Professional Documents
Culture Documents
Society For Research in Child Development, Wiley Child Development
Society For Research in Child Development, Wiley Child Development
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/1130451?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Society for Research in Child Development, Wiley are collaborating with JSTOR to
digitize, preserve and extend access to Child Development
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
Changes in Friendship during a School Year:
Effects on Children's and Adolescents'
Impressions of Friendship and Sharing
with Friends
Thomas J. Berndt
Purdue University
Jacquelyn A. Hawkins
Yale University
Sally G. Hoyle
Ohio State University
BERNDT, THOMAS J.; HAWKINS, JACQUELYN A.; and HOYLE, SALLY G. Changes in Friendship during
a School Year: Effects on Children's and Adolescents' Impressions of Friendship and Sharing with
Friends. CHILD DEVELOPMENT, 1986, 57, 1284-1297. To examine the changes in friendship during a
school year, fourth and eighth graders were paired with a close friend in the fall of the year. During
individual interviews, the children were asked a standard set of open- and closed-ended questions
about these specific friendships. In another session, each pair was observed as they worked on a task
that allowed them to share or to compete with each other. The same pairs of children were inter-
viewed and observed again in the spring. Roughly two-thirds of the children remained close friends
with their partners throughout the year. The children who did not remain close friends responded
less positively in the spring than in the fall when interviewed about their friendships. The responses
of children with stable and unstable friendships could be distinguished even in the fall, however,
which suggests that friendship stability can be predicted from children's statements about a friend-
ship. Observations of the children's interactions confirmed a hypothesis that the preference of
friends for equal sharing over competition increases between middle childhood and early adoles-
cence. In the spring, close friends shared less at fourth grade but more at eighth grade than children
who were no longer close friends. The relevance of the findings for theories of friendship develop-
ment is discussed.
Most friendships among children and ships and their preference for sharing over
adolescents show a clear pattern of changes
competition with friends. A second purpose
over time. Friendships form when two chil-of the study was to examine the differences
dren become acquainted with each other; between friendships in middle childhood and
early adolescence, using multiple methods
they are maintained for some period of time;
and measures.
they wane or end when, for various reasons,
the friends have fewer contacts with each
other or they form closer friendships with In many previous studies, sociometric
other children. Recent research has focused
methods were used to assess friendship sta-
on the features of existing friendships (see
bility and change. After reviewing this litera-
ture, Epstein (1986) concluded that friend-
Hartup, 1983) or, less often, on their formation
(Gottman, 1983). Investigations of multiple
ship stability increases regularly with age but
aspects of the changes in friendships over
is fairly low even during adolescence. By con-
time are relatively rare. One purpose of thistrast, Berndt (1982) concluded that friendship
study was to examine the extent to which the stability is fairly high during childhood and
changes in friendship during a school year in-adolescence and does not change markedly
fluence children's impressions of their friend-
with age. These reviewers apparently came to
The research reported in this paper was supported in part by National Science Foundation
grant BNS78-24157 and by a grant from the Spencer Foundation. The authors wish to thank Ann
Verrilli and Bridgett Perry for their assistance in coding and analyzing the data, and David Rowe for
his comments on an earlier version of the paper. Requests for reprints should be sent to Thomas J.
Berndt, Department of Psychological Sciences, Purdue University, West Lafayette, IN 47907.
[Child Development, 1986, 57, 1284-1297. ? 1986 by the Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/86/5705-0013$01.00]
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
Berndt, Hawkins, and Hoyle 1285
different conclusions because they made dif- macy of actual friendships increases with age.
ferent assumptions about the methodological Although this conclusion is reasonable, data
adequacy of several early studies (e.g., Hor- on children's impressions of their own friend-
rocks & Buker, 1951). All children in these ships would provide a better test of the hy-
studies were told to name three best friends, pothesis.
even though they may have had more or
fewer than that number. The error introduced In the present study, children were asked
by requiring a fixed number of best-friend a series of open-ended questions about one of
nominations could lead to underestimates of their closest friendships. Children were also
asked to rate various features of their friend-
friendship stability, because children with
more than three best friends would not neces- ships on structured response scales (see Fur-
sarily nominate the same ones on two occa- man & Bierman, 1984). The age changes in
sions. In other words, changes due to the children's responses were expected to be
unreliability of children's nominations could similar to those in previous studies of friend-
be mistaken for, or confounded with, real ship conceptions. In addition, the responses
changes in friendship. were expected to reflect the changes in
friendships during the school year. When a
On the other hand, recent findings on friendship weakened, as judged by changes
friendship stability cannot be completely ex- in best-friend nominations and ratings of lik-
plained in terms of the unreliability of mea- ing, positive comments about the friendship
sures. The changes in best-friend nomina- were expected to decrease and negative com-
tions are greater over a 6-month interval than ments were expected to increase. Ratings of
a 1-month interval (Bukowski & Newcomb, the friendship were also expected to become
1984), as would be expected if they indexed less positive. One goal of the study was to
real changes in friendship. Moreover, recip- determine if certain types of positive and
rocal nominations are more stable than unilat- negative comments, or ratings of certain
eral nominations (Bukowski & Newcomb, features, were especially sensitive to the
1984; Kandel, 1978). If nominations were un- changes in friendships.
reliable, however, the likelihood of two chil-
dren repeating the same (reciprocal) choice A third method for exploring changes in
would be lower than the likelihood that a sin- friendships is to observe friends' interactions.
Most previous studies of friends' interactions
gle nomination would be repeated.
were designed to test the common-sense hy-
In the present study, friendships were pothesis that friends share with each other
defined by a combination of best-friend nomi- more than nonfriends do. The results of sev-
nations and ratings of liking. In past research eral studies supported the hypothesis but, in
with the same two criteria (Berndt, 1981a), others, friends shared significantly less than
most friendships among first and fourth grad- nonfriends (see Hartup, 1983; Radke-Yarrow,
ers were stable over several months during a Zahn-Waxler, & Chapman, 1983). Berndt
school year. Among the fourth and eighth (1982) suggested that children share less with
graders included in this study, friendship sta- friends than nonfriends in a specific situa-
bility was also expected to be high. tional context, namely, when they regard
A second method for examining changes themselves as in a competition that they
in friendship is to ask children to describe might lose by sharing (see also Tesser, 1984;
their own friendships. A variant of this Tesser & Smith, 1980). Children view them-
method was used in many previous studies to selves as equal to their friends and, therefore,
explore children's conceptions of friendship they are especially concerned about their per-
formance relative to that of their friends. Al-
(Berndt, 1982; Hartup, 1983; Selman, 1981).
Children usually were asked open-minded though friends can often maintain their equal-
questions, such as "What is a friend?" ity by sharing with each other (Youniss, 1980),
there are some situations in which sharing
(Youniss, 1980). Their responses were coded
with a friend increases the risk of performing
into categories for specific features of friend-
ship. Age differences in category usage wereless well than the friend and so seeming in-
typically interpreted as evidence aboutferior the to him or her. In such situations, chil-
dren may choose to compete with the friend
development of friendship. For example,
many researchers reported an increase with rather than to share equally.
age in comments about intimate self- Sullivan (1953) proposed that friendships
disclosure or mutual understanding between
become less competitive and more mutually
friends (Berndt, 1982). This change in friend-
responsive as children enter adolescence.
ship conceptions has been taken as support
Drawing on Sullivan's work, Berndt (1982)
for Sullivan's (1953) hypothesis that thehypothesized
inti- that the development of mutual
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
1286 Child Development
SAll but 11 children who received their parents' permission to participate in the study had a
classmate in the sample who met the criteria for a close friend. In other words, 88% of the children
in the potential sample were used in the final sample. The 11 children who did not have a close
friend in the sample may have had close friends who were not participating in the study. Never-
theless, to avoid excluding these children they were assigned a partner and given the same proce-
dures as the other children, but their data were not used.
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
Berndt, Hawkins, and Hoyle 1287
pair. Pairs with higher scores were judged tome when I'm in a fight" or "she'll never leave
be closer friends. me for somebody else"); and (h) similarity in
activities, interests, values, or personality
Interview on friendship: Open-ended("we both like baseball").
questions.-In the fall, shortly after the pairs
of friends were identified, all children were With two exceptions, the negative catego-
interviewed individually. The children first ries were the converse of the positive ones.
were asked a series of open-ended questions
A single category for hostile behavior was
about their partners in the study and their created that corresponded to the two catego-
friendships with those partners. They were ries for prosocial behavior and the absence of
asked three questions about their partners: (a)conflicts. A category for "unstimulating play"
what kind of person he or she was; (b) whatwas not used, because children usually ex-
things were especially good about him or her pressed this idea by criticizing their partner's
(or made him or her fun to be with); and (c)personality (e.g., saying he or she was "bor-
what things they or other kids disliked abouting") rather than talking about the friendship
him or her. Next, they were asked three ques- itself.
tions about the friendship: (a) how they could
tell that their partner was their friend; (b) To reduce the effects of individual differ-
what they liked best about being friends with ences in verbal fluency and the tendency to
him or her; and (c) what they did not like give repetitive answers, present-absent scor-
about being friends with him or her. After theing was done for each question. That is, chil-
children's first response to each question,dren received scores of 1 or 0 for each cate-
they were asked a standard probe question gory for each of the six questions about their
(e.g., "What else can you tell me?"). Probing partners and their friendships. Therefore, the
continued until children gave at least three potential range for final scores was 0 to 6. Two
persons independently coded the responses
responses or said they could not think of any-
thing else. All of the children's responses to the open-ended questions from 20 inter-
were tape-recorded and then transcribed. views. Their agreement on the appropriate
category for each response was 82%.
In the spring, all children were inter-
viewed again, using the same questions about Interview on friendship: Closed-ended
questions.-The interview continued with a
the same partners as in the fall. At the begin-
ning of the spring interview, however, chil-series of closed-ended questions about five
dren were asked if their partners were (still)features of friendship that were emphasized
best friends. If they said no, they were askedin past research on friendship conceptions: (a)
first to explain why the partner was no longerprosocial behavior ("How often does he/she
a friend, and then whether there were any help you when you have a lot to do or you
positive features of the current relationship.can't do something alone?"); (b) aggressive
behavior ("How often does he/she fight or ar-
After an examination of the responses
gue with you?"); (c) intimacy ("How often
and a review of previous coding systems (e.g.,
does he/she talk with you about problems that
Berndt, 1981c; Bigelow & La Gaipa, 1980;
are important to him/her?"); (d) loyalty ("How
"Youniss, 1980), categories for the positive and
often does he/she take your side in a fight or
negative features of the children's friendship
argument?"); and (e) similarity ("How often
were defined.2 There were eight categories
do you and he/she feel differently or think
for positive comments and six categories for
differently about something?").
negative comments. The positive categories
referred to the friends' (a) liking or friendship Two close-ended questions were in-
("we're best friends"); (b) frequency of in- cluded for each feature of friendship. Chil-
teraction ("I go over to her house a lot"); (c) dren made their responses on a five-point
how stimulating they found their interactions scale ranging from never (0) to very often (4).
("we have a good time together"); (d) proso- As the examples illustrate, some questions
cial behavior (e.g., "we help each other"); (e)
were stated positively and others were stated
the absence of conflicts between them ("we negatively. When the data were coded, the
get along"); (f) intimacy ("he tells me when ratings for the negatively worded questions
he has problems"); (g) loyalty or faithfulnessand the questions on aggressive behavior
when around other people ("he'll stick up for were reversed. Then the ratings for the two
2 When children commented about a partner's personality, rather than their friendship with the
partner, the comments were coded with a different set of categories. Children were also asked a
series of closed-ended questions about their partner's personality. These data are not presented
because their relevance to this study is limited.
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
1288 Child Development
3 In the fall, the children received small toys as prizes after all the interviews and behavioral
observations were completed. The child in each pair who received more nickels during the task was
given a prize slightly larger or more expensive-looking than the other child. In the spring, all
children received the same prize.
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
Berndt, Hawkins, and Hoyle 1289
asked to guess their partner's primary and than unilateral nominations in the fall (78%
secondary motives. At the conclusion of thevs. 46%, p < .05).
task, the children were asked to rank the mo-
tives again, indicating what they had tried toChildren's Reports on Their Friendships
The measures derived from the inter-
do and what they thought their partner had
tried to do. views with children were used as dependent
variables in a series of repeated-measures
The primary motives that children re-analyses of variance. The between-subject
factors were grade, sex, and friendship stabil-
ported for themselves and their partner were
scored 2, the secondary motives were scoredity. The last factor contrasted relatively stable
1, and the other two motives were scored 0.friendships (i.e., pairs that met the criteria for
Because children reported similar motivesfriendship in both the fall and the spring)
before and after the task, mean scores for bothwith relatively unstable friendships (i.e., pairs
times were used in all analyses. that met the friendship criteria in the fall but
not in the spring). Time was a within-subject
On the posttask questionnaire, children factor.
were also asked a series of questions about
their reactions to their own and their partner'sA univariate analysis was done for the
ratings of interaction frequency. Separate
behavior. Using a five-point scale, children
multivariate analyses followed by univariate
indicated whether they thought their partner
should have shared more than he or she did,analyses were done for the categories of posi-
whether they should have shared more, tive comments, the categories of negative
whether their partner took advantage of them, comments, and the ratings on the five friend-
and whether they took advantage of him orship features. The multivariate approach was
her. These four questions were strongly cor- chosen because it yields significance levels
related (mean r = .54), so they were com- that are less affected by violations of assump-
bined. Higher mean ratings indicate greater tions than is a univariate analysis with more
dissatisfaction with the outcome of the task. than two levels for any within-subject factor
(McCall & Applebaum, 1973). The multi-
Results variate analysis can be less powerful, how-
ever. To take account of this potential loss of
The Stability of Friendship power, all multivariate effects with p < .10
The stability of the particular friendships
are presented, but the results of univariate
identified in the fall of the school year wastests are presented only when p < .05. Table
evaluated in two ways. First, because the 1 summarizes the significant main effects and
friendship scores were continuous variables, interactions in these analyses.
they served as particularly sensitive measures
of friendship change. These scores were used Effects of friendship stability and
as the dependent variables in a repeated- time.-A main effect of friendship stability
was found in all analyses. This main effect
measures analysis of variance with grade, sex,
and no interactions were found for comments
and time (fall or spring) as factors. The only
about liking and interaction frequency (Table
significant effect was for time, F(1,41) = 13.1,
p < .001. As expected, friendship scores were2). Children who remained close friends
lower in the spring than in the fall (M's throughout
= the school year commented more
8.42 and 9.36, respectively), indicating thatoften on the liking for each other and the fre-
some fall friendships had weakened or ended quency of their interactions than children
by the spring. who were close friends only in the fall.
Second, the original criteria for a close Main effects for friendship stability and
friendship, based on best-friend nominations
no interactions were also found for negative
and ratings of liking, were used to judge comments about intimacy and ratings of inti-
which pairs of children were still close macy (see Table 2). Children in unstable
friends in the spring. Overall, 69% of the pairs friendships made more comments about a
that were formed in the fall still met the crite- lack in intimacy in their relationships than
ria for a close friendship in the spring. The children in stable friendships. Similarly, chil-
percentage of pairs meeting the criteria was dren in unstable friendships rated their rela-
somewhat lower for eighth-grade males (56%) tionships as less intimate than did children in
than for eighth-grade females (69%), or for stable friendships. The absence of interac-
fourth-grade males (77%) and females (70%), tions with time implies that the differences
but none of the differences among these between stable and unstable friendships were
groups was significant (p's > .20). Stability roughly as great in the fall as in the spring.
was higher for pairs that had reciprocal rather Consequently, the children's fall reports
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
1290 Child Development
TABLE 1
MEASURES
INTERACTION
Positive Negative Feature FREQUENCY
Comments Comments Ratings RATINGS
F p F p F p F p
Main effects:
Friendship
stability ............ 2.88 <.01 3.07 <.01 2.97 <.025 45.39 <.001
Grade .............. 2.63 <.025 1.33 N.S. 2.01 <.09 1.51 N.S.
Sex ............... 2.04 <.06 2.37 <.05 2.14 <.07 <1 N.S.
Time ............... 1.03 N.S. 3.60 <.01 3.04 <.025 24.92 <.001
Two-way interactions:
Friendship stability
x grade ............ 1.06 N.S. <1 N.S. 2.62 <.05 12.76 <.001
Friendship stability
x time ............. 1.95 <.07 2.09 <.07 2.08 <.08 7.41 <.01
Three-way interactions:
Friendship stability
x grade x sex ...... 1.24 N.S. 1.17 N.S. 1.01 N.S. 5.35 <.025
Friendship stability
x sex x time ....... 1.29 N.S. <1 N.S. 2.51 <.05 <1 N.S.
Grade x sex
x time ............. 1.11 N.S. 1.67 N.S. 3.17 <.025 <1 N.S.
NOTE.--Multivariate F ratios are reported, except in the analysis of interaction
(8,75) for positive comments, (6,77) for negative comments, (5,78) for feature ratin
quency ratings.
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
Berndt, Hawkins, and Hoyle 1291
TABLE 2
MEAN SCORES FOR THE INTERVIEW RESPONSES CONCERNING STABLE AND UNSTABLE FRIENDSHIPS
AT EACH TIME
in the
about the intimacy of their friendships spring than in the fall, M's = 2.57 and
than
fourth graders, M's = .99 and .28, 3.17.
p < .001.
The comparable ratings by boys did not
Eighth graders also commented more often
change significantly over time, M's = 2.99
and they
than fourth graders on the stimulation 2.66, respectively. This result is consis-
derived from friends, M's = 2.02 andtent with
1.52, p previous evidence that girls are
especially
< .05. By contrast, eighth graders gave lower concerned about the loyalty of
ratings than fourth graders for theirfriends
friends'(see Coleman, 1980).
prosocial behavior, M's = 2.51 and 2.84, p <
For ratings of interaction frequency,
.05; and their similarity to their friends, M's =
there was a friendship stability x grade in-
2.21 and 2.56, p < .01.
teraction, and an interaction of these factors
Girls commented more often on intimacy
with sex. Pairwise comparisons for the three-
in friendship than boys did, M's =way
.85 interaction
and revealed no significant ef-
.42, p < .025. Girls also referred more fects,
often however.
to The same was true for the
their liking or friendship with their friendship
partners, stability x grade interaction for
M's = .90 and .56, p < .05. In addition, girls
feature ratings. Tests for the grade x sex x
expressed more concern about the disloyalty
time interaction for feature ratings yielded
or unfaithfulness of friends than boys uninterpretable
did, M's results.
= 1.30 and .72, p < .01. Finally, girls gave
Relations among the interview mea-
higher ratings for their friends' prosocial be-
sures.-Scores for the overall frequencies of
havior than boys did, M's = 2.88 and 2.47, p
< .01. positive and negative comments, and the dif-
ference in the frequency of positive and nega-
tive comments, were correlated with mean
Other effects.-The friendship stability
x sex x time interaction was significantratings
for for all the closed-ended questions, in-
feature ratings. A significant univariatecluding
in- those for interaction frequency.
teraction was found for loyalty ratings. Girls
These aggregate scores were expected to pro-
vide the best estimates of the relations be-
with unstable friendships rated loyalty lower
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
1292 Child Development
TABLE 3
Total positive
comments ....... ... .41** .80*** .26*
Total negative
comments ....... .27* ... -.21* - .08
Positive-negative
comments........82*** - .32**.33***
Mean ratings ....... .45*** -.25* .59***
NOTE.-The correlations for the fall are above the diagonal. The correlations
spring are below the diagonal.
* p < .05.
** p < .01.
*** p < .001.
tween the measures derived from the open- friends shared more than eighth graders who
and closed-ended questions (see Rushton, were no longer close friends, M's = 334 ver-
Brainerd, & Pressley, 1983). sus 300 sec. A planned comparison estab-
lished the significance of these mean differ-
The intercorrelations of the aggregate
scores are shown in Table 3. The children ences, t(41) = 2.04, p < .05. Moreover, the
who made more positive comments and fewer
mean score that was closest to equal sharing
negative comments on the open-ended ques-
(360 sec) was, as expected, that for eighth
graders paired with friends. A separate analy-
tions gave higher ratings on the closed-ended
sis of the data showed neither a main effect
questions. Children who made more positive
for sex nor any interactions with sex (p's >
comments also made more negative com-
.20).
ments, however, which suggests that children
with closer friendships responded more fully Motives and reactions to the spring shar-
to both the positive and the negative ques-ing task.-To determine if the grade differ-
tions about friendship. The correlationences forin sharing reflected differences in mo-
positive and negative comments was weaker tives, children's reports on their own motives
in the spring than in the fall, and the correla-
and their attributions about their partner's
tions of those scores with mean ratings were
motives were analyzed. Attributions that the
correspondingly higher. Still higher correla-
partner tried for equality in rewards showed a
tions were found for mean ratings and scores
marginally significant grade x friendship in-
for the difference between the frequencies of
teraction, F(1,82) = 3.58, p < .07. Eighth
positive and negative comments, especially in who were still friends assumed that
graders
the spring when the range in responses was their partner tried for equality more often
greater. than eighth graders who were no longer
Sharing between Friends friends (M's = 1.41 and .91). Fourth graders
who were still friends assumed that their part-
Based on the general hypothesis about a
ner tried for equality slightly less often (M =
change with age in friends' preferences for
equal sharing over competition, a specific pre-
1.22) than did fourth graders who were no
diction was made about the duration of shar- longer friends (M = 1.29). When tested by a
ing in the spring by children who had or had planned comparison like that for actual shar-
not remained close friends as defined by the ing, these differences in attributions were
original criteria.4 As predicted, fourth graders
significant, t(86) = 2.11, p < .05.
who had remained close friends shared less Children's reports on their own motives
with their partners than fourth graders whose for equality varied only with friendship,
friendships had weakened, M's = 261 versusF(1,82) = 5.82, p < .05; and sex, F(1,82) =
317 sec; eighth graders who were still close
4.43, p < .05. Children who had remained
4 Sharing in the fall did not vary significantly with sex, but fourth graders tended to share less
than eighth graders, F(1,41) = 3.98, p < .06, M's = 280 and 328 sec, respectively.
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
Berndt, Hawkins, and Hoyle 1293
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
1294 Child Development
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
Berndt, Hawkins, and Hoyle 1295
in friendships over time. On a more specific tion was also revealed, although less strongly,
level, however, the focus was different. The by the data on self-reported motives and at-
changes in friendship during the school year tributions about the partners' motives. Eighth
were important because they afforded a test of graders more often attributed a motive for
a hypothesis about developmental changes in equality to their partners when the partner
friends' interactions in a particular context. was still a close friend rather than no longer a
When competition rather than sharing is en- friend. Fourth graders showed a slight trend
couraged by the situational context, the pref- in the opposite direction. Attributions of
erence of friends for equal sharing over com- competitive motivation to partners and self-
petition is assumed to increase between reported competitive motivation did not vary
middle childhood and early adolescence. The with friendship, however. Children's reports
results strongly supported this hypothesis. on their own motives for equality varied with
friendship but not with grade. Across both
When working on a task that offered a
grades, children said they tried for equality
choice between sharing equally and trying to
more when their partner was a close friend
get more rewards than a partner, fourth grad-
ers who had remained close friends with their rather than a former friend. Although these
responses suggest that equality is important
partners shared less in the spring than fourth
in friendships across a broad age range (see
graders who were no longer close friends.
Youniss, 1980), it is worth reiterating that only
Under the same conditions, eighth graders
who had remained close friends shared more the eighth graders, and not the fourth graders,
made similar attributions about their partners'
than eighth graders who were no longer close
motives for equality.
friends. Similarly, children who made more
positive comments about their friendships, The fourth graders were expected to
and children who had a greater balance of share less with friends than nonfriends be-
positive comments over negative comments, cause they were concerned about getting
shared less than other children at fourth grade fewer rewards than a friend and thus appear-
but more than other children at eighth grade. ing inferior to him or her. An alternative hy-
The findings for fourth graders replicate pothesis is that the fourth graders shared less
those in previous studies, although the friend- because they enjoyed competing with friends.
nonfriend contrast was created in previous re- The evidence on children's reactions to the
search by pairing children at one time point task cast doubt on the alternative hypothesis.
either with a close friend or with another Fourth graders did not express more satisfac-
classmate who was not a close friend (Berndt, tion with the outcome of the task when they
1981b; Staub & Noerenberg, 1981). The were paired with close friends rather than for-
young children in previous studies mightmer friends. The eighth-grade girls who were
have shared more with classmates than paired with former friends shared little with
friends because they wanted to gain a new their partners and expressed considerable dis-
friend. This explanation does not apply satisfaction
as with the outcome of the task. In
well to the current results, because the non- the entire sample, children who shared less
friends in the spring were not potential new with each other reported greater dissatisfac-
friends. Rather, they were old friends who tion no (r = -.43, p < .01). The correlations for
longer considered themselves close friends. fourth and eighth graders, -.33 and -.59, re-
The old friends were not enemies: Only two spectively, do not differ significantly. These
of them (or 7%) indicated that they disliked data imply that neither fourth nor eighth grad-
their partners (i.e., rated them 1). Their mean ers viewed competing with their partners as
rating of liking for each other was 3.0, which an acceptable alternative to equal sharing.
is similar to the mean rating for the classmates
studied by Berndt (1981b). Consequently, the The entire set of findings for sharing, re-
ported motives, and reactions to the task is
evidence from this study and previous studies
can be summarized most accurately by saying
most consistent with the hypothesis that
friends are more concerned about their per-
that fourth graders are likely to share less with
formance relative to each other than are non-
close friends than with peers whom they like
friends. How friends respond to this concern
to a moderate degree, if they are working on
tasks that elicit competition. Eighth graders about relative performance changes with age.
When placed in a situation where they risk
are likely to share more with close friends
than with other peers when working on the losing to a friend, young children try to re-
same tasks. duce the risk of losing by competing rather
than sharing equally with the friend. In the
The developmental change in friends'same circumstances, adolescents try instead
preferences for equal sharing over competi- to assure that they achieve equality in re-
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
1296 Child Development
wards-rather than winning or losing-by (Eds.), Friendship and social relations in chil-
sharing equally. This age change is consistentdren (pp. 15-44). New York: Wiley.
with the general hypothesis that mutual re-Bukowski, W. M., & Newcomb, A. F. (1984). Stabil-
sponsiveness between friends increases as ity and determinants of sociometric status and
children enter adolescence (Sullivan, 1953; friendship choice: A longitudinal perspective.
Youniss, 1980). Developmental Psychology, 20, 941-952.
Coleman, J. C. (1980). Friendship and the peer
Of course, the current results are based group in adolescence. In J. Adelson (Ed.),
on children's behavior when working on a Handbook of adolescent psychology. New
structured, artificial task. The apparent shift inYork: Wiley.
friends' preferences for equality over compe- Diaz, R. M., & Berndt, T. J. (1982). Children's
tition should also be examined in more natu- knowledge of a best friend: Fact or fancy? De-
ralistic settings. For example, athletics is one velopmental Psychology, 18, 787-794.
major arena for competition during childhoodDuck, S. W. (1975). Personality similarity and
and adolescence. How friends deal with com- friendship choices by adolescents. European
petition in athletic events has received little Journal of Social Psychology, 5, 351-365.
attention. The academic arena offers other op-Epstein, J. L. (1986). Friendship selection: Devel-
portunities for competition and social com- opmental and environmental influences. In E.
parison. Recent research suggests that ado- Mueller & C. R. Cooper (Eds.), Process and
lescents limit their academic competition outcome in peer relationships (pp. 129-160).
with friends by choosing friends who are New York: Academic Press.
fairly equal in ability and then assuming that Furman, W., & Bierman, K. L. (1984). Children's
they and their friends have complementary conceptions of friendship: A multidimensional
strengths and weaknesses (Tesser, Campbell, study of developmental changes. Develop-
& Smith, 1984). How successful such tech- mental Psychology, 20, 925-931.
niques are in controlling competition be-Gottman, J. M. (1983). How children become
tween friends is unknown. Therefore, further friends. Monographs of the Society for Re-
investigation of the effects of competition on search in Child Development, 48(3, Serial No.
friendship would be desirable. In general, the 201).
degree to which children experience prob-Hallinan, M. T., & Tuma, N. B. (1978). Classroom
lems in maintaining friendships over a school effects on changes in children's friendships.
year, accommodating to friends' expectations Sociology of Education, 51, 270-281.
for an intimate and mutually responsive rela-Hartup, W. W. (1983). Peer relations. In E. M.
tionship, and mastering the balance between Hetherington (Ed.), P. H. Mussen (Series Ed.),
equality and competition merits more exten- Handbook of child psychology: Vol. 4. Social-
sive investigation. ization, personality, and social development
(pp. 103-196). New York: Wiley.
References Hill. C. T., Rubin, Z., & Peplau, L. A. (1976).
Breakups before marriage: The end of 103 af-
Berg, J. H. (1984). Development of friendship
fairs. Journal of Social Issues, 32, 147-168.
among roommates. Journal of Personality and
Horrocks, J., & Buker, A. (1951). A study of friend-
Social Psychology, 46, 346-356.
ship fluctuations of preadolescents. Journal of
Berndt, T. J. (1981a). Age changes and changes
Genetic Psychology, 78, 131-144.
over time in prosocial intentions and behavior
Kahn, A., Nelson, R. E., & Gaeddert, W. P. (1980).
between friends. Developmental Psychology,
17, 408-416. Sex of subject and sex composition of the group
as determinants of reward allocations. Journal
Berndt, T. J. (1981b). The effects of friendship on
of Personality and Social Psychology, 38, 737-
prosocial intentions and behavior. Child Devel-
750.
opment, 52, 636-643.
Kandel,
Berndt, T. J. (1981c). Relations between social cog- D. B. (1978). Homophily, selection, and
nition, nonsocial cognition, and social behav- socialization in adolescent friendships. Ameri-
ior: The case of friendship. In J. H. Flavell & Journal of Sociology, 84, 427-436.
can
Knight, G. P., Dubro, A. F., & Chao, C-C. (1985).
L. D. Ross (Eds.), Social cognitive develop-
ment: Frontiers and possible futures (pp. 176- Information processing and the development
199). Cambridge: Cambridge University Press.of cooperative, competitive, and individualistic
Berndt, T. J. (1982). The features and effects social
of values. Developmental Psychology, 21,
37-45.
friendships in early adolescence. Child Devel-
opment, 53, 1447-1460. McCall, R. B., & Applebaum, M. I. (1973). Bias in
the analysis of repeated-measures designs:
Bigelow, B. J., & La Gaipa, J. J. (1980). The devel-
opment of friendship values and choice. In Some alternative approaches. Child Develop-
H. C. Foot, A. J. Chapman, & J. R. Smithment, 44, 401-415.
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms
Berndt, Hawkins, and Hoyle 1297
This content downloaded from 36.73.99.250 on Sat, 06 Apr 2019 07:34:34 UTC
All use subject to https://about.jstor.org/terms