb

sample consisted of 102 preadolescents and 70 adolescents.

^'TtJ^o st^^rocedure w fHendship intoacy^'lnitidl^, which pairs of students reciprocally reported ,__

lescent social competence has been the unmethods have been developed to assess the skills needed by younger cbildren to gain peer-group acceptance (Asher, 1983; Dodge, Pettit, McClaskey, & Brown, 1986; Putallaz & Gottman, 1981), little work bas been done to develop measures that specifically assess the interpersonal competencies called for in intimate relationships (Ford, 1982). Recently, lege-aged adolescents that specifically portant in both close fnendships and romantic
relationships. This measure was modified . ^"^

cedures were used to identify relationships in that they were friends. Students tben rated the intimacy of their relationships with reciprocal friends. This method of assessment is similar to that employed by Bemdt (1981), and differs in important ways (which will be discussed later) from tbe more common practice of baving students rate tbeir seif-proclaimed "bestfriends."Students also rated

Procedures ^- P " * J ^"""'l.

"""-e gathered in I

slightly in the present study to investigate e cross-sectional study reported -ps of preadolescents and adolese administered several questioninvestigate age differences in the trengths of associations among friendship incentral suL ously, the data also addressed tv

sex schoolmates participating in the study and instructed to indicate which peers tbey considered close friends. Cbildren were told that "close friends are kids you know very well, spend a lot of time witb in and out of school, and wbo you talk to about things that happen

1104

Child Development

loiescent'sZ in fte studv II g ^ e r s a t the fifHi/sixth

Adolescent Interpersonal Competence Cuestionnoire (AlCP^.-This newly devel"P*"* * ^ " ' " questionnaire assessed the fol'""''"S ^"^ domains of competence that are : in close relationships (sample items

fHends that they have been neglectful or... considerater), and initiation of friendships ("How good istfiisperson at phoning friends ' things togethe??"). The AICQ was developed by modifying Interpersonal Competence Questionnaire ' t was o r ^ a l l y devel, ^ciose^friendships and romantic

preadolescents and

with reciprocal close iriends. li schoolmates whom subjects had noniinated fhends (but wbo bad not reciprocated the nomination) were identified so every child had two peers to rate on questionnaires (see below). These nonreciprocated fiiendships
were not included in the analyses. Instances

of reciprocal fairly goodfriendswere also identified, and students rated one fairly good fnend on the fnendship and competence
measures. The findings for fairly good fnends are not reported here, however, bpranse nf

Studrelationships (Buhrmester et al., 1988). 1 "^ * e ^ " ^ 8 ^ ICQ demonstra " " demonstrate that its jf^ they c co J ^<^l^ " " adequately reliable, they l t " ^ ^ *^* predicted i5ve-factar simple struc'f"'- ^^ *''>' con-elsrte in predictable and discriminant ways with theoretically related jariables. In rewordingtiieICQ it f " ^ " ' '"ff * vocabulary appronate for young adolescents without chang" *<= substantive content of questions.

space limitations and becai very few significant

ratings. Measures

ere were ith these

Fricm.

.,

tionnaire consisted of L _ _

man's (1978) S-point rating scale to indies Companionship, Intimate the level of competence and comfort that ea... target child would have in handling each type of situation (e.g., "1 = Poor at this; would be 1985). An Ulustiative intimate disclosure ii reads; "How often do you share secrets and so uncomfortable and unable to handle tbis private feelings with this person?" Subjects rated the qualities of their relationships with identified fnends using a 5-point Likert-type scale (e.g., "1 = Never or hardly ever" to "5 = VERY often or EXTREMKLY much").
tionships Inventory (Furmai-

Duane Buhrmester 1105

alpba coefficients were computed separately for the two age groups and were bigb: preadolescents (.93) and adolescents (.92).

Hostility, r = - .06 and - .37 (among preadolescents and adolescents, respectively); Sociar = -.15 and

bility-Anxiety/Depression,

''^'^' Hostility-Anxiety/Depression, r = .62

adolescents*Vand'adoiescen" (.12)"'"'"

Tt2tiTZ"l}^t7^''ll!^Cf.!^]

Socioemotional Adjustment (SA).A dimensions of adjustment, althougb the mod40-item self-report questionnaire was devel- esttemoderate size of the associations sugoped for this study as a relatively brief yet gests that the scales nonetheless assess disbroad measure of several spberes of self- tinguishable aspects of adjustment. Whereas perceptions of socioemotional adjustment. It information about the validity of these scales was necessary to develop this measure beis limited, the available data indicate that cause existing measures of child adjustment scale scores are correlated in theoretically were either too narrow in scope (i.e., assess- predictable ways with other mfo'^m,.^ nf -ning only one sphere of adjustment) or took cialfrinctioning(E ' " more Ume to administer than was allotted by school personnel. Items similar to those found Results 'le Child Behavior Check List, the Child _ ,^^.,,^,^,^ , . , , ,

point Likert-tyS' s^c'Se"'('?grn = "Never*OT^ " P <^ *'"' Consistent with findings from not at all" to "5 = Very often or very much"). P^'* ''^'^''^' ^ two-way analysis (ANOVA) of Factor analyses revealed three orthogo- sex, F(l,168) = 4.02, p < .05, witb girls (M = nal dimensions labeled Sociability, Hostility, 3.22) rating theirfriendshipsas mr ' ' and Anxiety/Depression (see Buhrmester, than boys (M = 3.0" ~ 1989, for details). Scale scores were computed tions, adolescents d by averaging items tbat assess eacb dimen- fiiendships as more i sion. Example items included: "How well do cents. Thisfindingis surprising b yon work witb other people?" (Sociability; studies examining * "' 10 items), "How often do you lose your simi' ,wofage

im .80 to .87forpreadolescents and .72

varied according to sex, age, and thw. - , ^ ^ ...

' '

rater (self vs. closefriend).There was a significant effect of type of rater, F(l,167) =

1106

CbUd Development

= 3.62). These differences are similar to those found for college students' ICQ ratings of self, andfriends(Buhrmester et al., 1988). There was also a significant efifect of gri "
F(l,167) = 16.25, p < 01 ith d l s

friendship intimacy. Fischer's procedure (with alpha set at .05) revealed that selfreported friendship " "

(M = _3.57) tihaii age 357) rated d as les.s competent^ p g di ). This age differof adjustment
studies where, as (Stipek & Maclver, between friend-rated

Significant age difierences

adjust-

intimacy and

ler, they lower ower their estimates h significant i i f i t 1989).^ There were no other effects or interactions. i

ment, seemingly because of the relatively

preadolescent

stronger correlations for the

.sample. .j.^^ correlations contained in Table 3 r r . " - f^ contention that interpersonal competence is related to friendship intimacy - ' lg adolescence. Both self- and fHend-

Associations among Friendship Intimacy,

friend-rated

friendship

intimacy scores for the

Adjustment, and Competence Table 2 provides clear evidence that friendship intimacy isrelatedto adjustment for adolescents. All four dimensions of .selfreported adjustment were significandy ' moderately (r = -.24 to .56) both self-andfriend; '

adolescent sample. It is noteworthy that selfratings of competence were related just as strongly to fHendratingsoffriendshipintimacy as to self-ratings offriendshipinUmacy. ., - J . .^ , ... . L u

Duane Buhrmester 1107

Sociabiuiy ..

nd ratings of fnendship intimacy, aithough

re evide'nt""^' ' "

e expecte

irechon

Secondary Analyses
Correspondence between self- andfriend

both age groups:fromB^ = .04 to R^ = .24 for

preadolescents (B^ change, F[l,101] = 12.46,

,__. .T.,.- extent - . ^ - ^ . of . r agreement. between p < .01) andfromB R' rongs.-The = = .19 to B" = .34 for change, F[l,69] = 7.08, p < self- andfriendratingsis of interest for both adolescents (H^ chan, conceptual , and methodological reasons. In- .03). Thus, _ although .03). Thus, although AICQ scores shared traclass correlations were computed between some variance with closefriends'FI scores, a
self-report and friend-reported FI scores, substantial proportion of a friend's appraisal

There were high levels of agreement among

adolescents'ratings,r = .81, and moderate

of a subject's competence was independent of

friendship.

the perceived intimacy of the

"

"

agpreadolescents'ratings,r = .38. " " " jdfrBm

These analyses indic?t *ot fi-.nrlc'Kn.! of competence and fri

Bemdt and Das (1987) have argued that the quality of children's friendships influences their perceptions of theirfriends'personalities. If this is true, thenfriends'judg-

whose fnendships were rate andfriendreports)as compa ing, and satisfying reported th competent, more sociable, le

1108

Child Development
how the results are interpreted. Subjects did not rate their self-identified "best" friends (which has been the most common practice in past studies) but rather rated "close" friends who had reciprocally nominated themselves as closefriends.This procedure may in part account for the M u r e tofindthe expected developmental increase in mean ratings of -^gy friendship intimacy. As bestfriendrelation-

anxious/depressed, and have higher selfesteem compared to peers involved in less intimatefriendships.Thesefindingsare consistent with the claim that the processes that create intimacy in adolescentfriendshipsare important determinants of mental health and the growth of competence.

many problems may overburdenfriendswith unpleasant disclosures or may be less able to decenter and proyouths lacking in interpersonal competence The lack of age differences in intimacy may have greater difficulty establishing and may have also been related to thatfactthat a maintaining intimatefriendships.Although lower proportion of adolescents'friendsparthe correlational nature of theresultsmake it ticipated in the study (owing to the lower proimpossible to draw conclusions about the di- portion of the total school population particirection of causation, it seems likely that a pating), thereby reducing the likelihood that ' ex- adolescents were paired (by chance) with iidi- their "best"friend.This aspect of sample iter- selection (which in large part was aresultof and, in tum, adjustment the inherently dtSerent institutional and sodetermine how successcioecological structiires of elementary and

, ,._ .

r^__,_^ . . , - , .

-inior h i * sehools) may have worked against

d, self-reportedfriendshipintire strongly related to adjustment .^.-sonal competence during adoles n preadolescence. These hndings support the view diat preadolescent friendships typically do not demand the interpersonal competencies called for in more mature forms of closerelationships.Tliese findings are also consistent with the view that, during preadolescence, parents pby a retovely

"f*" macy. Aimougn mese were noc necessaruy suojects' very bestfriendships,they appear to i,averepresentedrelsUionshipsfromsmall cir^i^ ^f ^igjgfiiendshipsamong these youAs. ^ ^ know this because: (a) most children nominated only a handfol of peers as close f^^^^ (j ^ ^bout four for preadrfescents and

(wo for adolescents), and (b) children's ratings

greater role than peers infrilfilltagchildren s of the intimacy of theseretoionshipswere social needs and shaping their acjustment ^^jy g^^y i^^j ^^^ the ratings of self(Buhrm (Buhrmester & Furman, 1987). On the o*er identified bestfriendshipsthat have been rehand, t i friendship intimacy, but merely modest ferences in the direction. direction. For ferences in expedsd the expedsd For , _ , , ^ , ^ UTOBUCU UJ aucuum .ur ii m dl frd reports of f intimacy i adolescents, friend were chins nrforclospfriendshinswe can be moderately correlated with adjustment and joiSbly c o n S n t that the p^cesses that competence. The discrepancies in findings thought to mediate the associations betw across self- andfriendreports cannot easily be friendship and socioemotional adapt^ explained. Further work is needed to repli- ^ disclosure-validation, provision of supcate or explain these findings. port, and the exercise of intarpersooal compeThe methods used to assess friendship tence) are ^iplicable to close as well as best andfriendshipintimacy have implicationsforfriendships.(My fmther research will reveal

Duane Buhrmester 1109

whether thefindingsreported here also bold true for best fnendships more narrowly defined. However, because children spend a great deal oftimeinteracting witb peers who fkll in the "closefriend"category, the results reported here describe a significant arena of children's social lives. meet standard criteria for internal consistency reliability and are correlated in meaningfril ways wilii theoretically related variables. The moderately strong agreement between selfratings and ratings by closefriendsfiirther suggests that the AICQ assesses meaningfril and visible dimensions of social bebavior.

child has unreciprocated (and often unreal- sures. These questionnt istic) feelings of closeness toward a peer. This perceptions of theirfriendships,competenis important because, in theory, unilateral cies, and adjustment, and accordingly the friendships are not expected to be positively findings might have differed if behavioral related to social adaptation. Infeet,the corre- methods were used. Even if children's perlations between intimacy ratings and other ceptions differ substantiallyfromobserved variablesreportedhere are, on average, some- behavior, however, children's perceptions are what sti-onger than those reported by investi- important to investigate in tbeir own rigbt gators gathering self-identified bestfriendrat- (Olson, 1977). There is substantial evidence ve appraisals ings_ (e.g., Bemdt, 1987; Blyth & Traeger, indicating that subjective appraisals ar are sig_.. _.. . . ts of behavior (Bar _

that, by excluding ratings of unilateral friend- 1986). One strength of the present stiidy was ships, the assessment procedure reduced a that by gathering both self-and peer ratings, I significant source of "noise" in the data that was able to determine the degree to which may have acted to attenuate the correlations children hold common perceptions of their found in previous studies. friendships and levels of competence. Tbe results show that, at least among adolescents, __ ^ ^ ..ot only was there agree-

-.-nt between sources of ratings, but similar

?e-re- pattems of con-elations were observed across mong and between self- and close-friend ratings, ir friend- These convergentfindingssuggest that these
'-" -" o p t i o n s are anchored in a shared social

found for adolescents than preadolescents. This age difference suggests that tbere is greater reciprocity in perceptions and feelings of closeness infriendshipamong adolescents than preadolescents. The surprising strength of the correlations between self- and friend ratings among adolescents suggests that adolescents are very aware of how close or distant

the qualities of their in^ so refiect a more dynamic relational process in which adolescents regulate their own feelings and behavior to\ fiiends so as totiyto matcb what they ceive as theirfriend'slevel of investment m Taken together, the results show tbat there is an initial basis for confidence in the reliability and validity ofthe AICQ as measure of perceptions of interpersonal tence, particularly for adolescents.

1110

ChUd Development
perceptions ofthe pereonal relationships in

andfriendshipon perceptions ofthe personal-

d relationships with parSalzinger, J, Antrobers, & Socio/ networks of chil-

(Eds.), Conversations of friends: Speadatiom Hartup, W. W. (1983). The peer system. In E. M. zation, personality, and social development (pp. 103-196). New Yorlc; Wiley. Hartup, W. W., & Sancilio, M. F. (1986). Children's -

on affective development (pp. 192-240). New York: Cambridge University Press. Hetherington (Ed), P. H. Mussen (Series Ed.), Handbook of child psychology: Vol. 4. Sociali-

id college students (pp.

171-194). Hillsdale, NJ: Erlbanm. uhrmester, D. (1989). Manual for the Sodoemo-

tional Adjustment Questionnaire. Unpublished manuscript. Program m Psychology, University of Texas at Dallas, Richardson, TX.
ingftincdonsof friendship in childhood: A

(Eds.), Social behavior in autism (pp. 61-80).

neo-SuUivanian perspective. In V.J.Derles '

sonality and Social Psychology, 55, 991-1008.

velopmental Psychology, 12,55

" T. J. Bemdt '& G. Ladd (Eds.),

males. Journal of Genetic Psychology, 133,

986). Social networks and social Exploring the effects of early adodships. American Journal of Com-

fiiends'hip

and altruism in preadolesoent girls.

Psychiatry. 42, 280-.284. McGuire, K., & Weisz, J. R. (1982). Social cognition

hology, 14, 607628. L, Pettit, G. S., McClaskey, C. L., &

hs of the Society for Related

and behavior correlates of preadolescent chumships. ' " ' " '

:perience. New York: Wiley.

ger & H. L. Raush (Eds.), Close

Dnane Buhrmester

1111

behavioral assessment of children and families (pp. 121-149). London:

Child Development,

58,283-305.

Steinberg, L. (1989). Adolescence. New York: Al-

laver, P. (1988). Intimacy as an ch in personal relationships (pp.

Sullivan, H. S. (1953). The interpersonal theortj of Youniss, J., & Smollar, J. (1985). Adolescent rela-

idon: Wiley.

tionships with mothers, fathers, and friends.