Friendships and Adaptation in the Life Course - Psychology ...
Friendships and Adaptation in the Life Course - Psychology ...
Friendships and Adaptation in the Life Course - Psychology ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Psychological Bunet<strong>in</strong> Copyright 1997 by <strong>the</strong> American Psychological Association, Inc.<br />
1997, Vol. 121, No. 3, 355-370 0033-2909/97/$3.00<br />
<strong>Friendships</strong> <strong>and</strong> <strong>Adaptation</strong> <strong>in</strong> <strong>the</strong> <strong>Life</strong> <strong>Course</strong><br />
Willard W. Hartup<br />
University of M<strong>in</strong>nesota, Tw<strong>in</strong> Cities<br />
Nan Stevens<br />
University of Nijmegen<br />
To consider friendships <strong>and</strong> <strong>the</strong>ir significance through <strong>the</strong> life course requires, first, differentiation<br />
of deep structure (i.e., reciprocity) from surface structure (i.e., <strong>the</strong> social exchange) <strong>and</strong>, second,<br />
assessment with<strong>in</strong> a multifaceted framework that simultaneously emphasizes hav<strong>in</strong>g friends, <strong>the</strong><br />
identity of one's friends, <strong>and</strong> relationship quality. Hav<strong>in</strong>g friends is correlated with a sense of well-<br />
be<strong>in</strong>g across <strong>the</strong> life span, but developmental outcome also depends on <strong>the</strong> identity of one's friends<br />
as well as <strong>the</strong> quality of one's relationships with <strong>the</strong>m. Greater attention needs to be given to <strong>the</strong><br />
manner <strong>in</strong> which friendships differ from one ano<strong>the</strong>r, cont<strong>in</strong>uities <strong>and</strong> changes across major develop-<br />
mental transitions, <strong>and</strong> differentiation of developmental pathways through which friendship experi-<br />
ence contributes to <strong>in</strong>dividual outcome.<br />
Most <strong>in</strong>dividuals build <strong>the</strong>ir lives around friends as well as<br />
families. A friendship consists ma<strong>in</strong>ly of be<strong>in</strong>g attracted to<br />
someone who is attracted <strong>in</strong> return, with parity govern<strong>in</strong>g <strong>the</strong><br />
social exchanges between <strong>the</strong> <strong>in</strong>dividuals <strong>in</strong>volved. <strong>Friendships</strong><br />
carry expectations that "best" friends will spend more time<br />
with one ano<strong>the</strong>r than o<strong>the</strong>r persons, offer<strong>in</strong>g one ano<strong>the</strong>r emo-<br />
tional support, <strong>in</strong>clud<strong>in</strong>g loyalty, trust, <strong>in</strong>timacy, <strong>and</strong> fun. Not<br />
everyone has friendships, but <strong>the</strong>se relationships are sought after<br />
<strong>and</strong> valued from early childhood through old age. <strong>Friendships</strong><br />
are ranked among <strong>the</strong> th<strong>in</strong>gs that matter most to children, adoles-<br />
cents, <strong>and</strong> adults (Kl<strong>in</strong>ger, 1977).<br />
Although friendships have been scrut<strong>in</strong>ized by social scien-<br />
tists for a century or more (Monroe, 1898 ), <strong>the</strong>ir developmental<br />
significance is difficult to specify beyond adolescence. First,<br />
studies with adults are not as numerous as studies with children<br />
<strong>and</strong> adolescents; relevant longitud<strong>in</strong>al studies are especially rare<br />
(Blieszner & Adams, 1992). Second, studies with adults are<br />
likely targeted on special groups (e.g., retirees <strong>and</strong> widows),<br />
thus render<strong>in</strong>g normative comparisons difficult. Third, adult<br />
friendships are studied with a relatively narrow range of methods<br />
(<strong>in</strong>terviews <strong>and</strong> questionnaires); child <strong>and</strong> adolescent friend-<br />
ships have been studied with more diverse methodologies, <strong>in</strong>-<br />
clud<strong>in</strong>g experimental <strong>and</strong> observational techniques. Fourth, sub-<br />
ject matter varies with age: Studies with adults deal with <strong>in</strong>ter-<br />
personal attraction, similarities between friends, <strong>and</strong> social<br />
support (Blieszner & Adams, 1992); studies with children <strong>and</strong><br />
adolescents deal with orig<strong>in</strong>s <strong>and</strong> behavioral manifestations,<br />
cognitive expectations, <strong>and</strong> developmental outcomes (Hartup,<br />
1996).<br />
Willard W. Hartup, Institute of Child Development, University of<br />
M<strong>in</strong>nesota, Tw<strong>in</strong> Cities; Nan Stevens, Department of Psychogerontology,<br />
University of Nijmegen, Nijmegen, The Ne<strong>the</strong>rl<strong>and</strong>s.<br />
Some sections of this article are based on presentations by WiUard<br />
W. Hartup <strong>and</strong> Nan Stevens at dedication ceremonies for <strong>the</strong> Rutten<br />
Institute for Psychological Research, University of Nijmegen, Nijmegen,<br />
The Ne<strong>the</strong>rl<strong>and</strong>s, on October 20, 1995.<br />
Correspondence concern<strong>in</strong>g this article should be addressed to Willard<br />
W. Hartup, Institute of Child Development, University of M<strong>in</strong>nesota, 51<br />
East River Road, M<strong>in</strong>neapolis, M<strong>in</strong>nesota 55455. Electronic mail may<br />
be sent via Internet to hartup@vx.cis.umn.edu.<br />
355<br />
Our major objective is to assess <strong>the</strong> developmental signifi-<br />
cance of hav<strong>in</strong>g friends, who one's friends are, <strong>and</strong> quality of<br />
friendship relations <strong>in</strong> a life-span perspective. Accord<strong>in</strong>gly, we<br />
discuss first what it means to take a life-course perspective on<br />
friendship relations. We suggest that life-course views <strong>in</strong> this<br />
area must <strong>in</strong>volve two levels of analysis--what might be called<br />
deep structure <strong>and</strong> surface structure. Second, we argue that<br />
friendship <strong>and</strong> life-course adaptation cannot be understood un-<br />
less a clear dist<strong>in</strong>ction is made among hav<strong>in</strong>g friends, <strong>the</strong> iden-<br />
tity of one's friends, <strong>and</strong> friendship quality. We exam<strong>in</strong>e <strong>the</strong>se<br />
three friendship dimensions <strong>in</strong> turn, extend<strong>in</strong>g <strong>the</strong> treatment<br />
given to <strong>the</strong>m from two earlier essays. In <strong>the</strong> first essay <strong>in</strong> this<br />
series (Hartup, 1995), <strong>the</strong> general case for multidimensional<br />
assessment was argued; <strong>in</strong> <strong>the</strong> second (Hartup, 1996), this<br />
three-dimensional framework was used to exam<strong>in</strong>e <strong>the</strong> signifi-<br />
cance of friendships <strong>in</strong> child <strong>and</strong> adolescent development. Now,<br />
<strong>in</strong> this article, we extend <strong>the</strong> analysis by differentiat<strong>in</strong>g deep<br />
<strong>and</strong> surface structures <strong>in</strong> friendship experience <strong>and</strong> by apply<strong>in</strong>g<br />
<strong>the</strong> three-dimensional framework across <strong>the</strong> life course.<br />
Friendship Relations <strong>in</strong> <strong>the</strong> <strong>Life</strong> <strong>Course</strong>: The Dist<strong>in</strong>ction<br />
Between Deep <strong>and</strong> Surface Structures<br />
To consider friendship <strong>in</strong> a life-span perspective <strong>in</strong>volves two<br />
ma<strong>in</strong> assumptions. First, one assumes that friendships bear on<br />
developmental outcome, beg<strong>in</strong>n<strong>in</strong>g <strong>in</strong> early childhood <strong>and</strong> ex-<br />
tend<strong>in</strong>g through old age. Some adaptational concomitants are<br />
concurrent, so one f<strong>in</strong>ds that friends support one ano<strong>the</strong>r <strong>in</strong><br />
coord<strong>in</strong>ated play <strong>and</strong> problem solv<strong>in</strong>g dur<strong>in</strong>g <strong>the</strong> early years <strong>and</strong><br />
that toddlers who are sought out as friends are more generally<br />
competent than those who are not (Howes, 1983). O<strong>the</strong>r conti-<br />
nuities extend across time <strong>and</strong> situation, so adults who have<br />
friends are known to meet various developmental challenges,<br />
such as widowhood, with better outcomes than <strong>in</strong>dividuals who<br />
do not have friends (Connidis & Davies, 1990; Dykstra, 1995a).<br />
Some of <strong>the</strong>se contributions may be cumulative <strong>in</strong> <strong>the</strong> sense that<br />
<strong>the</strong> contributions of one friendship augment <strong>the</strong> contributions<br />
of o<strong>the</strong>rs. One does not necessarily assume that friendships have<br />
<strong>the</strong> same significance throughout life but that <strong>the</strong>se relationships<br />
are developmental resources at all ages.<br />
Second, one assumes that developmental changes <strong>in</strong> <strong>in</strong>dividu-
356 HARTUP AND STEVENS<br />
als trigger changes <strong>in</strong> relationships <strong>and</strong> that <strong>the</strong>se changes extend<br />
through <strong>the</strong> life course. For example, (a) among very young<br />
children, new coord<strong>in</strong>ations <strong>and</strong> comb<strong>in</strong>ations <strong>in</strong> memory<br />
emerge at about 2 years of age, appear<strong>in</strong>g to serve as a basis for<br />
an <strong>in</strong>crease of coord<strong>in</strong>ation <strong>and</strong> collaboration <strong>in</strong> <strong>the</strong> <strong>in</strong>teraction<br />
between toddlers <strong>and</strong> <strong>the</strong>ir friends (Brownell, 1986); (b)<br />
changes associated with puberty trigger <strong>in</strong>creases <strong>in</strong> <strong>in</strong>timacy<br />
between opposite-gender friends but not same-gender friends<br />
(Sharabany, Gershoni, & Hofman, 1981 ); <strong>and</strong> (c) when strength<br />
<strong>and</strong> mobility decl<strong>in</strong>e <strong>in</strong> old age, one' s friends must <strong>in</strong>itiate social<br />
contact <strong>and</strong> activities, <strong>the</strong>reby chang<strong>in</strong>g relationship reciproci-<br />
ties (O'Connor, 1993; Rawl<strong>in</strong>s, 1992).<br />
To consider friendships across <strong>the</strong> life course requires <strong>the</strong><br />
differentiation of deep structure, which characterizes <strong>the</strong>se rela-<br />
tionships, from surface structure. We use deep structure to refer<br />
to <strong>the</strong> social mean<strong>in</strong>g (essence) of relationships <strong>and</strong> surface<br />
structure to refer to <strong>the</strong> social exchanges that characterize <strong>the</strong>m<br />
at any given moment or <strong>in</strong> any given situation--a convention<br />
that is similar to <strong>the</strong> one used <strong>in</strong> l<strong>in</strong>guistics (Chomsky, 1965).<br />
The friendship deep structure can be identified by researchers<br />
ask<strong>in</strong>g participants to describe a friend or friends--especially<br />
an ideal or hypo<strong>the</strong>tical friend--<strong>and</strong> <strong>the</strong>n content analyz<strong>in</strong>g <strong>the</strong><br />
results. Accord<strong>in</strong>gly, among young children, friendship expecta-<br />
tions have been found to center ma<strong>in</strong>ly on common activities<br />
<strong>and</strong> concrete reciprocities ("We play." "And I give <strong>the</strong>m food,<br />
so <strong>the</strong>y give me food back"; Goodnow & Burns, 1985, p. 120).<br />
School-aged children describe friends as underst<strong>and</strong><strong>in</strong>g, loyal,<br />
<strong>and</strong> trustworthy; children expect to spend time with <strong>the</strong>ir friends,<br />
share <strong>in</strong>terests, <strong>and</strong> engage <strong>in</strong> self-disclosure with <strong>the</strong>m ("A<br />
good friend is someone who likes you <strong>and</strong> spends time with<br />
you <strong>and</strong> forgives you <strong>and</strong> doesn't actually bash you up"; p.<br />
120). Children do not use words like <strong>in</strong>timate to describe <strong>the</strong>ir<br />
friends, but such constructs beg<strong>in</strong> to differentiate <strong>the</strong>se relation-<br />
ships shortly before adolescence (Bigelow, 1977; Selman,<br />
1980). Older <strong>in</strong>dividuals describe an ideal friend ma<strong>in</strong>ly as<br />
be<strong>in</strong>g supportive (dependable, underst<strong>and</strong><strong>in</strong>g, <strong>and</strong> accept<strong>in</strong>g), a<br />
confidant, <strong>and</strong> trustworthy. Most important, high school seniors,<br />
newlyweds, middle-aged parents, <strong>and</strong> soon-to-be retirees differ<br />
relatively little from one ano<strong>the</strong>r <strong>in</strong> <strong>the</strong>ir emphasis on <strong>the</strong>se<br />
reciprocities when <strong>the</strong>y describe an ideal friend. Similarity be-<br />
tween friends (shar<strong>in</strong>g <strong>in</strong>terests, experiences, <strong>and</strong> activities as<br />
well as communicative compatibility) is regarded, however, as<br />
an important friendship attribute among adolescents but de-<br />
creases subsequently (L. Weiss & Lowenthal, 1975).<br />
Close exam<strong>in</strong>ation of <strong>the</strong>se results reveals differences as well<br />
as similarities <strong>in</strong> <strong>the</strong> manner <strong>in</strong> which friends are described by<br />
younger <strong>and</strong> older <strong>in</strong>dividuals. Differences occur ma<strong>in</strong>ly <strong>in</strong> <strong>the</strong><br />
number of psychological constructs used, complexity <strong>and</strong> orga-<br />
nization of <strong>in</strong>formation <strong>and</strong> ideas, <strong>and</strong> flexibility with which<br />
this <strong>in</strong>formation is used to describe a friend, These age differ-<br />
ences are similar to those recorded <strong>in</strong> person perceptions (Lives-<br />
ley & Bromley, 1973), which probably reflect general cognitive<br />
development. Similarities across age occur <strong>in</strong> <strong>the</strong> extent to which<br />
reciprocity <strong>and</strong> mutuality emerge <strong>in</strong> <strong>the</strong> mean<strong>in</strong>g structure. Reci-<br />
procity does not have narrow connotations <strong>in</strong> <strong>the</strong>se descriptions;<br />
most <strong>in</strong>dividuals do not describe friendships as exchange rela-<br />
tionships <strong>in</strong> which <strong>in</strong>puts <strong>and</strong> resource exchanges must be equiv-<br />
alent or one <strong>in</strong>dividual's behavior must exactly match ano<strong>the</strong>r's.<br />
Never<strong>the</strong>less, children <strong>and</strong> adults of all ages consider <strong>the</strong>se rela-<br />
tionships to be marked by reciprocation, that is, mutuality--<br />
<strong>the</strong> giv<strong>in</strong>g <strong>and</strong> tak<strong>in</strong>g, <strong>and</strong> return<strong>in</strong>g <strong>in</strong> k<strong>in</strong>d or degree. On this<br />
basis, <strong>the</strong>n, we argue that <strong>the</strong> friendship deep structure <strong>in</strong> West-<br />
ern cultures is best described as "symmetrical reciprocity" (see<br />
H<strong>in</strong>de, 1979; <strong>and</strong> Youniss, 1980).<br />
Because reciprocity constitutes <strong>the</strong> friendship deep structure<br />
across <strong>the</strong> life course, certa<strong>in</strong> outcomes or consequences should<br />
be evident at all ages. Social reciprocities should be significant<br />
sources of security (Bowlby, t969), self-worth (Sullivan,<br />
1953), <strong>and</strong> well-be<strong>in</strong>g (R. S. Weiss, 1973). These conditions,<br />
<strong>in</strong> turn, should support successful cop<strong>in</strong>g, especially with devel-<br />
opmental transitions such as school entrance, puberty, workforce<br />
entrance, marriage, child bear<strong>in</strong>g, spouse' s death, <strong>and</strong> retirement<br />
(Magnusson, Statt<strong>in</strong>, & Allen, 1985). On <strong>the</strong> basis that friend-<br />
ship reciprocities support cop<strong>in</strong>g <strong>in</strong> this fashion, one expects<br />
<strong>the</strong>se relationships to promote good outcomes regardless of age.<br />
Surface structures--<strong>the</strong> actual exchanges that occur between<br />
friends--differ from situation to situation <strong>and</strong> from early child-<br />
hood to old age. Social reciprocities between two toddlers, for<br />
example, are manifested ma<strong>in</strong>ly <strong>in</strong> time spent toge<strong>the</strong>r <strong>and</strong> con-<br />
nected <strong>in</strong>teractions (Howes, 1989). Reciprocities between k<strong>in</strong>-<br />
dergarten friends are more elaborate socially but rema<strong>in</strong> basi-<br />
cally concrete, consist<strong>in</strong>g of play <strong>and</strong> shar<strong>in</strong>g (Howes, 1983).<br />
Reciprocities among adolescent friends consist of common ac-<br />
tivities (especially socializ<strong>in</strong>g), augmented by self-disclosure<br />
<strong>and</strong> expectations of loyalty <strong>and</strong> trust (Berndt, 1989). Reciproci-<br />
ties among adult friends are centered on work activities, <strong>and</strong><br />
many friendships become "fused" or "blended" with work<br />
(Hess, 1972; W<strong>in</strong>stead, Derlega, & Montgomery, 1995). O<strong>the</strong>r<br />
fusions occur with marriage; husb<strong>and</strong>s <strong>and</strong> wives frequently<br />
share friends with whom issues <strong>in</strong> marriage <strong>and</strong> family relations<br />
are stressed (Hess, 1972; Rawl<strong>in</strong>s, 1992). Reciprocities among<br />
older persons are more separated from family <strong>and</strong> work but<br />
concern support issues <strong>and</strong> companionship. Surface structures<br />
at this time <strong>in</strong>clude friends exchang<strong>in</strong>g letters <strong>and</strong> gifts, talk<strong>in</strong>g<br />
on <strong>the</strong> telephone, do<strong>in</strong>g favors for one ano<strong>the</strong>r, <strong>and</strong> express<strong>in</strong>g<br />
affection <strong>and</strong> mutual respect (Rawl<strong>in</strong>s, 1992; Shea, Thomp-<br />
son, & Blieszner, 1988). What friends talk about <strong>and</strong> what <strong>the</strong>y<br />
do with one ano<strong>the</strong>r thus change with age <strong>and</strong> circumstance.<br />
We believe that <strong>the</strong>se changes <strong>in</strong> surface structure ma<strong>in</strong>ly<br />
reflect changes <strong>in</strong> <strong>the</strong> developmental tasks (Havighurst, 1953)<br />
confront<strong>in</strong>g <strong>the</strong> <strong>in</strong>dividuals <strong>in</strong>volved. So, among young children,<br />
friendship reciprocities support <strong>the</strong> acquisition of new social<br />
skills, especially expertise <strong>in</strong> cooperation <strong>and</strong> o<strong>the</strong>r socially co-<br />
ord<strong>in</strong>ated skills (e.g., conflict resolution). Among older children<br />
<strong>and</strong> adolescents, <strong>the</strong> emphasis shifts to <strong>in</strong>timacy <strong>and</strong> o<strong>the</strong>r ex-<br />
changes that support a sense of self-identity, sensitivity with<br />
respect to <strong>the</strong> needs of o<strong>the</strong>rs, <strong>and</strong> ma<strong>in</strong>tenance of mutually<br />
oriented relationships with "agemates"--<strong>in</strong>clud<strong>in</strong>g members<br />
of both <strong>the</strong> same <strong>and</strong> opposite genders (Sullivan, 1953). Among<br />
young adults, friendshiPS center on a new collection of develop-<br />
mental tasks--work <strong>and</strong> family issues (Hess, 1972). F<strong>in</strong>ally,<br />
when older <strong>in</strong>dividuals confront major life transitions that re-<br />
quire reorganization of one's lifestyle <strong>and</strong> expectations, friend-<br />
ship content changes once aga<strong>in</strong>. Clearly, <strong>the</strong> socializ<strong>in</strong>g func-<br />
tions of <strong>the</strong>se relationships are closely tied to <strong>the</strong> developmental<br />
challenges that <strong>the</strong> <strong>in</strong>dividuals <strong>in</strong>volved face, whatever <strong>the</strong>ir age.<br />
Some researchers suggested that friendship surface structures<br />
are ma<strong>in</strong>ly socializ<strong>in</strong>g <strong>in</strong> childhood <strong>and</strong> adolescence because<br />
<strong>the</strong>y support <strong>in</strong>novation <strong>in</strong> both social underst<strong>and</strong><strong>in</strong>g <strong>and</strong> social<br />
skill, but <strong>the</strong>y are ma<strong>in</strong>ly susta<strong>in</strong>ed <strong>in</strong> old age, through which
normative traditionalism is supported (Hess, 1972). Research<br />
shows, however, that <strong>the</strong>se differences may be overstated: So-<br />
cialization, for example, rema<strong>in</strong>s an important friendship func-<br />
tion throughout <strong>the</strong> life course. Older <strong>in</strong>dividuals assist one<br />
ano<strong>the</strong>r <strong>in</strong> <strong>the</strong> reorganization of lifestyles <strong>and</strong> identities, such<br />
as when retirees seek <strong>the</strong> company of those already retired<br />
(Adams, 1987), recent widows are aided by long-term widows<br />
<strong>in</strong> reorganiz<strong>in</strong>g <strong>the</strong>ir lives as s<strong>in</strong>gle women (Bankoff, 1983;<br />
Stevens, 1995), <strong>and</strong> new residents <strong>in</strong> hous<strong>in</strong>g for older persons<br />
are assisted <strong>in</strong> <strong>the</strong>ir adaptation to <strong>the</strong> new sett<strong>in</strong>g by long-term<br />
residents (Armstrong & Goldsteen, 1990; Hochschild, 1973).<br />
Similarly, susta<strong>in</strong><strong>in</strong>g functions may occur among friends across<br />
a wide age range, not merely among older adults <strong>and</strong> <strong>the</strong>ir<br />
friends. Certa<strong>in</strong>ly, among older adults, "values, which <strong>in</strong> con-<br />
temporary terms are dated, can be freely expressed with people<br />
whose life span has given <strong>the</strong>m comparable experiences <strong>and</strong><br />
outlook" (Jerrome, 1981, p. 190), but friendships also support<br />
normative commitments among children <strong>and</strong> adolescents (Kan-<br />
del, 1978b).<br />
To summarize, <strong>the</strong> friendships of toddlers <strong>and</strong> <strong>the</strong> friendships<br />
of older adults are both similar <strong>and</strong> different. On <strong>the</strong> one h<strong>and</strong>,<br />
social <strong>in</strong>teraction between both younger <strong>and</strong> older friends is<br />
marked by symmetrical reciprocity (deep structure). On <strong>the</strong><br />
o<strong>the</strong>r h<strong>and</strong>, <strong>the</strong> social exchanges that occur between friends<br />
(surface structure) reflect salient developmental tasks. The<br />
friendship deep structure is thus developmentally stable,<br />
whereas friendship surface structure is not. To consider friend-<br />
ships <strong>in</strong> life course terms, <strong>the</strong>n, requires that attention be given<br />
to both cont<strong>in</strong>uities <strong>and</strong> discont<strong>in</strong>uities <strong>in</strong> <strong>the</strong>ir manifestations<br />
<strong>and</strong> developmental implications. Two hypo<strong>the</strong>ses, can be ad-<br />
vanced. (1) Good friends support well-be<strong>in</strong>g from early child-<br />
hood onward through <strong>the</strong> reciprocities that occur between <strong>the</strong><br />
friends; consequences may extend from immediate to future<br />
adaptations. (2) Friends support one ano<strong>the</strong>r <strong>in</strong> cop<strong>in</strong>g with <strong>the</strong><br />
developmental (time-limited) challenges that confront <strong>the</strong>m, <strong>the</strong><br />
consequences of which may also extend from present to future<br />
adaptations. We exam<strong>in</strong>e empirical evidence relat<strong>in</strong>g to <strong>the</strong>se<br />
hypo<strong>the</strong>ses <strong>in</strong> subsequent sections.<br />
Three Friendship Dimensions<br />
Researchers <strong>and</strong> cl<strong>in</strong>icians commonly differentiate among<br />
<strong>in</strong>dividuals--both children <strong>and</strong> adults--accord<strong>in</strong>g to whe<strong>the</strong>r<br />
or not <strong>the</strong>y have friends. Cl<strong>in</strong>icians, for example, want to know<br />
whe<strong>the</strong>r a troubled child or adult client has friends, a circle of<br />
friends, or a best friend--but usually little else <strong>in</strong> this doma<strong>in</strong>.<br />
Researchers enter "has friends" <strong>in</strong>to <strong>the</strong>ir regression equations<br />
but seldom more. Whe<strong>the</strong>r or not one has friends receives this<br />
emphasis for two reasons: First, mak<strong>in</strong>g <strong>and</strong> keep<strong>in</strong>g friends<br />
requires good social skills, so hav<strong>in</strong>g friends is an <strong>in</strong>dication<br />
of (a proxy for) good social adjustment. Second, acquir<strong>in</strong>g<br />
friends <strong>and</strong> ma<strong>in</strong>ta<strong>in</strong><strong>in</strong>g <strong>the</strong>se relationships require a person to<br />
be o<strong>the</strong>r oriented as well as self-oriented, espouse egalitarian<br />
attitudes, <strong>and</strong> manage conflicts with one's companions <strong>in</strong> con-<br />
structive ways. On <strong>the</strong> basis of <strong>the</strong>se arguments, hav<strong>in</strong>g friends<br />
<strong>in</strong>creases one's social skills <strong>and</strong> well-be<strong>in</strong>g, which, <strong>in</strong> turn,<br />
<strong>in</strong>crease one's likelihood of mak<strong>in</strong>g <strong>and</strong> keep<strong>in</strong>g friends.<br />
Friendship experience, however, cannot be reduced only to<br />
hav<strong>in</strong>g friends. <strong>Friendships</strong> vary greatly from <strong>in</strong>dividual to '<strong>in</strong>di-<br />
FRIENDSHIPS IN THE LIFE COURSE 357<br />
vidual as well as from companion to companion (Hartup,<br />
1996). First, enormous variation occurs <strong>in</strong> who one's friends<br />
are. Some companions are cooperative, outgo<strong>in</strong>g, <strong>and</strong> socially<br />
skilled; o<strong>the</strong>rs are not. Sometimes companions are antisocial;<br />
at o<strong>the</strong>r times, <strong>the</strong>y are not. Such differences are obvious, but<br />
<strong>the</strong>ir significance is often overlooked. New evidence suggests,<br />
though, that <strong>the</strong> identity of one's friends (i.e., <strong>the</strong>ir personal <strong>and</strong><br />
social characteristics) accounts for more outcome variance than<br />
whe<strong>the</strong>r one has a friend; antisocial children <strong>and</strong> adolescents<br />
are examples (Dishion, Andrews, & Crosby, 1995).<br />
Second, friendships differ from one ano<strong>the</strong>r qualitatively<br />
(Hartup, 1996), that is, <strong>in</strong> <strong>the</strong>ir content or normative functions<br />
(e.g., what <strong>the</strong> two <strong>in</strong>dividuals do toge<strong>the</strong>r), <strong>the</strong>ir construc-<br />
tiveness (e.g., whe<strong>the</strong>r one normally resolves conflicts with<br />
one's friends us<strong>in</strong>g negotiation or <strong>the</strong> "power-assertion" <strong>in</strong>flu-<br />
ence strategy), <strong>the</strong>ir closeness (e.g., if one's companions spend<br />
time toge<strong>the</strong>r, <strong>the</strong>y engage <strong>in</strong> many different activities as opposed<br />
to a few <strong>and</strong> <strong>the</strong>ir exchanges <strong>in</strong>volve self-disclosure), <strong>the</strong>ir sym-<br />
metry (e.g., whe<strong>the</strong>r friends <strong>in</strong>fluence one ano<strong>the</strong>r equally or<br />
"social power" is distributed unequally), <strong>and</strong> <strong>the</strong>ir affective<br />
character (e.g., whe<strong>the</strong>r friendships are supportive <strong>and</strong> secure<br />
or nonsupportive <strong>and</strong> conflict ridden). Qualitative features are<br />
well recognized <strong>in</strong> research on mo<strong>the</strong>r-<strong>in</strong>fant <strong>and</strong> marital rela-<br />
tionships (Bowlby, 1969; Gottman, 1979) but not <strong>in</strong> relation to<br />
friendships <strong>and</strong> <strong>the</strong>ir developmental implications.<br />
<strong>Friendships</strong> may thus be assets or liabilities, depend<strong>in</strong>g on<br />
who one's friends are <strong>and</strong> <strong>the</strong> quality of one's relationships with<br />
<strong>the</strong>m. Overly romanticized views of <strong>the</strong>se relationships distort<br />
what <strong>the</strong>y may contribute to development across <strong>the</strong> life course.<br />
Indeed, <strong>the</strong> identity of one's friends <strong>and</strong> <strong>the</strong> quality of one's<br />
relationships with <strong>the</strong>m may be related to developmental out-<br />
come more closely than hav<strong>in</strong>g friends.<br />
Occurrence<br />
Hav<strong>in</strong>g Friends<br />
Consider <strong>the</strong> occurrence of hav<strong>in</strong>g friends <strong>and</strong> what research-<br />
ers know about it. First, toddlers sometimes <strong>in</strong>teract with one<br />
ano<strong>the</strong>r preferentially <strong>and</strong> <strong>the</strong> word friend enters <strong>the</strong>ir vocabu-<br />
lary dur<strong>in</strong>g <strong>the</strong> third or fourth year (Howes, 1983). Various<br />
data sets suggest that about 75% of nursery school children are<br />
<strong>in</strong>volved <strong>in</strong> reciprocated friendships, as measured by observed<br />
time spent <strong>in</strong> one ano<strong>the</strong>r's company, nursery school teachers'<br />
reports, <strong>and</strong> maternal <strong>in</strong>terviews (H<strong>in</strong>de, Titmus, Easton, &<br />
Tampl<strong>in</strong>, 1985; Howes, 1983). This figure rises slightly through<br />
adolescence when 80% to 90% of teenagers report hav<strong>in</strong>g mu-<br />
tual friends, usually <strong>in</strong>clud<strong>in</strong>g one or two best friends <strong>and</strong> several<br />
"close" friends or "good" friends (Van der L<strong>in</strong>den & Dijkman,<br />
1989). Frequencies rema<strong>in</strong> high (90%) among adults <strong>in</strong> midlife<br />
<strong>and</strong> <strong>the</strong>n decl<strong>in</strong>e, but older adults more commonly have friends<br />
than not (Wright, 1989; Schutze & Lang, 1993). Small numbers<br />
of <strong>in</strong>dividuals have no friends as adults (6-7%)--a number<br />
that <strong>in</strong>creases to 12% for women <strong>and</strong> 22% for men over Age<br />
65 <strong>in</strong> <strong>the</strong> United States (Fischer & Phillips, 1982) <strong>and</strong> 19%<br />
for older persons <strong>in</strong> <strong>the</strong> Ne<strong>the</strong>rl<strong>and</strong>s (Pearl Dykstra, personal<br />
communication, December 7, 1995).<br />
Second, friendship networks are small among preschool chil-<br />
dren, averag<strong>in</strong>g about 1.7 friends for boys <strong>and</strong> 0.9 for girls,<br />
whereas school-aged children average 3.0-5.0 best friends, de-
358 HARTUP AND STEVENS<br />
pend<strong>in</strong>g on whe<strong>the</strong>r one counts unreciprocated as well as recip-<br />
rocated choices (Hall<strong>in</strong>an, 1980). This figure rema<strong>in</strong>s relatively<br />
constant dur<strong>in</strong>g adolescence <strong>and</strong> early adulthood (Cairns,<br />
Lueng, Buchanan, & Cairns, 1995), although to make compari-<br />
sons across studies is difficult because of methodological varia-<br />
tions. Network size is greatest among newlyweds, who report<br />
an average of 7.6 friends; <strong>the</strong>n decl<strong>in</strong>es to 4.7 aga<strong>in</strong> by middle<br />
age; <strong>and</strong> rises slightly (to 6.0) among persons about to retire<br />
(Lowenthal, Thurnher, & Chiriboga, 1975). Network size de-<br />
cl<strong>in</strong>es among older adults between Ages 55 <strong>and</strong> 90 (Dickens &<br />
Perlman, 1981; Dykstra, 1995b) due to mostly <strong>the</strong> loss of "ca-<br />
sual" friends. Close friends, however, are reta<strong>in</strong>ed <strong>in</strong>to old, old<br />
age, <strong>and</strong> older persons have about <strong>the</strong> same number of close<br />
friends as middle-aged adults (Field, 1995; Lang & Carstensen,<br />
1994). Girls' networks are ord<strong>in</strong>arily smaller <strong>and</strong> more exclu-<br />
sive than boys' dur<strong>in</strong>g early <strong>and</strong> middle childhood (Eder &<br />
Hall<strong>in</strong>an, 1978), but this situation reverses dur<strong>in</strong>g adolescence<br />
(Cairns et al., 1995 ). Gender differences <strong>in</strong> network size among<br />
adults are not consistent across <strong>the</strong> relevant studies.<br />
Third, time spent with friends varies over <strong>the</strong> life course,<br />
although estimates are extremely difficult to compare from <strong>in</strong>-<br />
vestigation to <strong>in</strong>vestigation; metrics <strong>and</strong> methods vary greatly.<br />
On <strong>the</strong> one h<strong>and</strong>, time spent with friends is greatest <strong>in</strong> middle<br />
childhood <strong>and</strong> adolescence, amount<strong>in</strong>g to 29% of time awake<br />
among teenagers. On <strong>the</strong> o<strong>the</strong>r h<strong>and</strong>, middle-aged adults spend<br />
only 7% of <strong>the</strong>ir time <strong>in</strong>teract<strong>in</strong>g with friends, <strong>and</strong> those over<br />
Age 65 spend 9% of <strong>the</strong>ir time this way (Larson & Bradney,<br />
1988; Larson, Zuzanek, & Mannell, 1985 ). Older women spend<br />
more time <strong>in</strong>teract<strong>in</strong>g with friends (<strong>and</strong> relatives) than men, <strong>and</strong><br />
widowed men <strong>and</strong> women spend more time with friends than<br />
married <strong>in</strong>dividuals (Altergott, 1988).<br />
Fourth, children who have friends at one age are likely to<br />
have <strong>the</strong>m at o<strong>the</strong>r ages (Elicker, Englund, & Sroufe, 1992),<br />
thus illustrat<strong>in</strong>g a cont<strong>in</strong>uity <strong>in</strong> social relations also suggested by<br />
retrospective <strong>in</strong>terviews with older persons. Based on <strong>in</strong>terviews<br />
with older participants, Mat<strong>the</strong>ws (1986) identified three "per-<br />
sonological" types as differentiated by friendship styles--<strong>in</strong>de-<br />
pendents enjoy friendly, satisfy<strong>in</strong>g social relationships through-<br />
out <strong>the</strong>ir lives but never have close or <strong>in</strong>timate friendS; discern-<br />
<strong>in</strong>g <strong>in</strong>dividuals report hav<strong>in</strong>g a small number of very close<br />
friends throughout childhood, adolescence, <strong>and</strong> adulthood; <strong>and</strong><br />
acquisitive <strong>in</strong>dividuals always have a relatively large number of<br />
friends <strong>and</strong> expect friends always to be available. Cont<strong>in</strong>uity is<br />
also suggested by <strong>the</strong> fact that close friends of many older<br />
persons are <strong>the</strong> same <strong>in</strong>dividuals who were <strong>the</strong>ir friends earlier<br />
<strong>in</strong> life (Field, 1995). Long-term friendships rest on shared his-<br />
tories, accumulated experiences, <strong>and</strong> simultaneously mov<strong>in</strong>g<br />
through major developmental transitions (Hess, 1972; Rawl<strong>in</strong>s,<br />
1992). Long-term cont<strong>in</strong>uities result from selection too: Old<br />
friends are consciously ma<strong>in</strong>ta<strong>in</strong>ed <strong>in</strong> preference to <strong>the</strong> mak<strong>in</strong>g<br />
of new ones <strong>in</strong>to very old age (Lang & Carstensen, 1994).<br />
Behavior With Friends <strong>and</strong> Nonfriends<br />
Behaviors that differentiate friends from nonfriends have been<br />
exam<strong>in</strong>ed <strong>in</strong> more than 80 studies with children <strong>and</strong> adolescents;<br />
<strong>the</strong> results generally verify <strong>the</strong> friendship deep <strong>and</strong> surface struc-<br />
tures described earlier. Reciprocity anchors <strong>the</strong>se relationships<br />
generally through <strong>the</strong> first 2 decades; surface content changes<br />
with age. Several narrative reviews of <strong>the</strong>se studies have been<br />
published (Hartup, 1989, 1996) as well as a meta-analysis (A.<br />
F. Newcomb & Bagwell, 1995 ). In <strong>the</strong>se studies, researchers did<br />
not demonstrate <strong>the</strong> developmental significance of friendships<br />
directly but specified some of <strong>the</strong> social processes through<br />
which friendship experience <strong>and</strong> adaptation are l<strong>in</strong>ked.<br />
We summarize one representative <strong>in</strong>vestigation <strong>in</strong> this article<br />
• along with <strong>the</strong> meta-analytic results as a basis for <strong>the</strong> rema<strong>in</strong>der<br />
of <strong>the</strong> review. In an empirical study, Hartup, Daiute, Zajac,<br />
<strong>and</strong> Sholl (1995) exam<strong>in</strong>ed conversations between friends <strong>and</strong><br />
nonfriends (10-year-olds) <strong>in</strong> an <strong>in</strong>nercity magnet school while<br />
<strong>the</strong> children wrote stories collaboratively on a computer, Stories<br />
were about <strong>the</strong> tropical ra<strong>in</strong> forest--a'subject matter that <strong>the</strong><br />
children had studied dur<strong>in</strong>g a 6-week science project. Children<br />
were assigned to three groups: (a) those who wrote stories<br />
<strong>in</strong>dividually on 4 different days; (b) those who wrote <strong>in</strong>dividu-<br />
ally on Day 1, <strong>the</strong>n <strong>the</strong> next 2 with a friend, <strong>and</strong> <strong>the</strong> f<strong>in</strong>al day<br />
alone; <strong>and</strong> (c) those who collaborated with a nonfriend ra<strong>the</strong>r<br />
than a friend. Results <strong>in</strong>dicate that friends did not talk more<br />
dur<strong>in</strong>g collaboration than nonfriends but, never<strong>the</strong>less, (a) en-<br />
gaged <strong>in</strong> more mutually oriented <strong>and</strong> fewer <strong>in</strong>dividualistic utter-<br />
ances; (b) agreed with one ano<strong>the</strong>r more often (but did not<br />
disagree more readily); (c) repeated <strong>the</strong>ir own <strong>and</strong> friend's<br />
assertions more often; (d) posed alternatives <strong>and</strong> provided elab-<br />
orations more frequently; (e) spent twice as much time as non-<br />
friends talk<strong>in</strong>g about writ<strong>in</strong>g content, <strong>the</strong> vocabulary used, <strong>and</strong><br />
writ<strong>in</strong>g mechanics; <strong>and</strong> (f) spent less time engaged <strong>in</strong> "off-<br />
task" talk. Pr<strong>in</strong>cipal component analyses confirm that friends'<br />
talk was assertively collaborative--a f<strong>in</strong>d<strong>in</strong>g that is rem<strong>in</strong>iscent<br />
of <strong>the</strong> dialogues between experts <strong>and</strong> novices reported <strong>in</strong> o<strong>the</strong>r<br />
social problem-solv<strong>in</strong>g studies (Rogoff, 1990). The stories also<br />
show that, overall, <strong>the</strong> ones written collaboratively by friends<br />
were better than those written by nonfriends--a difference that<br />
seems to rest on <strong>the</strong> better use of st<strong>and</strong>ard English by friends<br />
ra<strong>the</strong>r than <strong>the</strong> narrative elements <strong>in</strong>cluded <strong>in</strong> <strong>the</strong> stories.<br />
The meta-analysis (A. E Newcomb & Bagwell, 1995) sum-<br />
marizes friend versus nonfriend differences among children <strong>and</strong><br />
adolescents <strong>in</strong> terms of four broad b<strong>and</strong> categories: positive<br />
engagement (i.e., more talk, smil<strong>in</strong>g, <strong>and</strong> laughter generally<br />
occur among friends than nonfriends), conflict management<br />
(i.e., friends use disengagement <strong>and</strong> negotiation vs. power asser-<br />
tion proportionally more frequently than nonfriends), task activ-<br />
ity (i.e. task orientation as opposed to off-task orientation is<br />
greater among friends than nonfriends), <strong>and</strong> relationship proper-<br />
ties (i.e., equality <strong>in</strong> <strong>the</strong> exchange as well as mutuality <strong>and</strong><br />
affirmation are greater among friends than nonfriends). The<br />
data with<strong>in</strong> <strong>and</strong> across studies show that, behaviorally speak<strong>in</strong>g,<br />
<strong>the</strong> relationship affordances of "be<strong>in</strong>g friends" differ from <strong>the</strong><br />
affordances of "be<strong>in</strong>g acqua<strong>in</strong>tances." Friends are socially ac-<br />
tive with one ano<strong>the</strong>r, <strong>and</strong> <strong>the</strong>ir <strong>in</strong>teraction is marked by mutual-<br />
ity, effective conflict management, <strong>and</strong> task orientation, <strong>the</strong>reby<br />
reflect<strong>in</strong>g <strong>the</strong> deep structure revealed <strong>in</strong> <strong>the</strong> descriptions of<br />
friends (L. Weiss & Lowenthal, 1975).<br />
Certa<strong>in</strong> changes <strong>in</strong> <strong>the</strong> <strong>in</strong>teraction between friends can be<br />
observed with <strong>the</strong> onset of adolescence, although <strong>the</strong>se ex-<br />
changes rema<strong>in</strong> anchored <strong>in</strong> reciprocation. First, although nega-<br />
tive gossip occurs more commonly <strong>in</strong> talk between friends than<br />
between nonfriends throughout childhood <strong>and</strong> adolescence, pos-<br />
itive gossip differentiates friends <strong>and</strong> acqua<strong>in</strong>tances only <strong>in</strong> ado-<br />
lescence. Second, self-disclosure newly differentiates among<br />
friends <strong>and</strong> acqua<strong>in</strong>tances <strong>in</strong> adolescence <strong>and</strong> so does m<strong>in</strong>dread-
<strong>in</strong>g, that is, attributions that people make to o<strong>the</strong>rs' actions,<br />
motives, or personalities (Gottman & Mettetal, 1986). Conflicts<br />
with friends do not occur as frequently as commonly supposed<br />
(Laursen & Coll<strong>in</strong>s, 1994) <strong>and</strong> are reported about as frequently<br />
by adolescents as children. When choos<strong>in</strong>g between competition<br />
or shar<strong>in</strong>g with <strong>the</strong>ir friends, though, adolescents compete less<br />
<strong>and</strong> share more than children (Berndt, Hawk<strong>in</strong>s, & Hoyle,<br />
1986).<br />
Differences <strong>in</strong> <strong>the</strong> exchanges that occur between adult friends<br />
<strong>and</strong> nonfriends have been studied ma<strong>in</strong>ly because <strong>the</strong>se establish<br />
<strong>the</strong> greater closeness of friends as compared with acqua<strong>in</strong>tances.<br />
On that basis, more numerous exchanges occur between friends<br />
than acqua<strong>in</strong>tances (Berg, 1984); self-disclosure occurs more<br />
frequently <strong>and</strong> is deeper (Altman & Taylor, 1973; Berg, 1984);<br />
friends are more directive <strong>and</strong> authoritative with one ano<strong>the</strong>r<br />
than nonfriends. More diverse resources are exchanged by<br />
friends than nonfriends (Berg, 1983), although <strong>the</strong> specific re-<br />
sources exchanged differ: Affection, services, <strong>and</strong> status are<br />
exchanged more frequently by friends than nonfriends but not<br />
goods, money, <strong>and</strong> <strong>in</strong>formation (Tornblom & Fredholm, 1984).<br />
Friends are more likely to give benefits to one ano<strong>the</strong>r on <strong>the</strong><br />
basis of need or desire than nonfriends, whereas <strong>the</strong> latter are<br />
more likely to give benefits depend<strong>in</strong>g on whe<strong>the</strong>r <strong>the</strong>y have<br />
received <strong>the</strong>m previously <strong>the</strong>mselves (Mills & Clark, 1982).<br />
F<strong>in</strong>ally, friends th<strong>in</strong>k of <strong>the</strong>mselves as a unit more frequently<br />
than acqua<strong>in</strong>tances, <strong>the</strong>ir exchanges are more satisfy<strong>in</strong>g <strong>in</strong>tr<strong>in</strong>si-<br />
cally, <strong>and</strong> relationships are more last<strong>in</strong>g than between mere<br />
acqua<strong>in</strong>tances (see also Berg & Clark, 1989). Results <strong>in</strong>dicate<br />
that <strong>the</strong> reciprocities undergird<strong>in</strong>g friendships among younger<br />
children <strong>and</strong> adolescents are amplified among adults by a com-<br />
plex set of attributions that concern closeness <strong>and</strong> <strong>the</strong> nature of<br />
relationships. The developmental orig<strong>in</strong>s of <strong>the</strong>se attributions<br />
<strong>and</strong> <strong>the</strong>ir elaborations have not been studied.<br />
Companionship <strong>and</strong> talk (referred to earlier as positive en-<br />
gagement) cont<strong>in</strong>ue to differentiate friends from nonfriends <strong>in</strong><br />
middle <strong>and</strong> old age (Blieszner & Adams, 1992). Relationship<br />
talk is more topical <strong>and</strong> role related among young adult <strong>and</strong><br />
middle-aged friends than among ei<strong>the</strong>r adolescents or older<br />
adults. Shar<strong>in</strong>g, resource exchanges, emotional support, <strong>and</strong><br />
o<strong>the</strong>r <strong>in</strong>dications of mutuality are salient <strong>in</strong> friendship <strong>in</strong>terac-<br />
tions, especially those among adult friends that occur dur<strong>in</strong>g<br />
crises such as divorce (G<strong>in</strong>sberg, 1986). Problem solv<strong>in</strong>g<br />
among older persons is notably more symmetrical with friends<br />
than with relatives (Kaye & Monk, 1991 ). Although conflicts<br />
occur frequently between adult friends (as among children),<br />
<strong>the</strong>y are less commonly acknowledged than conflicts <strong>in</strong> marital<br />
relationships. Conflicts among adult friends are concentrated <strong>in</strong><br />
two areas (Argyle & Furnham, 1983): emotional disagreements<br />
(beliefs, values, <strong>and</strong> social partners) <strong>and</strong> criticisms (over life-<br />
style, habits, <strong>and</strong> personal issues). Among older persons, con-<br />
flicts ma<strong>in</strong>ly concern normative expectations that apply to older<br />
persons <strong>and</strong> resource <strong>in</strong>equities affect<strong>in</strong>g older persons (C. B.<br />
Fisher, Reid, & Melendez, 1989).<br />
To summarize, social <strong>in</strong>teractions between friends, as com-<br />
pared with nonfriends, are more engaged, mutually oriented, <strong>and</strong><br />
symmetrical across <strong>the</strong> life course. Social exchanges between<br />
friends, that is, <strong>the</strong>ir content, change over time but certa<strong>in</strong> struc-<br />
tural dimensions--especially <strong>the</strong>ir mutuality <strong>and</strong> symmetrical<br />
reciprocity--characterize friendships at all ages. Observational<br />
studies have seldom been conducted with adults, except <strong>in</strong> a<br />
FRIENDSHIPS IN THE LIFE COURSE 359<br />
few <strong>in</strong>stances with college roommates (G<strong>in</strong>sberg & Gottman,<br />
1986) <strong>and</strong> older residents of an apartment complex (Hochschild,<br />
1973). Consequently, age differences <strong>in</strong> behavior with friends<br />
cannot always be specified from direct observation. Observa-<br />
tional studies are badly needed, however, because of experimen-<br />
tal evidence show<strong>in</strong>g that exchange equivalences <strong>and</strong> reciprocal<br />
self-disclosure are not always concomitants of social attraction<br />
(Clark & Mills, 1979; G<strong>in</strong>sberg & Gottman, 1986).<br />
Developmental Significance<br />
Security <strong>and</strong> self-validation <strong>in</strong> social relations have long been<br />
regarded as necessary to <strong>the</strong> growth of social competence<br />
(Sroufe & Fleeson, 1986). To show that friends <strong>in</strong>teract commu-<br />
nally with one ano<strong>the</strong>r, however, does not demonstrate <strong>the</strong>ir<br />
developmental significance. Ra<strong>the</strong>r, two o<strong>the</strong>r k<strong>in</strong>ds of studies<br />
are needed to do this: (a) comparisons between <strong>in</strong>dividuals who<br />
have friends with those who do not, especially with respect to<br />
self-esteem <strong>and</strong> social competence, <strong>and</strong> (b) demonstrations that<br />
hav<strong>in</strong>g friends supports better cop<strong>in</strong>g with challenges or stress<br />
than not hav<strong>in</strong>g friends <strong>and</strong> promotes good outcome over time.<br />
Hav<strong>in</strong>g friends versus not hav<strong>in</strong>g friends. Cross-sectional<br />
(concurrent) comparisons show that, first, children who have<br />
friends are more socially competent than those who do not; <strong>the</strong>y<br />
are more sociable, cooperative, altruistic, self-confident, <strong>and</strong> less<br />
lonely (A. E Newcomb & BagweU, 1995). Second, children,<br />
adolescents, <strong>and</strong> adults seek<strong>in</strong>g cl<strong>in</strong>ical referrals or o<strong>the</strong>r forms<br />
of assistance with psychosocial problems are more likely to be<br />
friendless than better adjusted <strong>in</strong>dividuals (Rutter & Garmezy,<br />
1983). Third, <strong>in</strong>dividuals with friends enjoy greater psychologi-<br />
cal well-be<strong>in</strong>g throughout adulthood <strong>and</strong> old age than <strong>in</strong>dividu-<br />
als who do not have friends (Brown, 1981; Gupta & Korte,<br />
1994; Larson, 1978).<br />
Certa<strong>in</strong> studies suggest, however, that <strong>the</strong> existence of sup-<br />
portive relationships may have less to do with well-be<strong>in</strong>g, espe-<br />
cially <strong>in</strong> old age, than <strong>the</strong> absence of problematic ones. Indeed,<br />
<strong>the</strong> number of problematic relationships one has <strong>and</strong> <strong>the</strong> fre-<br />
quency with which one <strong>in</strong>teracts with problematic persons are<br />
more closely related (negatively) to well-be<strong>in</strong>g than <strong>the</strong> number<br />
of supportive persons available <strong>and</strong> opportunities to <strong>in</strong>teract<br />
with <strong>the</strong>m (Pagel, Erdly, & Becker, 1987; Rook, 1984; Stephens,<br />
K<strong>in</strong>ney, Ritchie, & Norris, 1987). Still o<strong>the</strong>r studies suggest<br />
that friendship support may be more necessary for some <strong>in</strong>divid-<br />
uals than o<strong>the</strong>rs. Lang <strong>and</strong> Carstensen (1994) found, for exam-<br />
ple, that among older <strong>in</strong>dividuals who no longer have liv<strong>in</strong>g<br />
relatives, number of close friends predicts feel<strong>in</strong>gs of social<br />
embeddedness, whereas this is not <strong>the</strong> case among <strong>in</strong>dividuals<br />
with families. Friends also provide emotional advantages for<br />
"spouseless" <strong>in</strong>dividuals, that is, persons who were never or<br />
were formerly married (Dykstra, 1995a; Gupta & Korte, 1994).<br />
Among older persons, widows <strong>and</strong> widowers spend more time<br />
with friends than married people (Gallagher & Gerstel, 1993;<br />
Wister, 1990) or those divorced (R. S. Weiss, 1975), with<br />
friends serv<strong>in</strong>g both confid<strong>in</strong>g <strong>and</strong> companionship functions<br />
(Connidis & Davies, 1990). Comparisons between married <strong>and</strong><br />
spouseless <strong>in</strong>dividuals are especially important because <strong>the</strong> re-<br />
sults support two widely discussed but <strong>in</strong>frequently verified<br />
hypo<strong>the</strong>ses: ( 1 ) <strong>Friendships</strong> may substitute for o<strong>the</strong>r (miss<strong>in</strong>g)<br />
relationships <strong>in</strong> social development <strong>and</strong> adaptation <strong>and</strong> (2)<br />
friendships are protective factors or "buffers" that mitigate
360 HARTUP AND STEVENS<br />
<strong>the</strong> effects of stress <strong>and</strong> privation <strong>in</strong> everyday life, especially<br />
relationship stra<strong>in</strong>s <strong>and</strong> losses.<br />
In general, <strong>the</strong>n, hav<strong>in</strong>g friends is correlated with psychologi-<br />
cal well-be<strong>in</strong>g from childhood through old age, consistent with<br />
<strong>the</strong> hypo<strong>the</strong>sis that <strong>the</strong> friendship deep structure (reciprocity)<br />
<strong>and</strong> certa<strong>in</strong> benefits associated with it do not change through<br />
<strong>the</strong> life course. These results are never<strong>the</strong>less difficult to <strong>in</strong>ter-<br />
pret. First, hav<strong>in</strong>g friends <strong>in</strong> <strong>the</strong>se studies is usually confounded<br />
with friendship quality; that is, hav<strong>in</strong>g friends usually means<br />
hav<strong>in</strong>g good friends. The significance of this confound is clearly<br />
demonstrated by <strong>the</strong> discovery that not hav<strong>in</strong>g problematic<br />
friendships is more closely related to well-be<strong>in</strong>g among older<br />
persons than hav<strong>in</strong>g supportive ones (Rook, 1984). Second,<br />
most empirical studies (especially with adolescents <strong>and</strong> adults)<br />
exam<strong>in</strong>e support networks that <strong>in</strong>clude friends but also <strong>in</strong>clude<br />
o<strong>the</strong>r persons, thus mak<strong>in</strong>g it difficult to specify <strong>the</strong> exact devel-<br />
opmental contributions made by <strong>the</strong> dyadic relationship. Third,<br />
causal direction is impossible to establish: Friendship experi-<br />
ence may contribute to self-esteem or well-be<strong>in</strong>g; but, at <strong>the</strong><br />
same time, confident <strong>and</strong> secure <strong>in</strong>dividuals may make friends<br />
more readily than less confident ones--both <strong>in</strong> childhood <strong>and</strong><br />
adulthood (see Elicker et al., 1992). Fourth, <strong>the</strong> correlation<br />
between hav<strong>in</strong>g friends <strong>and</strong> well-be<strong>in</strong>g is moderated by o<strong>the</strong>r<br />
conditions, <strong>in</strong>clud<strong>in</strong>g <strong>the</strong> availability of o<strong>the</strong>r supportive rela-<br />
tionships (e.g., with spouses or relatives), time spent with<br />
friends, <strong>and</strong> <strong>the</strong> k<strong>in</strong>ds of events that require support (Staud<strong>in</strong>ger,<br />
Marsiske, & Baltes, 1995). Although <strong>the</strong> correlational evidence<br />
suggests that friendships constitute social <strong>and</strong> emotional re-<br />
sources across <strong>the</strong> life course, important reservations must be<br />
attached to this conclusion.<br />
Outcomes across developmental transitions. Relatively few<br />
longitud<strong>in</strong>al studies conta<strong>in</strong> <strong>in</strong>formation about hav<strong>in</strong>g friends<br />
<strong>and</strong> developmental outcome. Short-term studies with children<br />
suggest that benefits accrue across certa<strong>in</strong> developmental transi-<br />
tions, for example, school entrance. School attitudes are better<br />
among k<strong>in</strong>dergarten children (5-year-olds) whose prior friends<br />
attend <strong>the</strong> same school <strong>and</strong> who ma<strong>in</strong>ta<strong>in</strong> <strong>the</strong>se relationships<br />
than among children who do not (Ladd, 1990). Hav<strong>in</strong>g friends<br />
also predicts <strong>in</strong>creases <strong>in</strong> self-esteem among preadolescents<br />
(Bukowski, Hoza, & Newcomb, 1991 ), <strong>and</strong> psychosocial distur-<br />
bances are less frequent when school changes occur <strong>in</strong> <strong>the</strong> com-<br />
pany of good friends than when <strong>the</strong>y do not (Berndt & Keefe,<br />
1992; Simmons, Burgeson, & Reef, 1988).<br />
O<strong>the</strong>r studies suggest that friendships <strong>in</strong> childhood may be<br />
precursors of romantic relationships <strong>in</strong> adolescence. First, ' 'hav-<br />
<strong>in</strong>g friends" <strong>and</strong> "hav<strong>in</strong>g a friend to confide <strong>in</strong>" dur<strong>in</strong>g middle<br />
childhood are reported more often by undergraduates also re-<br />
port<strong>in</strong>g childhood sexual encounters (with o<strong>the</strong>r children) than<br />
undergraduates not report<strong>in</strong>g early sexual experience (Hau-<br />
gaard & Tilly, 1988). This concordance between hav<strong>in</strong>g friends<br />
<strong>and</strong> early sexual experience may reflect <strong>in</strong>dividual differences <strong>in</strong><br />
self-esteem or social competence ra<strong>the</strong>r than a causal connection<br />
between friendship experience <strong>and</strong> sexual socialization. Alterna-<br />
tively, hav<strong>in</strong>g friends may <strong>in</strong>crease feel<strong>in</strong>gs of self-worth (see<br />
below), <strong>the</strong>reby sett<strong>in</strong>g <strong>the</strong> stage for early sexual engagement.<br />
Cross-lagged longitud<strong>in</strong>al data (Neeman, Hubbard, & Kojet<strong>in</strong>,<br />
1991) clarify this situation somewhat: Hav<strong>in</strong>g same-gender<br />
friends dur<strong>in</strong>g middle childhood forecasts hav<strong>in</strong>g romantic rela-<br />
tionships dur<strong>in</strong>g adolescence; subsequently, hav<strong>in</strong>g same-gender<br />
friends dur<strong>in</strong>g adolescence forecasts romantic relationships <strong>in</strong><br />
early adulthood. Because same-gender friendships forecast ro-<br />
mantic relationships but not vice versa, results are consistent<br />
with Sullivan's (1953) notion that same-gender friendships dur-<br />
<strong>in</strong>g <strong>the</strong> "juvenile era" support <strong>the</strong> formation <strong>and</strong> function<strong>in</strong>g<br />
of heterosexual relationships--ma<strong>in</strong>ly by <strong>the</strong> establishment of<br />
<strong>in</strong>timacy needs.<br />
F<strong>in</strong>ally, <strong>in</strong> one o<strong>the</strong>r longitud<strong>in</strong>al <strong>in</strong>vestigation, Bagwell, New-<br />
comb, <strong>and</strong> Bukowski (1996) exam<strong>in</strong>ed hav<strong>in</strong>g friends dur<strong>in</strong>g<br />
preadolescence broadly as a predictor of social adaptation <strong>in</strong><br />
early adulthood ( 12 years after <strong>the</strong> <strong>in</strong>itial test<strong>in</strong>g). Results show<br />
that (a) childhood sociometric status (popular vs. rejected) sig-<br />
nificantly predicted school performance, job success, aspira-<br />
tions, <strong>and</strong> sociability <strong>in</strong> early adulthood, although hav<strong>in</strong>g friends<br />
did not; <strong>and</strong> (b) childhood friendships predicted good attitudes<br />
toward family members <strong>and</strong>, most important, general feel<strong>in</strong>gs<br />
of self-worth <strong>and</strong> depressive symptoms <strong>in</strong> early adulthood; so-<br />
ciometric status, however, did not predict <strong>the</strong>se outcomes. Corre-<br />
lations between hav<strong>in</strong>g friends <strong>in</strong> childhood <strong>and</strong> one's self-<br />
attitudes <strong>in</strong> adulthood rema<strong>in</strong>ed significant, even when <strong>the</strong> parti-<br />
cipants' perceptions of <strong>the</strong>ir social competence as children were<br />
factored out.<br />
These results suggest, first, that sociometric status is a more<br />
important predictor of social skill than friendship experience<br />
across <strong>the</strong> transition between childhood <strong>and</strong> adulthood. Friend-<br />
ship experience, however, contributes to developmental outcome<br />
<strong>in</strong> two o<strong>the</strong>r doma<strong>in</strong>s--family relations <strong>and</strong> self-esteem (as<br />
well as depression, a disorder frequently accompanied by low<br />
self-esteem). The results thus augment <strong>the</strong> cross-sectional stud-<br />
ies (see above) to suggest that hav<strong>in</strong>g friends contributes devel-<br />
opmentally to <strong>the</strong> <strong>in</strong>dividual's sense of well-be<strong>in</strong>g. More im-<br />
portant, <strong>the</strong> results are consistent with <strong>the</strong> hypo<strong>the</strong>sis that friend-<br />
ships enhance cop<strong>in</strong>g with <strong>the</strong> specific developmental challenges<br />
that confront <strong>the</strong> <strong>in</strong>dividuals <strong>in</strong>volved--identity issues, <strong>in</strong> this<br />
case. With <strong>the</strong>se participants--studied between childhood <strong>and</strong><br />
early adulthood--<strong>the</strong> results support one of <strong>the</strong> most basic prop-<br />
ositions <strong>in</strong> S ullivan's (1953) <strong>the</strong>ory of <strong>in</strong>terpersonal relations,<br />
namely, that preadolescent friendships "provide opportunities<br />
for validation of self-worth <strong>and</strong> a unique context for exploration<br />
<strong>and</strong> development of personal strengths" (Bagwell et al., 1996,<br />
p. 22). Results also show <strong>the</strong>se effects to be relatively long<br />
last<strong>in</strong>g.<br />
O<strong>the</strong>r longitud<strong>in</strong>al studies are more circumscribed. Short-<br />
term prospective studies with adults reveal that friendship net-<br />
works are associated with better postdivorce adaptation among<br />
adult women than are family networks, presumably because<br />
friendship collectives are more diversified <strong>and</strong> less closely knit<br />
(Wilcox, 1981 ). Retrospective accounts among older persons<br />
show that <strong>in</strong>dividuals who report <strong>the</strong>mselves as hav<strong>in</strong>g been<br />
<strong>in</strong>creas<strong>in</strong>gly happy after Age 60 had greater numbers of friends<br />
earlier than those report<strong>in</strong>g a decreas<strong>in</strong>g happ<strong>in</strong>ess (Lebo,<br />
1953). Friends are actually more effective <strong>in</strong> enhanc<strong>in</strong>g self-<br />
esteem among older persons than family members (Felton &<br />
Berry, 1992). These results cont<strong>in</strong>ue to suggest that friendships<br />
promote a sense of well-be<strong>in</strong>g when <strong>the</strong> relationship reciproci-<br />
ties are centered <strong>in</strong> developmentally relevant surface structure.<br />
To summarize, correlational studies show that hav<strong>in</strong>g friends<br />
is correlated with good psychosocial function<strong>in</strong>g <strong>and</strong> a sense of<br />
well-be<strong>in</strong>g across <strong>the</strong> life course. These studies are not easy to<br />
<strong>in</strong>terpret for several reasons: (a) The impact of hav<strong>in</strong>g friends<br />
is difficult to disentangle from that of friendship quality, (b)
hav<strong>in</strong>g a best friend is often difficult to differentiate from be<strong>in</strong>g<br />
part of a friendship network, <strong>and</strong> (c) directional effects are<br />
difficult to specify. Short-term longitud<strong>in</strong>al studies suggest that<br />
hav<strong>in</strong>g friends supports good outcomes across developmental<br />
transitions--both normative (e.g., school entrance) <strong>and</strong> non-<br />
normative (illness, divorce, or family member's death). More<br />
substantial evidence on <strong>the</strong>se effects is needed. Cross-lagged<br />
longitud<strong>in</strong>al studies are rare, especially ones that span substan-<br />
tial lengths of time. Some results suggest, however, that <strong>the</strong><br />
specific contributions made by hav<strong>in</strong>g friends reflect <strong>the</strong> devel-<br />
opmental status of <strong>the</strong> <strong>in</strong>dividuals <strong>in</strong>volved: Childhood friend-<br />
ships, for example, are now known to contribute to self-esteem<br />
as well as better family attitudes <strong>and</strong> romantic relationships<br />
dur<strong>in</strong>g adolescence <strong>and</strong> early adulthood. Whe<strong>the</strong>r similar devel-<br />
opmental sequelae are associated with hav<strong>in</strong>g friends at o<strong>the</strong>r<br />
times is not likely.<br />
The Identity of One's Friends<br />
Several questions need to be asked about <strong>the</strong> identity of one's<br />
friends: With whom does one become friends? Can <strong>the</strong> identity<br />
of one's friends be forecast from what one knows about oneself?<br />
What is <strong>the</strong> adaptational significance of <strong>the</strong> identity of one's<br />
friends? These questions are especially important because of<br />
<strong>the</strong> grow<strong>in</strong>g awareness that <strong>the</strong> behavioral similarities exist<strong>in</strong>g<br />
between friends (known among sociologists as "homophilies" )<br />
have considerable long-term significance for <strong>the</strong> <strong>in</strong>dividuals <strong>in</strong>-<br />
volved (Cairns & Cairns, 1994).<br />
Who Are One's Friends?<br />
Consider, first, that common ground is necessary for <strong>the</strong> for-<br />
mation <strong>and</strong> ma<strong>in</strong>tenance of friendships throughout <strong>the</strong> life<br />
course. Consequently, friends ought to be similar to one ano<strong>the</strong>r<br />
<strong>in</strong> age, gender, <strong>and</strong> ethnicity as well as abilities <strong>and</strong> behavior.<br />
Indeed, <strong>the</strong> weight of <strong>the</strong> evidence shows friends to be concor-<br />
dant for age, gender, socioeconomic status, <strong>and</strong> ethnicity--be-<br />
g<strong>in</strong>n<strong>in</strong>g <strong>in</strong> childhood <strong>and</strong> extend<strong>in</strong>g through old age (Adams &<br />
Blieszner, 1995; Kupersmidt, DeRosier, & Patterson, 1995; Mat-<br />
<strong>the</strong>ws, 1995). Demographic concordances are especially great<br />
<strong>in</strong> midlife because friends frequently are coworkers or family<br />
related (Nahemow & Lawton, 1975).<br />
Behavioral concordances are extensive too, but <strong>the</strong>se vary<br />
from age to age <strong>and</strong> attribute to attribute--<strong>in</strong> most cases based<br />
on normative salience (i.e., <strong>the</strong> normative salience of <strong>the</strong> attri-<br />
bute <strong>in</strong> <strong>the</strong> social networks to which <strong>the</strong> <strong>in</strong>dividual belongs or<br />
<strong>the</strong> salience of <strong>the</strong> attribute to determ<strong>in</strong>e social reputations).<br />
Among young children, Challman (1932) found friends to be<br />
concordant <strong>in</strong> social cooperation (an attribute with considerable<br />
normative significance) but not <strong>in</strong> <strong>in</strong>telligence (an attribute with<br />
none). Among boys, friends were more concordant <strong>in</strong> physical<br />
activity than nonfriends but not among girls, for whom similarity<br />
<strong>in</strong> attractiveness of personality <strong>and</strong> social network size was<br />
greater among friends than nonfriends.<br />
Most concordance studies with school-aged children are<br />
about personal construct use (Erw<strong>in</strong>, 1985), self-reported simi-<br />
larities (Hymel & Woody, 1991), or similarities with<strong>in</strong> social<br />
networks (Cairns & Cairns, 1994). Some data (Haselager,<br />
Hartup, Van Lieshout, & Riksen-Walraven, 1996) suggest, how-<br />
ever, that behavioral concordances with<strong>in</strong> <strong>the</strong> friendship dyad<br />
FRIENDSHIPS IN THE LIFE COURSE 361<br />
are considerably stronger <strong>in</strong> childhood than previously thought.<br />
Peer rat<strong>in</strong>gs were obta<strong>in</strong>ed from a large number of 11-year-olds,<br />
center<strong>in</strong>g on prosocial behavior, antisocial behavior, shyness-<br />
dependency, depressive symptoms, <strong>and</strong> sociometric status. First,<br />
friends were more similar to one ano<strong>the</strong>r than nonfriends with<strong>in</strong><br />
each construct cluster (i.e., both mean difference scores were<br />
significantly smaller, <strong>and</strong> correlations with<strong>in</strong> dyads were sig-<br />
nificantly greater). Second, friends were more similar to one<br />
ano<strong>the</strong>r than nonfriends <strong>in</strong> rat<strong>in</strong>g <strong>the</strong>ir classmates as well as <strong>in</strong><br />
classmates' rat<strong>in</strong>gs of <strong>the</strong>m. Third, correlations between friends<br />
were greater for antisocial behavior (i.e., fight<strong>in</strong>g, disruption,<br />
<strong>and</strong> bully<strong>in</strong>g) than prosocial behavior (i.e., cooperation <strong>and</strong><br />
offer<strong>in</strong>g help to o<strong>the</strong>rs) or social withdrawal (i.e., shyness, de-<br />
pendency, <strong>and</strong> be<strong>in</strong>g "victimized" ). These correlational differ-<br />
ences may reflect differences among <strong>the</strong> three attributes <strong>in</strong> nor-<br />
mative salience: Fight<strong>in</strong>g, for example, is more salient among<br />
children <strong>in</strong> terms of <strong>the</strong>ir reputations <strong>and</strong> <strong>the</strong>ir membership <strong>in</strong><br />
<strong>the</strong> social networks that exist with<strong>in</strong> classrooms than is ei<strong>the</strong>r<br />
cooperation or shyness (Coie, Dodge, & Kupersmidt, 1990).<br />
Fourth, certa<strong>in</strong> gender differences were evident: (a) Friends were<br />
more similar to one ano<strong>the</strong>r <strong>in</strong> both prosocial <strong>and</strong> antisocial<br />
behavior among girls than boys, <strong>and</strong> (b) Friends were more<br />
similar <strong>in</strong> shyness among boys than girls. These gender differ-<br />
ences are consistent with <strong>the</strong> normative salience hypo<strong>the</strong>sis too<br />
because uncooperativeness <strong>and</strong> aggression are more socially<br />
problematic for girls than boys (Huston, 1983). One would<br />
expect <strong>the</strong> reverse with shyness, however, because this attribute<br />
presents more serious social difficulties for boys than girls<br />
(Caspi, Elder, &Bem, 1988).<br />
Behavioral concordances among adolescents also vary from<br />
attribute to attribute. Adolescents are most similar to <strong>the</strong>ir<br />
friends <strong>in</strong> school-related attitudes, aspirations, <strong>and</strong> achievement<br />
(Epste<strong>in</strong>, 1983; K<strong>and</strong>el, 1978b) along with normative behaviors<br />
that def<strong>in</strong>e <strong>the</strong>ir lifestyles, for example, smok<strong>in</strong>g, dr<strong>in</strong>k<strong>in</strong>g, drug<br />
use, dat<strong>in</strong>g, <strong>and</strong> sexual activity (Billy, Rodgers, & Udry, 1984;<br />
K<strong>and</strong>el, 1978b; Tolson & Urberg, 1993). In general, personality<br />
assessments are not very similar among adolescent friends, al-<br />
though similarities <strong>in</strong> normatively significant behaviors are sub-<br />
stantial, for example, aggression <strong>and</strong> del<strong>in</strong>quency (K<strong>and</strong>el,<br />
1978a, 1978b; Dishion et al., 1995).<br />
Although similarities <strong>in</strong> attitudes <strong>and</strong> values are known to be<br />
<strong>the</strong> basis of attraction among adults (Berscheid & Walster, 1968;<br />
T. M. Newcomb, 1961 ), attempts to verify behavioral <strong>and</strong> attitu-<br />
d<strong>in</strong>al similarities among adult friends have yielded weak out-<br />
comes <strong>and</strong> <strong>in</strong>consistent results (Brown, 1981; Nahemow &<br />
Lawton, 1975). In describ<strong>in</strong>g "real" friends, adults emphasize<br />
similarities, such as shared experiences <strong>and</strong> activities, ease of<br />
communication, similar general behaviors <strong>and</strong> <strong>in</strong>terests--more<br />
so <strong>in</strong> fact than reciprocities (L. Weiss & Lowenthal, 1975). It<br />
is also <strong>the</strong> case, however, that common ground <strong>and</strong> shared <strong>in</strong>ter-<br />
ests are not mentioned as frequently as a basis for friendships<br />
by adults than children (Lowenthal et al., 1975; Werner & Par-<br />
melee, 1979). Whe<strong>the</strong>r adult friendships depend more on work<br />
<strong>and</strong> family communaiities than earlier friendships--<strong>the</strong>reby re-<br />
duc<strong>in</strong>g behavioral <strong>and</strong> attitud<strong>in</strong>al concordances between friends<br />
<strong>in</strong> adulthood--is unknown. The bulk of <strong>the</strong> evidence on behav-<br />
ioral similarity among adult friends emanates from studies of<br />
college roommates ra<strong>the</strong>r than older adult friends, perhaps to<br />
expla<strong>in</strong> why actual behavioral similarities are not consistently<br />
reported.
362 HARTUP AND STEVENS<br />
Similarities Between Friends: Sources<br />
The developmental significance of similarities between<br />
friends cannot be specified without <strong>the</strong> acknowledgment that<br />
friendship similarities (at least among children <strong>and</strong> adolescents)<br />
stem from three sources: (a) sociodemographic conditions that<br />
br<strong>in</strong>g similar <strong>in</strong>dividuals <strong>in</strong>to contact with one ano<strong>the</strong>r, (b) so-<br />
cial selection through which <strong>in</strong>dividuals select friends who are<br />
similar to <strong>the</strong>mselves <strong>and</strong> simultaneously avoid relationships<br />
with <strong>in</strong>dividuals who are different, <strong>and</strong> (c) mutual socialization<br />
through which <strong>in</strong>dividuals become similar to <strong>the</strong>ir friends by<br />
<strong>in</strong>teract<strong>in</strong>g with <strong>the</strong>m.<br />
Sociodemographic conditions. Sociodemographic forces de-<br />
term<strong>in</strong>e <strong>the</strong> neighborhood <strong>in</strong> which people live, <strong>the</strong> schools <strong>in</strong><br />
which people are enrolled as children <strong>and</strong> adolescents, <strong>the</strong> <strong>in</strong>sti-<br />
tutions people work for, <strong>and</strong> <strong>the</strong> neighborhoods people live <strong>in</strong><br />
when <strong>the</strong>y retire <strong>and</strong> become widowed. Concordances among<br />
persons <strong>and</strong> <strong>the</strong>ir friends <strong>in</strong> socioeconomic status, ethnicity, <strong>and</strong><br />
chronological age thus derive to some extent from social forces<br />
that br<strong>in</strong>g similar <strong>in</strong>dividuals <strong>in</strong>to contact with one ano<strong>the</strong>r <strong>in</strong><br />
neighborhoods <strong>and</strong> <strong>in</strong>stitutions. Culture restricts social choices<br />
to similar <strong>in</strong>dividuals <strong>in</strong> two ways: First, it creates <strong>in</strong>dividuals<br />
who have similar beliefs <strong>and</strong> attitudes, <strong>in</strong>clud<strong>in</strong>g <strong>the</strong> belief that<br />
<strong>in</strong>terpersonal similarity optimizes gratification as well as <strong>the</strong><br />
belief that dissimilarity optimizes tension <strong>and</strong> discomfort (Ro-<br />
senbaum, 1986); second, it creates neighborhoods <strong>and</strong> commu-<br />
nity arrangements that maximize opportunities for similar <strong>in</strong>di-<br />
viduals to socialize with one ano<strong>the</strong>r <strong>and</strong> m<strong>in</strong>imize contacts<br />
between dissimilar <strong>in</strong>dividuals. Residential choices are <strong>the</strong> <strong>in</strong>-<br />
strumentalities through which culture br<strong>in</strong>gs about <strong>the</strong>se concor-<br />
dances; <strong>the</strong> <strong>in</strong>stitutions with which <strong>in</strong>dividuals affiliate (work<br />
organizations, schools, <strong>and</strong> churches) extend <strong>the</strong>m.<br />
Selection. Similarities between friends also derive from se-<br />
lection choice, that is, <strong>the</strong> tendency among <strong>in</strong>dividuals to choose<br />
associates who resemble <strong>the</strong>mselves. Among children attend<strong>in</strong>g<br />
schools that are of mixed ages, races, <strong>and</strong> socioeconomic sta-<br />
tuses, for example, friends are more similar to one ano<strong>the</strong>r than<br />
nonfriends (Goldman, 1981; McC<strong>and</strong>less & Hoyt, 1961 ). Simi-<br />
larly, older residents of nurs<strong>in</strong>g homes are more likely to make<br />
friends with similar than dissimilar <strong>in</strong>dividuals (Chown, 1981;<br />
Mat<strong>the</strong>ws, 1995).<br />
The similarity-attraction dynamic beg<strong>in</strong>s <strong>in</strong> childhood. For<br />
example, among <strong>the</strong> 8-year-old children <strong>in</strong> one experiment who<br />
began a series of sessions as strangers, differential attraction<br />
was evident <strong>in</strong> some groups (40%) after <strong>the</strong> first meet<strong>in</strong>g. With<strong>in</strong><br />
<strong>the</strong>se groups, <strong>the</strong> social <strong>and</strong> cognitive dimensions of play were<br />
more similar when <strong>the</strong> children were attracted to one ano<strong>the</strong>r<br />
than when <strong>the</strong>y were not (Rub<strong>in</strong>, Lynch, Coplan, Rose-<br />
Krasnor, & Booth, 1994). O<strong>the</strong>r studies show that children be-<br />
come friends <strong>in</strong> direct relation to <strong>the</strong> number of attributes (both<br />
demographic <strong>and</strong> behavioral) <strong>the</strong>y share (Kupersmidt et al.,<br />
1995).<br />
Re<strong>in</strong>forcement <strong>the</strong>orists argued that <strong>the</strong>se homophilies stem<br />
from two sources: <strong>the</strong> rewards that emanate from recognition<br />
by <strong>in</strong>dividuals of common statuses <strong>and</strong> values (Lazarsfeld &<br />
Merton, 1954) <strong>and</strong> <strong>the</strong> aversiveness that stems from recognition<br />
of status dissimilarities (Rosenbaum, 1986). Rewards are both<br />
<strong>in</strong>tr<strong>in</strong>sic <strong>and</strong> extr<strong>in</strong>sic: For example, most young girls f<strong>in</strong>d <strong>the</strong>ir<br />
<strong>in</strong>terests <strong>in</strong> dramatic play to be more supported by <strong>in</strong>teraction<br />
with o<strong>the</strong>r girls than with boys (Sears, Alpert, & Rau, 1964),<br />
whereas girls receive more approval from both adults <strong>and</strong> age-<br />
mates for associat<strong>in</strong>g with o<strong>the</strong>r girls than boys (Fagot, 1978;<br />
Thorne, 1986). Selection similarities based on avoidance also<br />
derive from both <strong>in</strong>tr<strong>in</strong>sic <strong>and</strong> extr<strong>in</strong>sic sources: Young girls,<br />
for example, f<strong>in</strong>d <strong>the</strong> rough <strong>and</strong> tumble play of boys to be<br />
aversive <strong>and</strong> a reason for <strong>the</strong>ir perference for <strong>the</strong> company of<br />
o<strong>the</strong>r young girls (Maccoby, 1990), whereas cross-gender so-<br />
cializ<strong>in</strong>g among nursery school children is criticized by o<strong>the</strong>r<br />
children of both genders (Fagot, 1978). These dynamics un-<br />
doubtedly contribute to <strong>the</strong> similarities that exist between<br />
friends throughout <strong>the</strong> life course (Hess, 1972).<br />
Based on <strong>the</strong> available evidence, we cannot assume that simi-<br />
larities between friends derive from carefully weighed decisions<br />
made by <strong>in</strong>dividuals, cumulat<strong>in</strong>g gradually over weeks <strong>and</strong><br />
months, to associate with o<strong>the</strong>rs who are similar <strong>and</strong> avoid those<br />
who are dissimilar to <strong>the</strong>mselves. Ra<strong>the</strong>r, network <strong>and</strong> friendship<br />
concordances resemble "shopp<strong>in</strong>g expeditions" (Dishion, Pat-<br />
terson, & Griesler, 1994) <strong>in</strong> which certa<strong>in</strong> selections "feel<br />
right" <strong>and</strong> o<strong>the</strong>rs do not (i.e., some activities <strong>and</strong> conversations<br />
suggest common ground, whereas o<strong>the</strong>rs do not). Antisocial<br />
children, for example, are more likely to make friends with o<strong>the</strong>r<br />
antisocial children than better socialized children, both because<br />
o<strong>the</strong>r children do not select <strong>the</strong>m <strong>and</strong> because, between <strong>the</strong>m-<br />
selves, <strong>the</strong>ir antisocial behaviors establish common ground. In-<br />
deed, <strong>the</strong> communalities that exist between antisocial adoles-<br />
cents <strong>and</strong> <strong>the</strong>ir friends suggest a k<strong>in</strong>d of merger between <strong>the</strong>m<br />
that results <strong>in</strong> <strong>the</strong> emergence of a "dyadic antisocial trait."<br />
Similar mergers occur among children for friendl<strong>in</strong>ess <strong>and</strong> coop-<br />
eration as well as shyness <strong>and</strong> depressive symptoms (Haselager<br />
et al., 1996).<br />
Selection choices are also embedded <strong>in</strong> complex assortative<br />
processes, about which relatively little is known. Assortments<br />
occur with<strong>in</strong> social networks, which <strong>in</strong> turn emerge from larger<br />
social units, such as classrooms, office blocks, <strong>and</strong> retirement<br />
residences. Greater similarity between friends than nonfriends<br />
thus emerges with<strong>in</strong> two <strong>in</strong>terlock<strong>in</strong>g selection systems: (a)<br />
dyadic <strong>in</strong>teraction <strong>and</strong> (b) assortative dialectics, which differen-<br />
tiate social networks from <strong>the</strong>ir larger aggregates. Little recogni-<br />
tion is given to, unfortunately, that friendship selection usually<br />
occurs with<strong>in</strong> higher order structures <strong>in</strong> which group decision<br />
mak<strong>in</strong>g <strong>and</strong> pressures also occur. Consequently, <strong>the</strong>oretical mod-<br />
els are lack<strong>in</strong>g to account for <strong>the</strong> manner <strong>in</strong> which network<br />
dynamics moderate friendship selection (Berndt, 1996). Given<br />
this state of affairs, researchers should not treat similarity be-<br />
tween best friends <strong>and</strong> with<strong>in</strong> friendship networks as reflections<br />
of <strong>the</strong> same construct.<br />
Mutual socialization. Similarities between friends derive<br />
from socialization as well as selection. K<strong>and</strong>el (1978a) studied<br />
changes over <strong>the</strong> course of 1 year <strong>in</strong> substance use, educational<br />
aspirations, <strong>and</strong> del<strong>in</strong>quency of adolescents, discover<strong>in</strong>g that<br />
similarity stems from both sources <strong>in</strong> approximately equal<br />
amounts. In o<strong>the</strong>r studies, selection seems to contribute more<br />
variance than socialization to cigarette <strong>and</strong> alcohol use (L. A.<br />
Fisher & Baumann, 1988). The sturdiest conclusion that can<br />
be drawn from <strong>the</strong>se studies is that <strong>the</strong> relative contribution<br />
of selection <strong>and</strong> socialization to friendship similarity must be<br />
estimated separately for each attribute <strong>and</strong>, most likely, each<br />
population assessed.<br />
Circumstantial evidence that demonstrates mutual socializa-<br />
tion effects is plentiful: Children, for example, who ascribe to
conventional norms move fur<strong>the</strong>r over time <strong>in</strong> <strong>the</strong> direction of<br />
normative behavior after <strong>the</strong>y associate with friends (Ball, 1981;<br />
Epste<strong>in</strong>, 1983; K<strong>and</strong>el & Andrews, 1987). Antisocial activities<br />
<strong>in</strong>crease over time among antisocial <strong>in</strong>dividuals <strong>and</strong> <strong>the</strong>ir friends,<br />
even though most people perceive friends as exert<strong>in</strong>g pressure<br />
toward desirable ra<strong>the</strong>r than undesirable conduct (Brown, Cla-<br />
sen, & Eicher, 1986). Similarly, socializ<strong>in</strong>g with friends smooths<br />
adaptation to new <strong>in</strong>stitutional sett<strong>in</strong>gs among both adolescents<br />
<strong>and</strong> older persons (Armstrong & Goldsteen, 1990; Simmons et<br />
al., 1988; Stacey-Konnert & Pynoss, 1992).<br />
Debate cont<strong>in</strong>ues whe<strong>the</strong>r friends truly socialize one ano<strong>the</strong>r<br />
or ra<strong>the</strong>r serve as "socialization supplements." Beth Hess<br />
(1972) argued that friends assist with learn<strong>in</strong>g, behavioral im-<br />
plementation, or both but cannot teach one ano<strong>the</strong>r because<br />
<strong>the</strong> difference between <strong>the</strong>m <strong>in</strong> expertise is not great enough.<br />
Although much evidence demonstrates that <strong>the</strong> co-construction<br />
of knowledge is more successful than is a solitary effort <strong>in</strong> many<br />
different situations (Rogoff, 1990), contemporary research does<br />
not tell def<strong>in</strong>itively whe<strong>the</strong>r two similar <strong>in</strong>dividuals can teach<br />
one ano<strong>the</strong>r effectively <strong>in</strong> all situations.<br />
One guesses that, from early childhood through old age, <strong>in</strong>di-<br />
viduals model normative behavior for <strong>the</strong>ir friends <strong>and</strong> simulta-<br />
neously receive re<strong>in</strong>forcement from <strong>the</strong>m. Once aga<strong>in</strong>, <strong>the</strong> exact<br />
mechanisms through which mutual socialization occurs between<br />
friends are not well documented. Friends engage <strong>in</strong> large<br />
amounts of talk, <strong>and</strong> conversations with friends dur<strong>in</strong>g problem<br />
solv<strong>in</strong>g are laced with both suggestions <strong>and</strong> criticisms (Nel-<br />
son & Aboud, 1985). Conflicts frequently occur <strong>and</strong> may be<br />
closely related to <strong>the</strong> socialization that occurs between friends.<br />
In one <strong>in</strong>vestigation with school-aged children (Azmitia &<br />
Montgomery, 1993), difficult deductive reason<strong>in</strong>g tasks were<br />
more frequently solved by friends work<strong>in</strong>g toge<strong>the</strong>r than non-<br />
friends. Although social <strong>in</strong>teraction differed between friends <strong>and</strong><br />
nonfriends <strong>in</strong> numerous ways, task success was significantly<br />
related to only one measure--transactive conflicts. Socializa-<br />
tion between friends thus may rest ma<strong>in</strong>ly on <strong>the</strong> free air<strong>in</strong>g of<br />
disagreements <strong>in</strong> a cooperative, task-oriented context ra<strong>the</strong>r than<br />
on model<strong>in</strong>g or re<strong>in</strong>forcement. O<strong>the</strong>r results suggest that friends<br />
also use coercion with one ano<strong>the</strong>r, but it is different from<br />
<strong>the</strong> criticisms <strong>and</strong> persuasions that mark <strong>in</strong>teractions between<br />
nonfriends; reason<strong>in</strong>g <strong>and</strong> explanations are more common (Dis-<br />
hion et al., 1995; Nelson & Aboud, 1985).<br />
Developmental Implications<br />
Given <strong>the</strong> demographic <strong>and</strong> social similarities that exist be-<br />
tween friends <strong>and</strong> that <strong>the</strong>se similarities extend from childhood<br />
through old age, what is <strong>the</strong>ir significance? First, friends who<br />
are well socialized <strong>and</strong> normatively conventional serve mutually<br />
as "protective" factors <strong>in</strong> development. For example, friends<br />
who are cooperative <strong>and</strong> nonadversarial become more so over<br />
time. Second, friends who are antisocial <strong>and</strong> socially unskilled<br />
seem to be "risk" factors. The developmental significance of<br />
certa<strong>in</strong> o<strong>the</strong>r similarities, however, is not known. Consider, for<br />
example, that no one knows whe<strong>the</strong>r shy friends socialize <strong>the</strong>m-<br />
selves toward <strong>in</strong>creased gregariousness or <strong>in</strong>creased shyness.<br />
Whatever <strong>the</strong> situation, though, <strong>the</strong> social exchanges that occur<br />
between two shy friends may improve ra<strong>the</strong>r than worsen <strong>in</strong>di-<br />
vidual adaptation by alleviat<strong>in</strong>g <strong>the</strong> debilitat<strong>in</strong>g lonel<strong>in</strong>ess that<br />
FRIENDSHIPS IN THE LIFE COURSE 363<br />
accompanies shyness among both children <strong>and</strong> adults (Asher,<br />
Parkhurst, Hymel, & Williams, 1990).<br />
So friendships can be mixed bless<strong>in</strong>gs: On <strong>the</strong> one h<strong>and</strong>,<br />
friends may support good developmental outcome through <strong>the</strong><br />
<strong>in</strong>timacy, companionship, social support, <strong>and</strong> <strong>in</strong>creased well-<br />
be<strong>in</strong>g that <strong>the</strong>y provide one ano<strong>the</strong>r. On <strong>the</strong> o<strong>the</strong>r h<strong>and</strong>, friends<br />
may be risk factors, depend<strong>in</strong>g on who one's friends are. Poorly<br />
socialized <strong>and</strong> antisocial friends place one at greater risk for<br />
social maladaptation than do well-socialized friends. Prac-<br />
titioners may not be able to <strong>in</strong>sist that <strong>the</strong>ir clients change friends<br />
(for ethical reasons or o<strong>the</strong>rwise), but developmental prognosis<br />
never<strong>the</strong>less dem<strong>and</strong>s that <strong>the</strong>y know who a client's friends are,<br />
not just whe<strong>the</strong>r <strong>the</strong> <strong>in</strong>dividual does or does not have friends.<br />
Assessment Issues<br />
Friendship Quality<br />
Qualitative assessment is more advanced to study friendships<br />
among children <strong>and</strong> adolescents than those among adults. Two<br />
ma<strong>in</strong> strategies are used: (a) dimensional assessment through<br />
which one determ<strong>in</strong>es whe<strong>the</strong>r particular features characterize<br />
<strong>the</strong> social <strong>in</strong>teraction between friends <strong>and</strong> with what frequency<br />
or regularity (e.g., companionship, <strong>in</strong>timacy, conflict, or power<br />
symmetries) <strong>and</strong> (b) typological or categorical assessment<br />
through which one identifies patterns or organizations <strong>in</strong> <strong>the</strong><br />
social <strong>in</strong>teractions between friends that contribute to social de-<br />
velopment <strong>and</strong> adaptation (Furman, 1996).<br />
Dimensional assessment. Most dimensional assessments are<br />
based on what Robert S. Weiss (1986) called "provisions" or<br />
features of relationships, for example, <strong>the</strong>ir closeness, <strong>in</strong>timacy,<br />
supportiveness, or content. Studies with children <strong>and</strong> adolescents<br />
<strong>in</strong>dicate that considerable mileage can be ga<strong>in</strong>ed simply by <strong>the</strong><br />
measurement of relationships <strong>in</strong> terms of <strong>the</strong>ir <strong>in</strong>timacy or sup-<br />
portiveness (Berndt, 1996; Furman, 1996). Work with adoles-<br />
cents suggests <strong>the</strong> <strong>the</strong>oretical relevance of connectedness, a<br />
composite compris<strong>in</strong>g closeness, <strong>in</strong>terdependence, <strong>and</strong> emo-<br />
tional tone (Coll<strong>in</strong>s & Rep<strong>in</strong>ski, 1991 ). Among adults, friend-<br />
ship quality is described ma<strong>in</strong>ly <strong>in</strong> terms of closeness or solidar-<br />
ity, def<strong>in</strong>ed operationally as <strong>the</strong> frequency <strong>and</strong> diversity of <strong>in</strong>ter-<br />
action as well as affective shar<strong>in</strong>g (Berscheid, Snyder, & Omoto,<br />
1989; Rep<strong>in</strong>ski, 1993).<br />
Dimensional structures that encompass both positive <strong>and</strong> neg-<br />
ative friendship attributes have also been exam<strong>in</strong>ed. W<strong>in</strong>dle<br />
(1994), for example, confirmed a factor structure among older<br />
adolescents that <strong>in</strong>cludes reciprocity of relations, self-disclo-<br />
sure, overt hostility, <strong>and</strong> covert hostility. O<strong>the</strong>r <strong>in</strong>struments de-<br />
signed to measure friendship quality <strong>in</strong>clude slightly different<br />
dimensions (Bukowski, Hoza, & Boiv<strong>in</strong>, 1994; Furman & Buhr-<br />
mester, 1985; Parker & Asher, 1993; Young, 1986), most tapp<strong>in</strong>g<br />
five or six doma<strong>in</strong>s (e.g., companionship, <strong>in</strong>timacy, commit-<br />
ment, affective tone, <strong>in</strong>strumental help, <strong>and</strong> conflict). Factor<br />
analysis does not always confirm that <strong>the</strong>se measurements are<br />
as multidimensional as <strong>the</strong>ir orig<strong>in</strong>ators wanted <strong>the</strong>m to be,<br />
although many reveal a well-differentiated structure (Ladd, Ko-<br />
chenderfer, & Coleman, 1996; W<strong>in</strong>dle, 1994).<br />
Closeness <strong>and</strong> supportiveness are emphasized <strong>in</strong> dimensional<br />
assessment because <strong>the</strong>se attributes reflect <strong>the</strong> friendship deep<br />
structure-reciprocity. This convention makes sense, although<br />
one or two reservations must be voiced: First, closeness <strong>and</strong>
364 HARTUP AND STEVENS<br />
<strong>in</strong>timacy are commonly def<strong>in</strong>ed <strong>in</strong> a fem<strong>in</strong>ized manner that<br />
emphasizes exclusivity, self-disclosure, social underst<strong>and</strong><strong>in</strong>g,<br />
<strong>and</strong> care (communal aspects). Concomitantly, adolescent girls<br />
rate <strong>the</strong>ir friendships more highly than do boys <strong>in</strong> self-disclosure<br />
as well as less highly <strong>in</strong> overt hostility (W<strong>in</strong>dle, 1994). Greater<br />
weight ought to be given to sociable communication, social<br />
<strong>in</strong>volvement with multiple companions, objective praise <strong>and</strong><br />
criticism, <strong>and</strong> effective conflict management, so <strong>the</strong> agentic di-<br />
mensions of closeness are measured as well as communal ones.<br />
The objective, <strong>in</strong> this <strong>in</strong>stance, is not to m<strong>in</strong>imize gender differ-<br />
ences <strong>in</strong> friendship closeness but only to ensure that agentic<br />
elements are recognized as strongly to measure this attribute as<br />
communal ones are.<br />
Second, closeness (reciprocity) must be measured <strong>in</strong> ways<br />
that are developmentally appropriate. Closeness needs to be<br />
exam<strong>in</strong>ed among children as well as adults. Considerable psy-<br />
chometric work must be conducted before o<strong>the</strong>r attributes, such<br />
as communality-agency, affective tone, symmetry, <strong>and</strong> conflict<br />
management, can be measured effectively from early childhood<br />
through old age. One must recognize, too, that different friend-<br />
ship qualities may be related to social adaptation among younger<br />
as compared with older <strong>in</strong>dividuals, reflect<strong>in</strong>g developmental<br />
changes that occur <strong>in</strong> <strong>the</strong> friendship surface structure (see<br />
Friendship Relations <strong>in</strong> <strong>the</strong> <strong>Life</strong> <strong>Course</strong>). Among <strong>the</strong> urgent<br />
needs of researchers are comparative developmental studies that<br />
establish <strong>the</strong> qualitative dimensions most closely related to good<br />
developmental outcome across time.<br />
Typological assessment. Many years ago, John Bowlby<br />
(1969) argued that <strong>in</strong>fant-caregiver relationships are based on<br />
"felt security" <strong>and</strong> that <strong>the</strong> significance of <strong>the</strong>se relationships<br />
can be detected <strong>in</strong> <strong>the</strong> <strong>in</strong>fant's behavior on two occasions--<br />
when separated from <strong>the</strong> caregiver <strong>and</strong> when reunited with him<br />
or her. Mary A<strong>in</strong>sworth (A<strong>in</strong>sworth, Blehar, Waters, & Wall,<br />
1978) subsequently demonstrated that variations <strong>in</strong> <strong>the</strong> organi-<br />
zation of separation- <strong>and</strong> reunion-<strong>in</strong>duced behavior (secure, re-<br />
sistant, <strong>and</strong> anxious attachments) forecast adaptational out-<br />
comes <strong>in</strong> child <strong>and</strong> adolescent development. Several efforts dem-<br />
onstrate <strong>the</strong> "secure base phenomenon" among friends. Nursery<br />
school children show fewer signs of distress, more positive af-<br />
fect, greater mobility, <strong>and</strong> more frequent talk<strong>in</strong>g when left <strong>in</strong> a<br />
strange situation with a friend than with a nonfriend (Ipsa, 1981;<br />
Schwartz, 1972). Among older adults, absence of friendship<br />
support is closely tied to feel<strong>in</strong>gs of lonel<strong>in</strong>ess (Dykstra,<br />
1995a). Whe<strong>the</strong>r friendship styles can be identified that are<br />
similar to <strong>the</strong> attachment types identified <strong>in</strong> mo<strong>the</strong>r-<strong>in</strong>fant rela-<br />
tionships has not been established. Although an attachment ty-<br />
pology (secure, anxious, <strong>and</strong> ambivalent) has been identified<br />
among adult romantic relationships (Hazan & Shaver, 1987),<br />
results have been controversial (Berscheid, 1994). Consider<strong>in</strong>g<br />
that <strong>the</strong> friendship deep structure consists of symmetrical reci-<br />
procities ra<strong>the</strong>r than security (see Friendship Relations <strong>in</strong> <strong>the</strong><br />
<strong>Life</strong> <strong>Course</strong>), we believe that <strong>the</strong> developmental significance of<br />
<strong>the</strong>se relationships is more likely to rest on variations <strong>in</strong> <strong>in</strong>timacy<br />
<strong>and</strong> its concomitants than variations <strong>in</strong> security.<br />
Two typological models--both grounded <strong>in</strong> <strong>the</strong> friendship<br />
deep structure--have been described <strong>in</strong> <strong>the</strong> literature, although<br />
nei<strong>the</strong>r is well validated. First, Rawl<strong>in</strong>s (1992) argued that con- _<br />
versations between friends can be classified on <strong>the</strong> basis of four<br />
dialectical axes: dependence versus <strong>in</strong>dependence (connected-<br />
ness vs. <strong>in</strong>dividuation), <strong>in</strong>strumental versus affective engage-<br />
ment, criticism versus acceptance, <strong>and</strong> expressiveness versus<br />
protectiveness. Configurations or patterns del<strong>in</strong>eated by <strong>the</strong>se<br />
four dimensions are believed to differentiate one friendship pair<br />
from ano<strong>the</strong>r <strong>and</strong> to be potentially useful to study relationship<br />
stages, cultural differences, cont<strong>in</strong>uities, <strong>and</strong> changes across <strong>the</strong><br />
life course. Assessment researchers who based <strong>the</strong>ir work on<br />
this system use qualitative ra<strong>the</strong>r than quantitative methods, <strong>and</strong><br />
applications have been made mostly to conversational vignettes.<br />
Second, Shulman (1993) constructed a typological model<br />
based on <strong>the</strong> balance between closeness <strong>and</strong> <strong>in</strong>timacy--on <strong>the</strong><br />
one h<strong>and</strong>--<strong>and</strong> <strong>in</strong>dividuality--on <strong>the</strong> o<strong>the</strong>r. The measurement<br />
model orig<strong>in</strong>ated <strong>in</strong> family systems <strong>the</strong>ory, <strong>and</strong> three types of<br />
friendships are identified: <strong>in</strong>terdependent ones <strong>in</strong> which cooper-<br />
ation <strong>and</strong> autonomy are balanced, disengaged ones <strong>in</strong> which<br />
friends are disconnected <strong>in</strong> spite of <strong>the</strong>ir efforts to ma<strong>in</strong>-<br />
ta<strong>in</strong> proximity with one ano<strong>the</strong>r, <strong>and</strong> consensus-sensitive or<br />
-enmeshed relationships <strong>in</strong> which agreement <strong>and</strong> cohesion are<br />
maximized. Empirical data that support this factor structure<br />
derive from adolescents' <strong>in</strong>teractions with <strong>the</strong>ir friends <strong>in</strong> a co-<br />
operative task.<br />
Developmental Significance<br />
Cross-sectional studies. Among children <strong>and</strong> adolescents,<br />
supportiveness between friends (high vs. low) is positively cor-<br />
related with school <strong>in</strong>volvement <strong>and</strong> achievement (Berndt &<br />
Hawk<strong>in</strong>s, 1991; Cauce, 1986) <strong>and</strong> negatively correlated with<br />
school-based problems (Kurdek & S<strong>in</strong>clair, 1988). Closeness is<br />
positively correlated with popularity <strong>and</strong> good social reputations<br />
( Cauce, 1986), with self-esteem (Mannar<strong>in</strong>o, 1978; McGuire &<br />
Weisz, 1982), <strong>and</strong> with psychosocial adjustment (Buhrmester,<br />
1990). Supportiveness is negatively correlated with identity<br />
problems (Pap<strong>in</strong>i, Farmer, Clark, Micke, & Barnett, 1990) as<br />
well as del<strong>in</strong>quency <strong>and</strong> depression (W<strong>in</strong>dle, 1994). Among<br />
middle-aged adults, closeness <strong>and</strong> communality at work (ra<strong>the</strong>r<br />
than equity) are correlated with job satisfaction, social satisfac-<br />
tion, <strong>and</strong> attitudes toward supervisors (W<strong>in</strong>stead et al., 1995).<br />
Among older adults, well-be<strong>in</strong>g <strong>and</strong> friendship satisfaction are<br />
associated with closeness (Lowenthal & Haven, 1968; Ro-<br />
berto & Scott, 1986), although <strong>the</strong> friendship quality measures<br />
<strong>in</strong> <strong>the</strong>se studies sometimes consist of hav<strong>in</strong>g close friends ra<strong>the</strong>r<br />
than friendship closeness. Results are thus consistent: Support-<br />
iveness between friends (<strong>the</strong> deep structure) <strong>and</strong> well-be<strong>in</strong>g<br />
are correlated from childhood through old age, support<strong>in</strong>g our<br />
hypo<strong>the</strong>sis that friendship prototypes (i.e., those most clearly<br />
reflect<strong>in</strong>g reciprocity) have similar affective concomitants<br />
across <strong>the</strong> life course.<br />
Correlations are negative between o<strong>the</strong>r friendship attributes<br />
<strong>and</strong> adaptive outcome. Among adolescents, for example, both<br />
overt <strong>and</strong> covert hostility <strong>in</strong> relation to one's friends are posi-<br />
tively correlated with alcohol use, del<strong>in</strong>quency, <strong>and</strong> depressive<br />
symptoms. Self-disclosure between friends is also positively<br />
correlated with alcohol use (W<strong>in</strong>dle, 1994), perhaps because<br />
friendship <strong>and</strong> alcohol both dis<strong>in</strong>hibit reticence. Also among<br />
adolescents, conflict <strong>and</strong> contention with friends are negatively<br />
correlated with attitudes toward school (Berndt, 1989).<br />
Once aga<strong>in</strong>, <strong>the</strong>se correlational results are impossible to <strong>in</strong>ter-<br />
pret. Psychological well-be<strong>in</strong>g may foster closeness between<br />
friends, especially when both <strong>in</strong>dividuals have high self-esteem,<br />
but <strong>the</strong> reverse may also be true. Similarly, problem behaviors
among adolescents (e.g., alcohol use) may foster self-disclosure<br />
between friends as well as tension <strong>and</strong> disagreement, but <strong>the</strong><br />
reverse may also occur.<br />
Longitud<strong>in</strong>al studies. Longitud<strong>in</strong>al studies that deal with<br />
friendship quality are centered mostly on school attitudes,<br />
<strong>in</strong>volvement, <strong>and</strong> achievement among children <strong>and</strong> adolescents.<br />
One cross-lagged <strong>in</strong>vestigation (Berndt, 1989) across <strong>the</strong> transi-<br />
tion from elementary to junior high school shows that variations<br />
<strong>in</strong> developmental outcome can be predicted with comprehensive<br />
friendship assessments that <strong>in</strong>clude friendship quality. Size of<br />
<strong>the</strong> friendship network, friendship stability, <strong>and</strong> self-reported<br />
friendship quality were simultaneously studied; outcome mea-<br />
sures <strong>in</strong>cluded school attitudes <strong>and</strong> achievement. First, network<br />
size was negatively related to friendship supportiveness <strong>and</strong> <strong>in</strong>ti-<br />
macy (friendship quality). Second, nei<strong>the</strong>r number of friends<br />
nor friendship stability contributed to changes <strong>in</strong> school adjust-<br />
ment--across ei<strong>the</strong>r <strong>the</strong> school transition or <strong>the</strong> first year <strong>in</strong> <strong>the</strong><br />
new school. (School adjustment was relatively stable across <strong>the</strong><br />
transition <strong>and</strong> was related to friendship stability cross-section-<br />
ally but not with earlier adjustment factored out.) Third, <strong>the</strong><br />
supportiveness of <strong>the</strong> child's friends, assessed shortly after en-<br />
trance to <strong>the</strong> new school, predicted <strong>in</strong>creas<strong>in</strong>g popularity <strong>and</strong><br />
<strong>in</strong>creas<strong>in</strong>gly positive attitudes toward classmates over <strong>the</strong> next<br />
year. Clearly, hav<strong>in</strong>g friends, <strong>the</strong> stability of <strong>the</strong>se relationships,<br />
<strong>and</strong> <strong>the</strong>ir supportiveness (i.e., quality) have different develop-<br />
mental implications. Consistent results have been reported for<br />
younger children (Ladd, 1990) as well as adolescents (Berndt &<br />
Keefe, 1992) over a period of 9 to 12 months.<br />
O<strong>the</strong>r researchers have exam<strong>in</strong>ed <strong>the</strong> relation between social<br />
support (<strong>in</strong>clud<strong>in</strong>g friends) <strong>and</strong> problem behavior. In general,<br />
<strong>in</strong>creases <strong>in</strong> social support experienced <strong>in</strong> childhood <strong>and</strong> adoles-<br />
cence are accompanied by <strong>in</strong>creas<strong>in</strong>gly better social adaptation<br />
(Dubow, Tisak, Causey, Hryshko, & Reid, 1991). Complex<br />
<strong>in</strong>teractions exist, however, to <strong>the</strong> effect that friendship quality<br />
<strong>and</strong> <strong>the</strong> identity of a child's friends <strong>in</strong>teract with stress occur-<br />
rence to determ<strong>in</strong>e outcome. For example, regression models<br />
(W<strong>in</strong>dle, 1992) show that, among adolescent boys experienc<strong>in</strong>g<br />
high stress, friend support encourages both alcohol use <strong>and</strong><br />
depression; among boys experienc<strong>in</strong>g moderate or low stress,<br />
nonsupportive friends encourage both alcohol use <strong>and</strong> depres-<br />
sion. Among girls, <strong>the</strong> results are more straightforward: Sup-<br />
portiveness among one's friends is positively correlated with<br />
alcohol use but negatively correlated with depression (with <strong>in</strong>i-<br />
tial adjustment factored out). One does not know whe<strong>the</strong>r <strong>the</strong><br />
stressed-out boys <strong>in</strong> this <strong>in</strong>vestigation also had stressed-out<br />
friends--a reasonable supposition, given <strong>the</strong> behavioral similar-<br />
ities known to exist between friends (Dishion et al., 1995).<br />
Never<strong>the</strong>less, <strong>the</strong> admonition that friendships are mixed bless-<br />
<strong>in</strong>gs should be recalled: Childhood friends with behavioral dif-<br />
ficulties may provide one ano<strong>the</strong>r with emotional support, but<br />
<strong>the</strong> <strong>in</strong>teractions that occur between <strong>the</strong>m may not simultaneously<br />
predict good outcome. Moreover, supportive friendships may<br />
have deleterious effects when <strong>the</strong> support emanates from a be-<br />
haviorally disturbed friend but good effects when emanat<strong>in</strong>g<br />
from a better adjusted friend. Only when <strong>the</strong> identity of an<br />
adolescent's friend is specified along with relationship quality<br />
(i.e., supportiveness) will <strong>the</strong>se results be <strong>in</strong>terpretabl e.<br />
Although causal <strong>in</strong>fluence seems to extend from friendship<br />
quality to <strong>in</strong>dividual adaptation, this does not rule out <strong>the</strong> possi-<br />
bility that attributes of <strong>in</strong>dividuals also affect friendship quality.<br />
FRIENDSHIPS IN THE LIFE COURSE 365<br />
W<strong>in</strong>dle (1994), for example, discovered that, among adolescent<br />
friends, behavior problems ev<strong>in</strong>ced by <strong>in</strong>dividuals predicted<br />
friendship characteristics across time ra<strong>the</strong>r than <strong>the</strong> reverse.<br />
Both del<strong>in</strong>quency <strong>and</strong> depression, for example, predicted overt<br />
<strong>and</strong> covert hostility with friends but not vice versa. Perhaps one<br />
select<strong>in</strong>g antisocial friends <strong>and</strong> socializ<strong>in</strong>g with <strong>the</strong>m br<strong>in</strong>gs<br />
about dissension between <strong>the</strong>se same <strong>in</strong>dividuals, which, <strong>in</strong> turn,<br />
affects developmental outcome <strong>in</strong>dividually. Results are some-<br />
what difficult to <strong>in</strong>terpret because friendship attributes <strong>and</strong> so-<br />
cial behaviors were both assessed by means of self-reports (thus<br />
subject to contam<strong>in</strong>ation) from s<strong>in</strong>gle <strong>in</strong>dividuals ra<strong>the</strong>r than<br />
both members of each friendship dyad (thus biased). More<br />
detailed study of <strong>the</strong> natural history of adolescent friendships<br />
is needed to disentangle <strong>the</strong>se effects (see W<strong>in</strong>dle, 1994).<br />
Relatively little is known about friendship quality as a pre-<br />
dictor of adult adjustment, even though o<strong>the</strong>r aspects of peer<br />
competence <strong>in</strong> adolescence (not childhood) generally predict<br />
sociability, good marital adjustment, <strong>and</strong> mental health status <strong>in</strong><br />
adulthood (Bagwell et al., 1996; Skolnick, 1986). Attitudes <strong>and</strong><br />
commitments of older persons' friends to one ano<strong>the</strong>r become<br />
<strong>in</strong>creas<strong>in</strong>gly positive as <strong>the</strong>se relationships become closer. Spe-<br />
cifically, a direct relation has been discovered between closeness<br />
changes <strong>and</strong> <strong>the</strong> exchange of resources among older participants<br />
(Shea et al., 1988). Friendship quality seems not to have been<br />
used, though, to forecast developmental outcome across transi-<br />
tions to retirement, widowhood, or residential liv<strong>in</strong>g, even<br />
though older <strong>in</strong>dividuals (e.g., widows) are known to use friends<br />
to support new activities on <strong>the</strong>se occasions (Bankoff, 1983;<br />
Stevens, 1989). Sometimes, developmental transitions <strong>the</strong>m-<br />
selves precipitate changes <strong>in</strong> friendship quality, as when a per-<br />
son's decl<strong>in</strong><strong>in</strong>g health <strong>and</strong> mov<strong>in</strong>g to a nurs<strong>in</strong>g home change<br />
<strong>the</strong> friendship network from close to casual friends (Adams,<br />
1987).<br />
To summarize, although relatively little is known about friend-<br />
ship quality <strong>and</strong> its developmental implications, <strong>the</strong> weight of<br />
<strong>the</strong> evidence suggests that good outcomes are most likely when<br />
one has friends, one's friends are well socialized, <strong>and</strong> when<br />
one's relationships with <strong>the</strong>se <strong>in</strong>dividuals are supportive <strong>and</strong><br />
<strong>in</strong>timate. But <strong>the</strong> amount of variance accounted for by each of<br />
<strong>the</strong>se relationship parameters may not be equal. Some dimen-<br />
sions, for example, hav<strong>in</strong>g friends, may account for relatively<br />
little outcome variance, whereas o<strong>the</strong>rs, for example, <strong>the</strong> identity<br />
of one's friends, may account for a great deal. More needs to<br />
be known, too, about <strong>the</strong> manner <strong>in</strong> which relationship attributes<br />
reflect <strong>the</strong> social histories of <strong>the</strong> <strong>in</strong>dividuals <strong>in</strong>volved.<br />
Regression models may differ too across different develop-<br />
mental transitions. Supportive friends may be <strong>the</strong> most important<br />
assets that a child can have when fac<strong>in</strong>g a school transition, but<br />
merely hav<strong>in</strong>g friends may be most important when one is mak-<br />
<strong>in</strong>g <strong>the</strong> transition to retirement. The identity of one's friends<br />
may be extremely important to determ<strong>in</strong>e whe<strong>the</strong>r an adolescent<br />
will move <strong>in</strong>to an adult crim<strong>in</strong>al career but relatively unimport-<br />
ant to determ<strong>in</strong>e good outcome across <strong>the</strong> transition to widow-<br />
hood. Comprehensive studies are badly needed, <strong>the</strong>refore, <strong>in</strong><br />
which friendship is studied <strong>in</strong> t<strong>and</strong>em with o<strong>the</strong>r predictors (e.g.,<br />
temperament <strong>and</strong> personality) ra<strong>the</strong>r than studied separately.<br />
Multivariate effects must be tracked through time. Children,<br />
adolescents, <strong>and</strong> adults need to participate more extensively <strong>in</strong><br />
longitud<strong>in</strong>al studies on friendship <strong>and</strong> its many vicissitudes.<br />
Birth-to-death studies are not necessarily needed, but short-
366 HARTUP AND STEVENS<br />
term longitud<strong>in</strong>al studies across major developmental transitions<br />
(e.g., adolescence, midlife, <strong>and</strong> retirement) would be extremely<br />
valuable.<br />
Conclusion<br />
The weight of <strong>the</strong> evidence suggests that friendships are de-<br />
velopmentally significant throughout <strong>the</strong> life course. First,<br />
friends are cognitive <strong>and</strong> affective resources from childhood<br />
through old age, foster<strong>in</strong>g self-esteem <strong>and</strong> a sense of well-be<strong>in</strong>g.<br />
Second, friends socialize one ano<strong>the</strong>r, especially with respect to<br />
age-related tasks that must be mastered for <strong>in</strong>dividuals to achieve<br />
good outcomes. Third, supportive <strong>and</strong> <strong>in</strong>timate relationships be-<br />
tween socially skilled <strong>in</strong>dividuals seem to be developmental<br />
advantages, whereas conflict-ridden relationships between trou-<br />
bled <strong>in</strong>dividuals seem to be disadvantages.<br />
The database on friendship <strong>and</strong> adaptation abounds with dis-<br />
cont<strong>in</strong>uities. Adults, for example, spend less time with <strong>the</strong>ir<br />
friends than children do <strong>and</strong> spend <strong>the</strong>ir time differently. Sup-<br />
portiveness between friends is correlated with good outcome,<br />
but this result depends on who one's friends are <strong>and</strong> what <strong>the</strong>se<br />
relationships are like. Consistency emerges, however, when one<br />
exam<strong>in</strong>es friendships with<strong>in</strong> a multidimensional framework.<br />
First, one must dist<strong>in</strong>guish between deep <strong>and</strong> surface struc-<br />
tures when th<strong>in</strong>k<strong>in</strong>g about <strong>the</strong>se relationships. Based on empiri-<br />
cal evidence, reciprocity is <strong>the</strong> deep structure, <strong>and</strong> certa<strong>in</strong> out-<br />
comes (e.g., <strong>the</strong> <strong>in</strong>dividual's sense of well-be<strong>in</strong>g) derive from<br />
<strong>the</strong>se reciprocities throughout <strong>the</strong> life course. Surface structures<br />
vary based on how old <strong>the</strong> <strong>in</strong>dividuals are: Four-year-olds, for<br />
example, engage <strong>in</strong> rough-<strong>and</strong>-tumble play; 14-year-olds social-<br />
ize <strong>in</strong> shopp<strong>in</strong>g malls <strong>and</strong> over <strong>the</strong> telephone; 34-year-olds relate<br />
to one ano<strong>the</strong>r by discuss<strong>in</strong>g jobs, parent<strong>in</strong>g, <strong>and</strong> golf scores;<br />
74-year-olds rem<strong>in</strong>isce <strong>and</strong> discuss <strong>the</strong>ir health <strong>and</strong> what <strong>the</strong><br />
world is com<strong>in</strong>g to. Surface structure changes, however, are<br />
precisely why <strong>the</strong>se relationships are adaptational advantages.<br />
Friendship reciprocities are cast <strong>and</strong> recast so that <strong>the</strong>y can<br />
support <strong>the</strong> <strong>in</strong>dividuals <strong>in</strong>volved <strong>in</strong> developmentally relevant<br />
ways.<br />
Second, one must recognize that, although hav<strong>in</strong>g friends<br />
may be a developmental advantage, all friendships are not alike.<br />
People differ <strong>in</strong> <strong>the</strong> company <strong>the</strong>y keep, that is, <strong>in</strong> who <strong>the</strong>ir<br />
friends are. Both children <strong>and</strong> adults also differ <strong>in</strong> that some<br />
friendships are close, stable, <strong>and</strong> symmetrical, while o<strong>the</strong>rs are<br />
not. Only when three dimensions--hav<strong>in</strong>g friends, identity of<br />
one's friends, <strong>and</strong> quality of <strong>the</strong> friendship--are taken <strong>in</strong>to ac-<br />
count does <strong>the</strong> literature yield a coherent account of friendship<br />
<strong>and</strong> adaptation <strong>in</strong> <strong>the</strong> life course.<br />
Significant gaps exist <strong>in</strong> <strong>the</strong> database. Some issues are unre-<br />
solved because too little is known about how friendships differ<br />
from one ano<strong>the</strong>r. Some conclusions are shaky because cross-<br />
sectional data have not been supplemented with longitud<strong>in</strong>al<br />
studies. O<strong>the</strong>r conclusions are tenuous because <strong>the</strong> database<br />
comes from children <strong>and</strong> adolescents but too little is known<br />
about <strong>the</strong> relevant issue among adults. Still o<strong>the</strong>r studies are<br />
<strong>in</strong>conclusive because too little is known about <strong>the</strong> complex man-<br />
ner <strong>in</strong> which relationships <strong>and</strong> <strong>the</strong>ir development <strong>in</strong>teract with<br />
<strong>the</strong> development of <strong>the</strong> <strong>in</strong>dividual.<br />
Among <strong>the</strong> many issues that need attention are <strong>the</strong>se three.<br />
First, more attention must be given to <strong>the</strong> manner <strong>in</strong> which<br />
friendships differ from one ano<strong>the</strong>r. New assessments of close-<br />
ness, supportiveness, <strong>and</strong> hostility <strong>in</strong> <strong>the</strong>se relationships, espe-<br />
cially among children <strong>and</strong> adolescents, are good beg<strong>in</strong>n<strong>in</strong>gs.<br />
Many o<strong>the</strong>r qualitative dimensions, however, need exam<strong>in</strong>ation.<br />
Friends vary based on how many <strong>and</strong> which attributes <strong>the</strong>y<br />
share: Are friends who are similar <strong>in</strong> many ways different from<br />
those who are similar <strong>in</strong> only one or two ways? What are <strong>the</strong><br />
long-term implications of a person shar<strong>in</strong>g one attribute with a<br />
friend versus shar<strong>in</strong>g many attributes? One can guess that, when<br />
many attributes are shared, <strong>in</strong>terpersonal attraction can be great<br />
(Kupersmidt et al., 1995), but does this mean that <strong>the</strong>se friends<br />
have certa<strong>in</strong> developmental advantages? Are certa<strong>in</strong> similarities<br />
more significant than o<strong>the</strong>rs? Do homophilies with depression<br />
place <strong>in</strong>dividuals at <strong>the</strong> same degree of risk as would those with<br />
antisocial behavior? F<strong>in</strong>ally, when one considers that friendships<br />
differ from one ano<strong>the</strong>r <strong>in</strong> myriad ways, which qualitative attri-<br />
butes carry <strong>the</strong> greatest developmental significance <strong>and</strong> when?<br />
Second, what normative cont<strong>in</strong>uities <strong>and</strong> discont<strong>in</strong>uities can<br />
be traced across <strong>the</strong> life course? What br<strong>in</strong>gs about normative<br />
change <strong>in</strong> friendship <strong>in</strong>teractions? Exist<strong>in</strong>g normative studies<br />
center too much on hav<strong>in</strong>g friends <strong>and</strong> <strong>the</strong> most superficial facts<br />
about <strong>the</strong>m. Normative trends over <strong>the</strong> life course can be de-<br />
duced only from diverse studies that vary greatly <strong>in</strong> methodology<br />
<strong>and</strong> quality. For example, observational studies tell many th<strong>in</strong>gs<br />
about <strong>the</strong> behavioral manifestations of children's friendships,<br />
but virtually noth<strong>in</strong>g is known about <strong>the</strong>se manifestations among<br />
middle-aged or older friends. Cross-sectional normative studies<br />
<strong>in</strong>clud<strong>in</strong>g children <strong>and</strong> adults would improve this state of affairs;<br />
but longitud<strong>in</strong>al studies are also needed. Whatever can be<br />
learned about <strong>the</strong> dist<strong>in</strong>ctiveness of friendship <strong>in</strong>teraction among<br />
adults <strong>and</strong> <strong>the</strong> changes that occur across major life transitions<br />
<strong>in</strong> adulthood (e.g., widowhood) would be extremely valuable.<br />
Third, friendships <strong>and</strong> <strong>the</strong>ir developmental significance need<br />
to be better understood. Developmental psychologists are study-<br />
<strong>in</strong>g behavioral outcomes <strong>in</strong> relation to temperament <strong>and</strong> early<br />
experience, social skills <strong>and</strong> sociometric status, family relation-<br />
ships, <strong>and</strong> <strong>the</strong> social context (most especially situations that<br />
<strong>in</strong>volve stress <strong>and</strong> o<strong>the</strong>r challenges). Multiple pathways that<br />
<strong>in</strong>volve comb<strong>in</strong>ations of <strong>the</strong>se conditions are be<strong>in</strong>g discovered<br />
that lead to good outcomes, meanwhile o<strong>the</strong>r pathways are be<strong>in</strong>g<br />
discovered that lead to poorer ones (Hartup & Van Lieshout,<br />
1995). Friendship experiences may contribute a significant vari-<br />
ance with<strong>in</strong> some pathways but not o<strong>the</strong>rs--a cont<strong>in</strong>gency that<br />
has somehow been ignored.<br />
Overall, our assessment shows documentation on friendships<br />
<strong>and</strong> <strong>the</strong>ir developmental significance to be sufficiently strong<br />
for <strong>the</strong> argument that friendships should not be ignored <strong>in</strong> devel-<br />
opmental analysis--nei<strong>the</strong>r <strong>in</strong> childhood nor throughout <strong>the</strong> life<br />
course. Causal models are weak, however, <strong>and</strong> need to be im-<br />
proved. Consideration of both deep <strong>and</strong> surface structures as<br />
well as multidimensional assessment <strong>in</strong> which hav<strong>in</strong>g friends,<br />
<strong>the</strong> identity of one's friends, <strong>and</strong> quality of <strong>the</strong> friendship are<br />
essential to this improvement.<br />
References<br />
Adams, R. G. (1987). Patterns of network change: A longitud<strong>in</strong>al study<br />
of friendships of elderly women. Gerontologist, 27, 222-227.<br />
Adams, R. G., & Blieszner, R. (1995). Midlife friendship patterns. In<br />
N. Vanzetti & S. Duck (Eds.), A lifetime of relationships (pp. 336-<br />
363). San Francisco: Brooks/Cole.
A<strong>in</strong>sworth, M.D.S., Blehar, M.C., Waters, E., & Wall, S. (1978).<br />
Patterns of attachment: A psychological study of <strong>the</strong> Strange Situa-<br />
tion. Hillsdale, NJ: Erlbaum.<br />
Altergott, K. (1988). Social action <strong>and</strong> <strong>in</strong>teraction <strong>in</strong> later life: Ag<strong>in</strong>g<br />
<strong>in</strong> <strong>the</strong> United States. In K. Altergott (Ed.), Daily life <strong>in</strong> later life:<br />
Comparative perspectives (pp. 117-146). Newbury Park, CA: Sage.<br />
Altman, I., & Taylor, D. A. (1973). Socialpenetration: The development<br />
of <strong>in</strong>terpersonal relationships. New York: Holt, R<strong>in</strong>ehart, & W<strong>in</strong>ston.<br />
Argyle, M., & Furnham, A. (1983). Sources of satisfaction <strong>and</strong> conflict<br />
<strong>in</strong> long-term relationships. Journal of Marriage <strong>and</strong> <strong>the</strong> Family, 45,<br />
481-493.<br />
Armstrong, M. J., & Goldsteen, K. S. (1990). Friendship support pat-<br />
terns of older American women. Journal of Ag<strong>in</strong>g Studies, 4, 391-<br />
404.<br />
Asher, S. R., Parkhurst, J. T., Hymel, S., & Williams, G. A. (1990). Peer<br />
rejection <strong>and</strong> lonel<strong>in</strong>ess <strong>in</strong> childhood. In S. R. Asher & J.D. Coie<br />
(Eds.), Peer rejection <strong>in</strong> childhood (pp. 253-273). New York: Cam-<br />
bridge University Press.<br />
Azmitia, M., & Montgomery, R. (1993). Friendship, transactive dia-<br />
logues, <strong>and</strong> <strong>the</strong> development of scientific reason<strong>in</strong>g. Social Develop-<br />
ment, 2, 202-221.<br />
Bagwell, C.L., Newcomb, A. E, & Bukowski, W.M. (1996). Pre-<br />
adolescent friendship <strong>and</strong> peer rejection as predictors of adult adjust-<br />
ment. Unpublished manuscript, University of Richmond, Department<br />
of <strong>Psychology</strong>, Richmond, VA.<br />
Ball, S.J. (1981). Beachside comprehensive. Cambridge, Engl<strong>and</strong>:<br />
Cambridge University Press.<br />
Bankoff, E. (1983). Social support <strong>and</strong> adaptation to widowhood. Jour-<br />
nal of Marriage <strong>and</strong> <strong>the</strong> Family, 45, 827-839.<br />
Berg, J. H. ( 1983, August). Attraction <strong>in</strong> relationships: As it beg<strong>in</strong>s so<br />
it goes. Paper presented at <strong>the</strong> 91 st Annual Convention of <strong>the</strong> American<br />
Psychological Association, Anaheim, CA.<br />
Berg, J. H. (1984). The development of friendship between roommates.<br />
Journal of Personality <strong>and</strong> Social <strong>Psychology</strong>, 46, 346-356.<br />
Berg, J.H., & Clark, M.S. (1989). Differences <strong>in</strong> social exchange<br />
between <strong>in</strong>timate <strong>and</strong> o<strong>the</strong>r relationships: Gradually evolv<strong>in</strong>g or<br />
quickly apparent? In V. J. Derlega & B. A. W<strong>in</strong>stead (Eds.), Friend-<br />
ship <strong>and</strong> social <strong>in</strong>teraction (pp. 101-128). New York: Spr<strong>in</strong>ger<br />
Verlag.<br />
Berndt, T. J. (1989). Obta<strong>in</strong><strong>in</strong>g support from friends dur<strong>in</strong>g childhood<br />
<strong>and</strong> adolescence. In D. Belle (Ed.), Children's social networks <strong>and</strong><br />
social supports (pp. 308-331 ). New York: Wiley.<br />
Berndt, T. J. (1996). Explor<strong>in</strong>g <strong>the</strong> effects of friendship quality on social<br />
development. In W. M. Bukowski, A. E Newcomb, & W. W. Hartup<br />
(Eds.), The company <strong>the</strong>y keep: <strong>Friendships</strong> <strong>in</strong> childhood <strong>and</strong> adoles-<br />
cence (pp. 346-365). New York: Cambridge University Press.<br />
Berndt, T. J., & Hawk<strong>in</strong>s, J. A. (1991). Effects of friendship on adoles-<br />
cents' adjustment to junior high school. Unpublished manuscript,<br />
Purdue University, Department of Psychological Sciences, West Lafay-<br />
ette, IN.<br />
Berndt, T. J., Hawk<strong>in</strong>s, J. A., & Hoyle, S. G. (1986). Changes <strong>in</strong> friend-<br />
ship dur<strong>in</strong>g a school year: Effects on children's <strong>and</strong> adolescents'<br />
impressions of friendship <strong>and</strong> shar<strong>in</strong>g with friends. Child Develop-<br />
ment, 57, 1284-1297.<br />
Berndt, T.J., & Keefe, K. (1992). Friends' <strong>in</strong>fluence on adolescents'<br />
perceptions of <strong>the</strong>mselves <strong>in</strong> school. In D. H. Schunk & J. L. Meece<br />
(Eds.), Students'perceptions <strong>in</strong> <strong>the</strong> classroom (pp. 51-73). Hillsdale,<br />
NJ: Erlbaum.<br />
Berscheid, E. (1994). Interpersonal relationships. Annual Review of<br />
<strong>Psychology</strong>, 45, 79-130.<br />
Berscheid, E., Snyder, M., & Omoto, A. M. (1989). The Relationship<br />
Closeness Inventory: Assess<strong>in</strong>g <strong>the</strong> closeness of <strong>in</strong>terpersonal relation-<br />
ships. Journal of Personality <strong>and</strong> Social <strong>Psychology</strong>, 57, 792-807.<br />
Berscheid, E., & Walster, E. (1968). Interpersonal attraction (2nd ed.).<br />
Read<strong>in</strong>g, MA: Addison-Wesley.<br />
FRIENDSHIPS IN THE LIFE COURSE 367<br />
Bigelow, B. J. (1977). Children's friendship expectations: A cognitive<br />
developmental study. Child Development, 48, 246-253.<br />
Billy, J. O. G., Rodgers, J. L., & Udry, J. R. (1984). Adolescent sexual<br />
behavior <strong>and</strong> friendship choice. Social Forces, 62, 653-678.<br />
Blieszner, R., & Adams, R. G. (1992). Adult friendship. Newbury Park,<br />
CA: Sage.<br />
Bowlby, J. (1969). Attachment <strong>and</strong> loss. Vol. 1: Attachment. New York:<br />
Basic Books.<br />
Brown, B. B. (1981). A life-span approach to friendship: Age-related<br />
dimensions of an ageless relationship. In H. Lopata & D. Ma<strong>in</strong>es<br />
(Eds.), Research <strong>in</strong> <strong>the</strong> <strong>in</strong>terweave of social roles (Vol. 2, pp. 23-<br />
50). Greenwich, CT:. JAI Press.<br />
Brown, B.B., Clasen, D.R., & Eicher, S.A. (1986). Perceptions of<br />
peer pressure, peer conformity dispositions, <strong>and</strong> self-reported behav-<br />
ior among adolescents. Developmental <strong>Psychology</strong>, 22, 521-530.<br />
Brownell, C. A. (1986). Convergent developments: Cognitive-develop-<br />
mental correlates of growth <strong>in</strong> <strong>in</strong>fant/toddler peer skills. Child Devel-<br />
opment, 57, 275-286.<br />
Buhrmester, D. (1990). Intimacy of friendship, <strong>in</strong>terpersonal compe-<br />
tence, <strong>and</strong> adjustment dur<strong>in</strong>g preadolescence <strong>and</strong> adolescence. Child<br />
Development, 61, 1101 - 1111.<br />
Bukowski, W. M., Hoza, B., & Boiv<strong>in</strong>, M. (1994). Measur<strong>in</strong>g friendship<br />
quality dur<strong>in</strong>g pre- <strong>and</strong> early adolescence: The development <strong>and</strong> psy-<br />
chometric properties of <strong>the</strong> Friendship Qualities Scale. Journal of<br />
Social <strong>and</strong> Personal Relationships, 11, 471-484.<br />
Bukowski, W. M., Hoza, B., & Newcomb, A.F. (1991). Friendship,<br />
popularity, <strong>and</strong> <strong>the</strong> "self" dur<strong>in</strong>g early adolescence. Unpublished<br />
manuscript, Concordia University, Department of <strong>Psychology</strong>, Mon-<br />
treal, Quebec, Canada.<br />
Cairns, R. B., & Cairns, B. D. (1994). <strong>Life</strong>l<strong>in</strong>es <strong>and</strong> risks. New York:<br />
Cambridge University Press.<br />
Cairns, R. B., Leung, M., Buchanan, L., & Cairns, B. D. (1995). Friend-<br />
ships <strong>and</strong> social networks <strong>in</strong> childhood <strong>and</strong> adolescence: Fluidity,<br />
reliability, <strong>and</strong> <strong>in</strong>terrelations. Child Development, 66, 1330-1345.<br />
Caspi, A., Eider, G. H., &Bem, D. J. (1988). Mov<strong>in</strong>g away from <strong>the</strong><br />
world: <strong>Life</strong>-course patterns of shy children. Developmental Psychol-<br />
ogy, 24, 824-831.<br />
Cauce, A. M. (1986). Social networks <strong>and</strong> social competence: Explor<strong>in</strong>g<br />
<strong>the</strong> effects of early adolescent friendships. American Journal of Com-<br />
munity <strong>Psychology</strong>, 14, 607-628.<br />
Challman, R.C. (1932). Factors <strong>in</strong>fluenc<strong>in</strong>g friendships among pre-<br />
school children. Child Development, 3, 146-158.<br />
Chomsky, N. (1965). Aspects of a <strong>the</strong>ory of syntax. Cambridge, MA:<br />
MIT Press.<br />
Chown, S. M. (1981). Friendship <strong>in</strong> old age. In S. Duck & R. Gilmour<br />
(Eds.), Personal relationships. Vol. 2: Develop<strong>in</strong>g personal relation-<br />
ships (pp. 231-246). London: Academic Press.<br />
Clark, M. S., & Mills, J. (1979). Interpersonal attraction <strong>in</strong> exchange<br />
<strong>and</strong> communal relationships. Journal of Personality <strong>and</strong> Social Psy-<br />
chology, 37, 12-24.<br />
Coie, J.D., Dodge, K.A., & Kupersmidt, J.B. (1990). Peer group<br />
behavior <strong>and</strong> social status. In S. R. Asher & J. D. Coie (Eds.), Peer<br />
rejection <strong>in</strong> childhood (pp. 17-59). New York: Cambridge University<br />
Press.<br />
Coll<strong>in</strong>s, W. A., & Rep<strong>in</strong>ski, D. (1991, April). Development of relation-<br />
ships dur<strong>in</strong>g <strong>the</strong> transition to adolescence: Processes of adaptation<br />
to <strong>in</strong>dividual change. Paper presented at <strong>the</strong> biennial meet<strong>in</strong>gs of <strong>the</strong><br />
Society for Research <strong>in</strong> Child Development, Seattle, WA.<br />
Connidis, I. A., & Davies, L. (1990). Confidants <strong>and</strong> companions <strong>in</strong><br />
later life: The place of family <strong>and</strong> friends. Journal of Gerontology:<br />
Social Sciences, 45, S141-S149.<br />
Dickens, W. J., & Perlman, D. (1981). Friendship over <strong>the</strong> life cycle.<br />
In S. Duck & R. Gilmour (Eds.), Personal relationships. Vol. 2:<br />
Develop<strong>in</strong>g personal relationships (pp. 91 - 122). London: Academic<br />
Press.<br />
Dishion, T. J., Andrews, D. W., & Crosby, L. (1995). Anti-social boys
368 HARTUPANDSTEVENS<br />
<strong>and</strong> <strong>the</strong>ir friends <strong>in</strong> early adolescence: Relationship characteristics,<br />
quality, <strong>and</strong> <strong>in</strong>teractional process. Child Development, 66, 139-151.<br />
Dishion, T. J., Patterson, G. R., & Griesler, P. C. (1994). Peer adapta-<br />
tions <strong>in</strong> <strong>the</strong> development of antisocial behavior: A confluence model.<br />
In L. R. Huesmann (Ed.), Current perspectives on aggressive behav-<br />
ior (pp. 61-95). New York: Plenum.<br />
Dubow, E. E, Tisak, J., Causey, D., Hryshko, A., & Reid, G. (1991).<br />
A two-year longitud<strong>in</strong>al study of stressful life events, social support,<br />
<strong>and</strong> social problem-solv<strong>in</strong>g skills: Contributions to children's behav-<br />
ioral <strong>and</strong> academic adjustment. Child Development, 62, 583-599.<br />
Dykstra, P. (1995a). Lonel<strong>in</strong>ess among <strong>the</strong> never <strong>and</strong> formerly married:<br />
The importance of supportive friendships <strong>and</strong> a desire for <strong>in</strong>depen-<br />
dence. Journals of Gerontology: Psychological Sciences <strong>and</strong> Social<br />
Sciences, 50B, $321-$329.<br />
Dykstra, P. (1995b). Network composition. In C. P. M. Knipscheer, J.<br />
de Jong Gierveld, T.G. van Tilburg, & P. Dykstra (Eds.), Liv<strong>in</strong>g<br />
arrangements <strong>and</strong> social networks of older adults (pp. 97-114).<br />
Amsterdam: Free University Press.<br />
Eder, D., & Hall<strong>in</strong>an, M. T. (1978). Sex differences <strong>in</strong> children's friend-<br />
ships. American Sociological Review, 43, 237-250.<br />
Elicker, J., Englund., M., & Sroufe, L. A. (1992). Predict<strong>in</strong>g peer com-<br />
petence <strong>and</strong> peer relationships <strong>in</strong> childhood from early parent-child<br />
relationships. In R. D. Parke & G. W. Ladd (Eds.), Family-peer rela-<br />
tionships: Modes of l<strong>in</strong>kage (pp. 77-106). Hillsdale, NJ: Erlbaum.<br />
Epste<strong>in</strong>, J. L. (1983). Exam<strong>in</strong><strong>in</strong>g <strong>the</strong>ories of adolescent friendship. In<br />
J.L. Epste<strong>in</strong> & N.L. Karweit (Eds.), Friends <strong>in</strong> school (pp. 39-<br />
61 ). San Diego, CA: Academic Press.<br />
Erw<strong>in</strong>, P. G. (1985). Similarity of attitudes <strong>and</strong> constructs <strong>in</strong> children's<br />
friendships. Journal of Experimental Child <strong>Psychology</strong>, 40, 470-485.<br />
Fagot, B. I. (1978). Re<strong>in</strong>forc<strong>in</strong>g cont<strong>in</strong>gencies for sex role behax~iors:<br />
Effect of experience with children. Child Development, 49, 30-36.<br />
Felton, B. J., & Berry, C. A. (1992). Do <strong>the</strong> sources of urban elderly's<br />
social support determ<strong>in</strong>e its psychological consequences? <strong>Psychology</strong><br />
<strong>and</strong> Ag<strong>in</strong>g, 7, 89-97.<br />
Field, D. (1995, August). Cont<strong>in</strong>uity <strong>and</strong> change <strong>in</strong> friendships <strong>in</strong> ad-<br />
vanced old age: F<strong>in</strong>d<strong>in</strong>gs from <strong>the</strong> Berkeley Older Generation Study.<br />
Paper presented at <strong>the</strong> 7th European Conference on Developmental<br />
<strong>Psychology</strong>, Krakow, Pol<strong>and</strong>.<br />
Fischer, C. S., & Phillips, S. L. (1982). Who is alone? Social characteris-<br />
tics of people with small networks. In L. A. Peplau & D. Perlman<br />
(Eds.), Lonel<strong>in</strong>ess: A sourcebook of current <strong>the</strong>ory, research, <strong>and</strong><br />
<strong>the</strong>rapy (pp. 21-39). New York: Wiley Interscience.<br />
Fisher, C. B., Reid, J. D., & Melendez, M. (1989). Conflict <strong>in</strong> families<br />
<strong>and</strong> friendships of later life. Family Relations, 38, 83-89.<br />
Fisher, L. A., & Baumann, K. E. (1988). Influence <strong>and</strong> selection <strong>in</strong> <strong>the</strong><br />
friend-adolescent relationship: F<strong>in</strong>d<strong>in</strong>gs from studies of adolescent<br />
smok<strong>in</strong>g <strong>and</strong> dr<strong>in</strong>k<strong>in</strong>g. Journal of Applied Social <strong>Psychology</strong>, 18,<br />
289-314.<br />
Furman, W. (1996). The measurement of friendship perceptions: Con-<br />
ceptual <strong>and</strong> methodological issues. In W. M. Bukowski, A. E New-<br />
comb, & W. W. Hartup (Eds.), The company <strong>the</strong>y keep: <strong>Friendships</strong><br />
<strong>in</strong> childhood <strong>and</strong> adolescence (pp. 68-70). New York: Cambridge<br />
University Press.<br />
Furman, W., & Buhrmester, D. (1985). Children's perceptions of <strong>the</strong><br />
personal relationships <strong>in</strong> <strong>the</strong>ir social networks. Developmental Psy-<br />
chology, 21, 1016-1022.<br />
Gallagher, S. K., & Gerstel, N. (1993). K<strong>in</strong>keep<strong>in</strong>g <strong>and</strong> friend keep<strong>in</strong>g<br />
among older women: The effect of marriage. Gerontologist, 33, 675-<br />
681.<br />
G<strong>in</strong>sberg, D. (1986). Friendship <strong>and</strong> postdivorce adjustment. In J. M.<br />
Gottman & J. G. Parker (Eds.), Conversations of friends (pp. 346-<br />
376). Cambridge, Engl<strong>and</strong>: Cambridge University Press.<br />
G<strong>in</strong>sberg, D., & Gottman, J. M. (1986). Conversations of college room-<br />
mates: Similarities <strong>and</strong> differences <strong>in</strong> male <strong>and</strong> female friendships. In<br />
J. M, Gottman & J. G. Parker (Eds.), Conversations of friends (pp.<br />
241-291 ). Cambridge, Engl<strong>and</strong>: Cambridge University Press.<br />
Goldman, J. A. (1981). The social <strong>in</strong>teraction of preschool children <strong>in</strong><br />
same-age versus mixed-age group<strong>in</strong>gs. Child Development, 52, 644-<br />
650.<br />
Goodnow, J.J., & Burns, A. (1985). Home <strong>and</strong> school: A child's eye<br />
view. Sydney, Australia: Allen & Unw<strong>in</strong>.<br />
Gottman, J.M. (1979). Marital relationships. New York: Academic<br />
Press.<br />
Gottman, J. M., & Mettetal, G. (1986). Speculations about social <strong>and</strong><br />
affective development: Friendship <strong>and</strong> acqua<strong>in</strong>tanceship through ado-<br />
lescence. In J. M. Gottman & J. G. Parker (Eds.), Conversations of<br />
friends (pp. 192-237). Cambridge, Engl<strong>and</strong>: Cambridge University<br />
Press.<br />
Gupta, V., & Korte, C. (1994). The effects of a confidant <strong>and</strong> a peer<br />
group on <strong>the</strong> well-be<strong>in</strong>g of s<strong>in</strong>gle elders. International Journal of<br />
Ag<strong>in</strong>g <strong>and</strong> Human Development, 39, 293-302.<br />
Hall<strong>in</strong>an, M. T. (1980). Patterns of cliqu<strong>in</strong>g among youth. In H. C. Foot,<br />
A. J. Chapman, & J. R. Smith (Eels.), Friendship <strong>and</strong> peer relations<br />
<strong>in</strong> children (pp. 321-342). New York: Wiley.<br />
Hartup, W. W. (1989). Behavioral manifestations of children's friend-<br />
ships. In T.J. Berndt & G. W. Ladd (Eds.), Peer relations <strong>in</strong> child<br />
development (pp. 46-70). New York: Wiley.<br />
Hartup, W. W. (1995). The three faces of friendship. Journal of Social<br />
<strong>and</strong> Personal Relationships, 12, 569-574.<br />
Hartup, W. W. (1996). The company <strong>the</strong>y keep: <strong>Friendships</strong> <strong>and</strong> <strong>the</strong>ir<br />
developmental significance. Child Development, 67, 1-13.<br />
Hartup, W. W., Daiute, C., Zajac, R., & Sholl, W. (1995). Collaboration<br />
<strong>in</strong> creative writ<strong>in</strong>g by friends <strong>and</strong> nonfriends. Unpublished manuscript,<br />
University of M<strong>in</strong>nesota, Tw<strong>in</strong> Cities, Institute of Child Development,<br />
M<strong>in</strong>neapolis, MN.<br />
Hartup, W. W., & Van Lieshout, C. E M. (1995). Personality develop-<br />
ment <strong>in</strong> social context. Annual Review of <strong>Psychology</strong>, 46, 655-687.<br />
Haselager, G. J. T., Hartup, W. W., Van Lieshout, C. E M., & Riksen-<br />
Walraven, M. (1996). Behavioral similarities between friends <strong>and</strong><br />
nonfriends <strong>in</strong> middle childhood. Unpublished manuscript, University<br />
of Nijmegen, Department of Developmental <strong>Psychology</strong>, Nijmegen,<br />
The Ne<strong>the</strong>rl<strong>and</strong>s.<br />
Haugaard, J. J., & Tilly, C. (1988) Characteristics predict<strong>in</strong>g children's<br />
responses to sexual encounters with o<strong>the</strong>r children. Child Abuse <strong>and</strong><br />
Neglect, 12, 209-218.<br />
Havighurst, R. (1953). Human development <strong>and</strong> education. New York:<br />
Longmans, Green.<br />
Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an<br />
attachment process. Journal of Personality <strong>and</strong> Social <strong>Psychology</strong>,<br />
52, 511-524.<br />
Hess, B. (1972). Friendship. In M. W. Riley, M. Johnson, & A. Foner<br />
(Eds.), Ag<strong>in</strong>g <strong>and</strong> society (Vol. 3, pp. 357-393). New York: Russell<br />
Sage.<br />
H<strong>in</strong>de, R. A. (1979). Towards underst<strong>and</strong><strong>in</strong>g relationships. New York:<br />
Academic Press.<br />
H<strong>in</strong>de, R. A., Titmus, G., Easton, D,, & Tampl<strong>in</strong>, A. (1985). Incidence<br />
of "friendship" <strong>and</strong> behavior with strong associates versus non-<br />
associates <strong>in</strong> preschoolers. Child Development, 56, 234-245.<br />
Hochschild, A. (1973). The unexpected community. Englewood Cliffs,<br />
NJ: Prentice-Hail.<br />
Howes, C. ( 1983 ). Patterns of friendship. Child Development, 54, 1041 -<br />
1053.<br />
Howes, C. (1989). Peer <strong>in</strong>teraction of young children. Monographs of<br />
<strong>the</strong> Society for Research <strong>in</strong> Child Development, 53(Serial No. 217).<br />
Hymel, S., & Woody, E. (1991, April). Friends versus nonfriends:<br />
Perceptions of similarity across self, teacher, <strong>and</strong> peers. Paper pre-<br />
sented at <strong>the</strong> biennial meet<strong>in</strong>gs of <strong>the</strong> Society for Research <strong>in</strong> Child<br />
Development, Seattle, WA.<br />
Huston, A. C. (1983). Sex-typ<strong>in</strong>g. In P. H. Mussen (Series Ed.) & E. M.<br />
He<strong>the</strong>r<strong>in</strong>gton (Vol. Ed.), H<strong>and</strong>book of child psychology. Vol. 4: So-<br />
cialization, personality, <strong>and</strong> social development (pp. 387-467). New<br />
York: Wiley.
Ipsa, J. (1981). Peer support among Soviet day care toddlers. Interna-<br />
tional Journal of Behavioral Development, 4, 255-269.<br />
Jerrome, D. (1981). The significance of friendship for women <strong>in</strong> later<br />
life. Age<strong>in</strong>g <strong>and</strong> Society, 1, 175-196.<br />
K<strong>and</strong>el, D. B. (1978a). Homophily, selection, <strong>and</strong> socialization <strong>in</strong> ado-<br />
lescent friendships. American Journal of Sociology, 84, 427-436,<br />
K<strong>and</strong>el, D. B. (1978b). Similarity <strong>in</strong> real-life adolescent pairs. Journal<br />
of Personality <strong>and</strong> Social <strong>Psychology</strong>, 36, 306- 312.<br />
K<strong>and</strong>el, D. B., & Andrews, K. (1987). Processes of adolescent socializa-<br />
tion by parents <strong>and</strong> peers. International Journal of <strong>the</strong> Addictions,<br />
22, 319-342.<br />
Kaye, L. W., & Monk, A. ( 1991 ). Social relations <strong>in</strong> enriched hous<strong>in</strong>g<br />
for <strong>the</strong> aged: A case study. Journal of Hous<strong>in</strong>g for <strong>the</strong> Elderly, 9,<br />
111-126.<br />
Kl<strong>in</strong>ger, E. (1977). Mean<strong>in</strong>g <strong>and</strong> void: Inner experience <strong>and</strong> <strong>the</strong> <strong>in</strong>cen-<br />
tives <strong>in</strong> people's lives. M<strong>in</strong>neapolis: University of M<strong>in</strong>nesota Press.<br />
Kupersmidt, J. B., DeRosier, M. E., & Patterson, C. P. (1995). Similarity<br />
as <strong>the</strong> basis for children's friendships: The roles of sociometric status,<br />
aggressive <strong>and</strong> withdrawn behavior, academic achievement <strong>and</strong> demo-<br />
graphic characteristics. Journal of Social <strong>and</strong> Personal Relationships,<br />
12, 439-452.<br />
Kurdek, L. A., & S<strong>in</strong>clair, R. J. (1988). Adjustment of young adolescents<br />
<strong>in</strong> two-parent nuclear, stepfa<strong>the</strong>r, <strong>and</strong> mo<strong>the</strong>r-custody families. Journal<br />
of Consult<strong>in</strong>g <strong>and</strong> Cl<strong>in</strong>ical <strong>Psychology</strong>, 56, 91-96.<br />
Ladd, G. W. (1990). Hav<strong>in</strong>g friends, keep<strong>in</strong>g friends, mak<strong>in</strong>g friends,<br />
<strong>and</strong> be<strong>in</strong>g liked by peers <strong>in</strong> <strong>the</strong> classroom: Predictors of children's<br />
early school adjustment? Child Development, 61, 1081-1100.<br />
Ladd, G. W., Kochenderfer, B. J., & Coleman, C. C. (1996). Friendship<br />
quality as a predictor of young children's early school adjustment.<br />
Child Development, 67, 1103-1118.<br />
Lang, F. R., & Carstensen, L. L. (1994). Close emotional relationships<br />
<strong>in</strong> late life: Fur<strong>the</strong>r support for proactive ag<strong>in</strong>g <strong>in</strong> <strong>the</strong> social doma<strong>in</strong>.<br />
<strong>Psychology</strong> <strong>and</strong> Ag<strong>in</strong>g, 9, 315-324.<br />
Larson, R. (1978). Thirty years of research on <strong>the</strong> subjective well-be<strong>in</strong>g<br />
of older Americans. Journal of Gerontology, 33, 109-125.<br />
Larson, R., & Bradney, N. (1988). Precious moments with family mem-<br />
bers <strong>and</strong> friends. In R. M. Milardo (Ed.), Families <strong>and</strong> social networks<br />
(pp. 107-126). Newbury Park, CA: Sage.<br />
Larson, R., Zuzanek, J., & Mannell, R. (1985). Be<strong>in</strong>g alone versus<br />
be<strong>in</strong>g with people: Disengagement <strong>in</strong> <strong>the</strong> daily experience of older<br />
adults. Journal of Gerontology, 40, 375-381.<br />
Laursen, B., & Coll<strong>in</strong>s, W.A. (1994). Interpersonal conflict dur<strong>in</strong>g<br />
adolescence. Psychological Bullet<strong>in</strong>, 115, 197-209.<br />
Lazarsfeld, P. F., & Merton, R. K. (1954). Friendship as a social process:<br />
A substantive <strong>and</strong> methodological analysis. In M. Berger, T. Abel, &<br />
C. H. Page (Eels.), Freedom <strong>and</strong> control <strong>in</strong> modern society (pp. 18-<br />
66). Pr<strong>in</strong>ceton, NJ: Van Nostr<strong>and</strong>.<br />
Lebo, D. (1953). Some factors said to make for happ<strong>in</strong>ess <strong>in</strong> old age.<br />
Journal of Cl<strong>in</strong>ical <strong>Psychology</strong>, 9, 384-390.<br />
Livesley, W. J., & Bromley, B. D. (1973). Person perception <strong>in</strong> child-<br />
hood <strong>and</strong> adolescence. London: Wiley.<br />
Lowenthal, M. F., & Haven, C. (1968). Interaction as adaptation: Inti-<br />
macy as a critical variable. American Sociological Review, 33, 20-<br />
30.<br />
Lowenthal, M. E, Thurnher, M., & Chiriboga, D. (Eds.). (1975). Four<br />
stages of life: A comparative study of women <strong>and</strong> men fac<strong>in</strong>g transi-<br />
tions. San Francisco: Jossey-Bass.<br />
Maccoby, E. E. (1990). Gender <strong>and</strong> relationships: A developmental ac-<br />
count. American Psychologist, 45, 513-520.<br />
Magnusson, D., Statt<strong>in</strong>, H., & Allen, V. L. (1985). Biological maturation<br />
<strong>and</strong> social development: A longitud<strong>in</strong>al study of some adjustment<br />
processes from mid-adolescence to adulthood. Journal of Youth <strong>and</strong><br />
Adolescence, 14, 267-283.<br />
Mannar<strong>in</strong>o, A. P. (1978). Friendship patterns <strong>and</strong> self-concept develop-<br />
ment <strong>in</strong> preadolescent males. Journal of Genetic <strong>Psychology</strong>, 133,<br />
105-110.<br />
FRIENDSHIPS IN THE LIFE COURSE 369<br />
Mat<strong>the</strong>ws, S. H. (1986). <strong>Friendships</strong> through <strong>the</strong> life course. Beverly<br />
Hills, CA: Sage.<br />
Mat<strong>the</strong>ws, S. H. (1995). <strong>Friendships</strong> <strong>in</strong> old age. In N. Vanzetti & S.<br />
Duck (Eds.), A lifetime of relationships (pp. 406-430). San Fran-<br />
cisco: Brooks/Cole.<br />
McC<strong>and</strong>less, B. R., & Hoyt, J. M. ( 1961 ). Sex, ethnicity <strong>and</strong> play prefer-<br />
ences of preschool children. Journal of Abnormal <strong>and</strong> Social Psychol-<br />
ogy, 62, 683-685.<br />
McGuire, K. D., & Weisz, J. R. (1982). Social cognition <strong>and</strong> behavior<br />
correlates of preadolescent chumship. Child Development, 53, 1478-<br />
1484.<br />
Mills, J., & Clark, M. S. (1982). Exchange <strong>and</strong> communal relationships.<br />
Review of Personality <strong>and</strong> Social <strong>Psychology</strong>, 36, 72-81.<br />
Monroe, W. S. (1898). Social consciousness <strong>in</strong> children. Psychological<br />
Review, 5, 68-70.<br />
Nahemow, L., & Lawton, M. P. (1975). Similarity <strong>and</strong> prop<strong>in</strong>quity <strong>in</strong><br />
friendship formation. Journal of Personality <strong>and</strong> Social <strong>Psychology</strong>,<br />
32, 205-213.<br />
Neeman, J. D., Hubbard, J., & Kojet<strong>in</strong>, B. A. (1991, April). Cont<strong>in</strong>uity<br />
<strong>in</strong> quality of friendships <strong>and</strong> romantic relationships from childhood<br />
to adolescence. Poster session presented at <strong>the</strong> biennial meet<strong>in</strong>gs of<br />
<strong>the</strong> Society for Research <strong>in</strong> Child Development, Seattle, WA.<br />
Nelson, J., & Aboud, F.E. (1985). The resolution of social conflict<br />
between friends. Child Development, 56, 1009-1017.<br />
Newcomb, A. F., & Bagwell, C. (1995). Children's friendship relations:<br />
A meta-analytic review. Psychological Bullet<strong>in</strong>, 117, 306-347.<br />
Newcomb, T. M. (1961). The acqua<strong>in</strong>tance process. New York: Holt,<br />
R<strong>in</strong>ehart & W<strong>in</strong>ston.<br />
O'Connor, P. (1993). Same-gender <strong>and</strong> cross-gender friendships among<br />
<strong>the</strong> frail elderly. Gerontologist, 33, 24-30.<br />
Pagel, M. D., Erdly, W. W., & Becker, J. (1987). Social networks: We<br />
get by with (<strong>and</strong> <strong>in</strong> spite of) a little help from our friends. Journal<br />
of Personality <strong>and</strong> Social <strong>Psychology</strong>, 53, 793-804.<br />
Pap<strong>in</strong>i, D. R., Farmer, E F., Clark, S. M., Micke, J. C., & Barnett, J. K.<br />
(1990). Early adolescent age <strong>and</strong> gender differences <strong>in</strong> patterns of<br />
emotional self-disclosure to parents <strong>and</strong> friends. Adolescence, 25,<br />
959-976.<br />
Parker, J. G., & Asher, S. R. (1993). Friendship <strong>and</strong> friendship quality<br />
<strong>in</strong> middle childhood: L<strong>in</strong>ks with peer group acceptance <strong>and</strong> feel<strong>in</strong>gs<br />
of lonel<strong>in</strong>ess <strong>and</strong> social dissatisfaction. Developmental <strong>Psychology</strong>,<br />
29, 611-621.<br />
Rawl<strong>in</strong>s, W.K. (1992). Friendship matters. New York: Ald<strong>in</strong>e de<br />
Gruyter.<br />
Rep<strong>in</strong>ski, D.J. (1993). Adolescents" close relationships with parents<br />
<strong>and</strong> friends. Unpublished doctoral dissertation, University of M<strong>in</strong>ne-<br />
sota, Tw<strong>in</strong> Cities, Institute of Child Development, M<strong>in</strong>neapolis.<br />
Roberto, K.A., & Scott, J.P. (1986). <strong>Friendships</strong> of older men <strong>and</strong><br />
women: Exchange patterns <strong>and</strong> satisfaction. <strong>Psychology</strong> <strong>and</strong> Ag<strong>in</strong>g,<br />
1, 103-109.<br />
Rogoff, B. (1990). Apprenticeship <strong>in</strong> th<strong>in</strong>k<strong>in</strong>g. New York: Oxford Uni-<br />
versity Press.<br />
Rook, K. S. (1984). The negative side of social <strong>in</strong>teraction: Impact on<br />
psychological well-be<strong>in</strong>g. Journal of Personality <strong>and</strong> Social Psychol-<br />
ogy, 46, 1097-1108.<br />
Rosenbaum, M. E. (1986). The repulsion hypo<strong>the</strong>sis: On <strong>the</strong> nondevel-<br />
opment of relationships. Journal of Personality <strong>and</strong> Social Psychol-<br />
ogy, 51, 1156-1166.<br />
Rub<strong>in</strong>, K. H., Lynch, D., Coplan, R., Rose-Krasnor, L., & Booth, C. L.<br />
(1994). "Birds of a fea<strong>the</strong>r . . . ": Behavioral concordances <strong>and</strong><br />
preferential personal attraction <strong>in</strong> children. Child Development, 65,<br />
1778-1785.<br />
Rutter, M., & Garmezy, N. (1983). Developmental psychopathology. In<br />
P. H. Mussen (Series Ed.) & E. M. He<strong>the</strong>r<strong>in</strong>gton (Vol. Ed.), H<strong>and</strong>book<br />
of child psychology. Vol. 4: Socialization, personality, <strong>and</strong> social de-<br />
velopment (pp. 775-911 ). New York: Wiley.
370 HARTUP AND STEVENS<br />
Schutze, Y., & Lang, E R. (1993). Freundschaft, alter und geschlecht<br />
[Friendship, age <strong>and</strong> gender]. Zeitschriftfur Soziologie, 22, 209-220,<br />
Schwartz, J. C. (1972). Effects of peer familiarity on <strong>the</strong> behavior of<br />
preschoolers <strong>in</strong> a novel situation. Journal of Personality <strong>and</strong> Social<br />
<strong>Psychology</strong>, 24, 276-284.<br />
Sears, R. R., Alpert, R., & Ran, L. (1964). Identification <strong>and</strong> child-<br />
rear<strong>in</strong>g. Stanford, CA: Stanford University Press.<br />
Selman, R. L. (1980). The growth of <strong>in</strong>terpersonal underst<strong>and</strong><strong>in</strong>g. New<br />
York: Academic Press.<br />
Sharabany, R., Gershoni, R., & Hofman, J. E. (1981). Girlfriend, boy-<br />
friend: Age <strong>and</strong> sex differences <strong>in</strong> <strong>in</strong>timate friendship. Developmental<br />
<strong>Psychology</strong>, 17, 800-808.<br />
Shea, L., Thompson, L., & Blieszner, R. (1988). Resources <strong>in</strong> older<br />
adults' old <strong>and</strong> new friendships. Journal of Social <strong>and</strong> Personal Rela-<br />
tionships, 5, 83-96.<br />
Shulman, S. (1993). Close friendships <strong>in</strong> early <strong>and</strong> middle adolescence:<br />
Typology <strong>and</strong> friendship reason<strong>in</strong>g. In B. Laursen (Ed.), Close friend-<br />
ships <strong>in</strong> adolescence (pp. 55-72). San Francisco: Jossey-Bass.<br />
Simmons, R.G., Burgeson, R., & Reef, M.J. (1988). Cumulative<br />
change at entry to adolescence. In M. Gunnar & W. A. Coll<strong>in</strong>s (Eds.),<br />
M<strong>in</strong>nesota Symposia on Child <strong>Psychology</strong> (Vol. 21, pp. 123-150).<br />
Hillsdale, NJ: Erlbaum.<br />
Skolnick, A. (1986). Early attachment <strong>and</strong> personal relationships across<br />
<strong>the</strong> life course. In D. Magnusson, H. Statt<strong>in</strong>, & V. L. Allen (Eds.),<br />
<strong>Life</strong> span development <strong>and</strong> behavior (Vol. 7, pp. 173-206). New<br />
York: Academic Press.<br />
Sroufe, L. A., & Fleeson, J. (1986). Attachment <strong>and</strong> <strong>the</strong> construction<br />
of relationships. In W. W. Hartup & Z. Rub<strong>in</strong> (Eds.), Relationships<br />
<strong>and</strong> development (pp. 57-71 ). Hillsdale, NJ: Erlbaum.<br />
Stacey-Konnert, C., & Pynoss, J. (1992). Friendship <strong>and</strong> social networks<br />
<strong>in</strong> a cont<strong>in</strong>u<strong>in</strong>g care retirement community. Journal of Applied Geron-<br />
tology, 11, 298-313.<br />
Staud<strong>in</strong>ger, U. M., Marsiske, M., & Baltes, P. B. (1995). Resilience <strong>and</strong><br />
reserve capacity <strong>in</strong> later adulthood: Potentials <strong>and</strong> limits of develop-<br />
ment across <strong>the</strong> life span. In D. Cicchetti & D.J. Cohen (Eds.),<br />
Developmental psychopathology. Vol. 2: Risk, disorder, <strong>and</strong> adapta-<br />
tion (pp. 801-847). New York: Wiley.<br />
Stephens, M. A.P., K<strong>in</strong>ney, J.M., Ritchie, S.W., & Norris, V.K.<br />
(1987). Social networks as assets <strong>and</strong> liabilities <strong>in</strong> recovery from<br />
stroke by geriatric patients. <strong>Psychology</strong> <strong>and</strong> Ag<strong>in</strong>g, 2, 125-129.<br />
Stevens, N. (1989). Well-be<strong>in</strong>g <strong>in</strong> widowhood: A question of balance.<br />
Doctoral dissertation, University of Nijmegen, Department of Psycho-<br />
gerontology, Nijmegen, The Ne<strong>the</strong>rl<strong>and</strong>s.<br />
Stevens, N. (1995). Gender <strong>and</strong> adaptation to widowhood <strong>in</strong> later life.<br />
Age<strong>in</strong>g <strong>and</strong> Society, 15, 37-58.<br />
Sullivan, H.S. (1953). The <strong>in</strong>terpersonal <strong>the</strong>ory of psychiatry. New<br />
York: Norton.<br />
Thorne, B. B. (1986). Boys <strong>and</strong> girls toge<strong>the</strong>r.., but mostly apart. In<br />
W. W. Hartup & Z. Rub<strong>in</strong> (Eds.), Relationships <strong>and</strong> development (pp.<br />
167-184). Hillsdale, NJ: Erlbaum.<br />
Tolson, J. M., & Urberg, K. A. (1993). Similarity between adolescent<br />
best friends. Journal of Adolescent Research, 8, 274-288.<br />
Tomblom, K. Y., & Fredholm, E. M. (1984). Attribution of friendship:<br />
The <strong>in</strong>fluence of <strong>the</strong> nature <strong>and</strong> comparability of resources given <strong>and</strong><br />
received. Social <strong>Psychology</strong> Quarterly, 47, 50-61.<br />
Van der L<strong>in</strong>den, E J., & Dijkman, T. A. (1989). Jong zijn en volwassen<br />
worden <strong>in</strong> Nederl<strong>and</strong> [Be<strong>in</strong>g young <strong>and</strong> becom<strong>in</strong>g an adult <strong>in</strong> The<br />
Ne<strong>the</strong>rl<strong>and</strong>s]. Nijmegen, The Ne<strong>the</strong>rl<strong>and</strong>s: Hoogveld.<br />
Weiss, L., & Lowenthal, M.E (1975). <strong>Life</strong>-course perspectives on<br />
friendship. In M. E Lowenthal, M. Thurnher, & D. Chiriboga (Eds.),<br />
Four stages of life: A comparative study of women <strong>and</strong> men fac<strong>in</strong>g<br />
transitions (pp. 48-61). San Francisco: Jossey-Bass.<br />
Weiss, R. S. (1973). Lonel<strong>in</strong>ess. Cambridge, MA: MIT Press.<br />
Weiss, R. S. (1975). Marital separation. New York: Basic Books.<br />
Weiss, R. S. (1986). Cont<strong>in</strong>uities <strong>and</strong> transformations <strong>in</strong> social relation-<br />
ships from childhood to adulthood. In W.W. Hartup & Z. Rub<strong>in</strong><br />
(Eds.), Relationships <strong>and</strong> development (pp. 95-110). Hillsdale, NJ:<br />
Erlbaum.<br />
Werner, C., & Parmelee, P. (1979). Similarity of activity preferences<br />
among friends: Those who play toge<strong>the</strong>r stay toge<strong>the</strong>r. Social Psychol-<br />
ogy Quarterly, 42, 62-66.<br />
Wilcox, B. L. ( 1981 ). Social support <strong>in</strong> adjust<strong>in</strong>g to marital disruption:<br />
A network analysis. In B.H. Gottlieb (Ed.), Social networks <strong>and</strong><br />
social support (pp. 97-115). Beverly Hills, CA: Sage.<br />
W<strong>in</strong>dle, M. (1992). A longitud<strong>in</strong>al study of stress buffer<strong>in</strong>g for adoles-<br />
cent problem behaviors. Developmental <strong>Psychology</strong>, 28, 522-530.<br />
W<strong>in</strong>dle, M. (1994). A study of friendship characteristics <strong>and</strong> problem<br />
behaviors among middle adolescents. Child Development, 65, 1764-<br />
1777.<br />
W<strong>in</strong>stead, B. A., Derlega, V. J., & Montgomery, M. J. (1995). The qual-<br />
ity of friendships at work <strong>and</strong> job satisfaction. Journal of Social <strong>and</strong><br />
Personal Relationships, 12, 199-215.<br />
Wister, A. (1990). Liv<strong>in</strong>g arrangements <strong>and</strong> <strong>in</strong>formal social support<br />
among <strong>the</strong> elderly. Journal of Hous<strong>in</strong>g for <strong>the</strong> Elderly, 6, 33-43.<br />
Wright, P.H. (1989). Gender differences <strong>in</strong> adults' same- <strong>and</strong> cross-<br />
gender friendships. In R.G. Adams & R. Blieszner (Eds.), Older<br />
adult friendship (pp. 197-221 ). Newbury Park, CA: Sage.<br />
Young, J.E. (1986). A cognitive-behavioral approach to friendship<br />
disorders. In V. J. Derlega & B. A. W<strong>in</strong>stead (Eds.), Friendship <strong>and</strong><br />
social <strong>in</strong>teraction (pp. 246-276). New York: Spr<strong>in</strong>ger Verlag.<br />
Youniss, J. (1980). Parents <strong>and</strong> peers <strong>in</strong> social development: A Sulli-<br />
van-Piaget perspective. Chicago, IL: University of Chicago Press.<br />
Received January 25, 1996<br />
Revision received July 25, 1996<br />
Accepted July 29, 1996 •