Chapter 6

The Peer Context

Willard W. Hartup
Socialization in the peer context varies from culture to culture, with considerable variation
existing in the onset of a child's earliest experiences with other children. In most societies,
children begin to socialize with one another in early childhood; sustained and coordinated
social interaction becomes evident in the years between 3 and 6. Both quantitative and
qualitative changes occur in middle childhood, and, between the ages of 6 and 12,
socialization in the peer context becomes a central issue in children's lives.
The peer system can be represented as a matrix of contexts and components (see Figure 6-1).
The vertical axis of this matrix consists of a hierarchical ordering (Hinde, 1976) of various
social contexts. The most basic of these are interactions, i.e., meaningful encounters between
two or more individuals. Relationships are interactions between individuals (known to each
other) that persist over time and that involve expectations, affects, and characteristic
configurations of interactions. Groups, subsuming both interactions and relationships,
possess structural and normative dimensions that are not evident in either of the other
contexts; most commonly, groups are polyadic rather than dyadic. Macrostructures are
higher-order social contexts, including entities that we commonly call institutions or
societies. These macrostructures consist of dynamic interrelationships among the
interactions, relationships, and groups that constitute them.
The peer system.
Peer contexts involve specific objects and events occurring in specific times and locations.
Three situational components can be identified: the setting, the ''problem,'' and the actors.
Settings consist of the milieux in which social activity occurs. Problems consist of the
challenges existing in these settings that activate or energize the individuals involved. The
actors are the individuals with whom the target child interacts.
Using this matrix we can describe the structure, content, affects, and diversity of the peer
system—in each case as a function of context and component. This matrix is not a model of
the peer system or a theory of social dynamics; it is a simple schematic that can be used to
describe the peer system as it has been examined empirically. Less a theoretical statement
than a pragmatic device, this matrix focuses attention on the various levels of the individual's
commerce with other children.
Accordingly, in the various sections of this chapter, peer relationships in middle childhood
are discussed with an emphasis on child-child interactions and their changes with age; close
relationships and their significance; group formation and functioning; and interconnections
between the peer system and two macrostructures—the family and the school. Peer
interaction and the socialization of the individual child are examined in relation to conditions
of the setting and the identity of the individuals with whom children interact. Methodological
issues are discussed, especially the problems encountered in obtaining naturalistic data.

Peer Contexts

Interactions
Child-child interaction differs from adult-child interaction in many ways. Barker and Wright
(1955) observed that children's actions toward adults are weighted mainly with appeals and
submissive acts; actions of adults toward children consist mainly of dominance and
nurturance. These interactions are thus concentrated in two complementary issues: the child's
dependency on the adult and the adult's need to control the child. In contrast, the most
common actions of children toward child associates are sociability and
assertiveness/aggression.
Cross-cultural observations (Whiting and Whiting, 1975) are consistent with these results, as
are recent interviews with American schoolchildren (Youniss, 1980). Children see
themselves as recipients of adult actions rather than vice versa. In contrast, child-child
interactions are seen as revolving around equal exchanges between the actors. Concordantly,
"kindnesses" in adult-child interactions are conceived as actions confirming complementary
expectations, whereas kindnesses in child-child interactions consist of actions confirming
more egalitarian expectations. Close relationships (e.g., friendships or parent-child
relationships) usually involve mixtures of complementary and equal interactions, but this
does not negate the thesis that children differentiate between adult-child and child-child
relationships mainly in terms of this dichotomy.
The time that children spend together and the nature of their interactions when they are on
their own are not well documented. Patterns of interaction have been most extensively
examined in ad hoc settings, mainly schools, and it is largely on an anecdotal basis that we
have concluded that more and more time is devoted to child-child interactions in middle
childhood. Observations of 8 children, each covering an entire day (Barker and Wright,
1955), revealed that approximately 85 percent of the children's activities were social and that
the proportion spent with child associates rose from 10 percent at age 2, to 20 percent at age
4, to slightly over 40 percent between ages 7 and 11. The school-age children engaged in an
average of 299 behavior episodes (i.e., interactive segments marked by constant direction
and intent) with other child associates in a typical school day, 45 of these with siblings and
the remainder with friends. Detailed records based on observations of one of these children
(Barker and Wright, 1951) indicate that most of these episodes consisted of play or "fooling
around" and that the interactions consisted mainly of sociability and dominance exchanges.
Although these measures are difficult to translate into time units, it can nevertheless be
concluded that time spent with child associates consumed hours rather than minutes.
More recent time-use studies clarify, to some extent, what children do with one another on
their own. Even so, the frequencies of their activities and the structures of the social
interactions remain unstudied. Using interviews with 764 sixth graders in Oakland,
California, Medrich et al. (1982) asked the children to enumerate "what you like to do when
you are with your friends." Responses covered a range indicating that, in contrast to time
spent alone, children spend their time with their friends engaged in physically active or
"robust" interactions. Team sports accounted for 45 percent and 26 percent of the
enumerations of boys and girls, respectively, although other types of robust interactions such
as "general play," "going places,'' and "socializing" were more commonly mentioned by girls
than by boys. These interactions occurred most often outside the home, although close to
home and more often in private than in public places (e.g., parks and playgrounds), and were
segregated by sex.
Overall, it appears that a substantial portion of the schoolchild's daily existence is spent in
peer interaction and that the content of this interaction consists mainly of play and
socializing. Children in many cultures also share work experiences in which they take turns
and substitute for one another to a greater extent than when they work with adults (Weisner,
1982). The proportion of child-child interactions spent in work and in play thus varies from
culture to culture, but the nature of this interaction seems universally to be more egalitarian
than the interaction that occurs between children and adults.
The two classes of child-child interaction most extensively studied in developmental terms
are aggression and prosocial activity. Overall, aggression decreases in middle childhood,
although both mode and content change (Parke and Slaby, 1983). Physical aggression (more
common among boys than among girls) and quarreling decrease, although abusive verbal
exchanges increase. Schoolchildren typically engage in instrumental aggression (directed
toward retrieving objects and the like) less frequently than younger children, although
person-directed hostile aggression is more common among older ones. Increasingly salient in
middle childhood are insults, derogation, and other threats to self-esteem (Hartup, 1974).
Between ages 6 and 12, aggressive boys are more ready to attribute hostile intent to others
than are nonaggressive boys. In addition, their associates more often attribute hostile
intentions to aggressive than nonaggressive boys, and the former are more often targets of
aggression (Dodge, 1980; Dodge and Frame, 1982). Individual differences in aggression
among school-age males thus come to be associated with social cognitive biases (a
willingness to perceive hostility in others), negative experiences with others, and "bad" social
reputations.
Certain studies indicate that sharing and other forms of altruism increase in middle
childhood; others suggest that these growth functions are more complex (Radke-Yarrow et
al., 1983). In one investigation, for example, no relationship was found between age and the
amount of sharing among kindergarten and second and fourth grade students, but among the
older children more individuals shared, more individuals shared in "complex" situations, and
more advanced reasoning was used to rationalize the altruism (Bar-Tal et al., 1980). Still
other evidence suggests that age differences in altruism vary according to context.
Kindergarten children were found to intervene directly in assisting another child with a
difficult task, but third graders differentiated their assistance according to the situation—the
older children would usually offer suggestions or assistance before actually giving it, and
sometimes intervention would be withheld altogether (Milburg, McCabe, and Kobasigawa,
unpublished data). Thus, there is evidence that complex attributions become increasingly
involved in prosocial interactions as well as aggressive interactions in middle childhood.
Competition, in contrast to cooperation, increases with age when rewards are allocated in
proportion to the number of points accumulated in a game (Avellar and Kagan, 1976) or
when points are tallied (McClintock, 1974). These age gradients occur more clearly in some
cultures (the United States, Japan) than in others (Mexico, Kenya). With outcome controlled,
competitive preferences are not as clearly evident, and in some tasks cooperation increases
with age (Kagan et al., 1977; Skarin and Moely, 1976). Chronological age and goal structure
thus seem to interact in children's cooperative and competitive choices.
Studies of American children reveal that increases in competition under winner-takes-all
conditions occur mainly in the preschool years; cooperation under shared reward conditions
increases mainly between ages 6 and 8; and age differences in individualistic (proportional
reward) conditions vary according to the manner in which a child's gains are linked to the
gains of others (McClintock and Moskowitz, 1976; McClintock et al., 1977). With time,
children differentially use strategies that coincide with the goal structures associated with
obtaining valued outcomes. Children do not simply become more competitive or cooperative
as they grow older. They become sensitive to the contingencies controlling the incentives
that are important to them. Again, cognitive and social factors seem to determine the nature
of child-child interactions in middle childhood.
More detailed studies, including direct assessment of the attributions made by children in
distress or conflict situations, are relatively rare. We know that (a) the social cues used in
interactions (e.g., facial expression, vocal intonation) are encoded with increasing accuracy
between ages 6 and 10 (Girgus and Wolf, 1975); (b) visual attention is increasingly utilized
in conversation (Levine and Sutton-Smith, 1973); (c) increases occur in speakers' abilities to
transmit information about simple problems to listeners and to respond appropriately to
queries from their listeners, and listeners' utilization of feedback improves (see, e.g.,
Karabenick and Miller, 1977); (d) abilities to infer motivation and intent in simulated social
situations increase in middle childhood, although these trends are most evident in cognitively
complex situations (see Shantz, 1983); and (e) increases occur in children's abilities to
integrate two sources of information as opposed to one (Brady et al., 1983). These results
suggest that a variety of cognitive constraints on social interaction become less evident in
middle childhood. These constraints appear to involve the coding, storage, and retrieval of
information as well as the integration and application of information in social situations.
Existing studies thus provide a general characterization of child-child interactions between
ages 6 and 12. First, time spent with child associates increases. Second, children become
more adept at sending and receiving messages, in utilizing information from a variety of
sources to determine their actions toward other children, in making causal attributions, and in
coordinating their actions with those of others. We are only beginning, however, to
understand the manner in which changes in the content and structure of child-child
interactions reflect changes in cognitive functioning (Hartup et al., 1983).

Relationships

Social Attraction
Sociometric techniques have been used to examine the characteristics that make children
attractive to one another. Usually, sociometric interviews are administered concurrently with
personality and intelligence tests or with behavioral ratings made by teachers, children, or
observers, and the various scores are correlated. We know three things.
First, social attractiveness is associated with sociocultural conditions. Social class is
positively correlated with attractiveness (Grossman and Wrighter, 1948), and recent studies
indicate that socioeconomic variations may exist in the values concomitant with sociometric
status. For example, popularity in middle-class schools is correlated with the use of positive
verbal overtures among children, whereas status in working-class schools is associated with
the use of nonverbal overtures. Middle-class children who engage in nonverbal interactions,
even though positive, are actually rejected more often than children who do not use these
techniques (Gottman et al., 1975).
Second, characteristics of the child are correlated with social attraction. Being liked is
associated with being physically attractive, socially outgoing, and supportive of others;
achievements in school and in sports are also positively associated with social attractiveness.
Being rejected is associated with being unattractive, immature, disruptive, and aggressive in
indirect ways. Rejected children, however, are not necessarily more aggressive in general
than their nonrejected peers (see Hartup, 1983).
Third, children's reputations mirror these differences. Sociometric "stars" and "average"
children have social reputations that are accepting and that allow them considerable
flexibility in dealing with their companions (New-comb and Rogosch, 1982). These children
are regarded by their associates as cooperative, supportive, and attractive (Cole et al., 1982).
Rejected children are restricted by their reputations and the negative expectations of their
companions. They tend to be perceived as disruptive and indirectly aggressive (Coie et al.,
1982).
The causal connections underlying these results are presumably bidirectional. That is, the
well-liked child appears to possess a repertoire of effective social skills and a positive social
reputation—conditions associated with a high probability that other children will behave
supportively. This support in turn maintains the competent behaviors and the child's social
reputation. Similarly, negative attributes seem to undergird negative social reputations,
nonsupportive feedback from one's associates, and in turn a continuation of the negative
behaviors (see Coie and Kupersmidt, 1983; Dodge and Frame, 1982).
Certain side effects of these conditions have been documented. Popular and rejected children,
for example, are members of distinctive social networks. Rejected children, compared with
popular children, socialize on the playground in small groups and more frequently interact
with younger and/ or unpopular companions. The social networks of popular children are
more likely to be composed of mutual friends and to be characterized as cliquish (Ladd,
1983). These distinctive networks suggest the existence of a self-maintaining cleavage
between popular and rejected children. Social attraction thus seems to involve a nexus of
social skills, social reputations, the extent to which one socializes with friends, and the
extensiveness of one's social world.
Most likely this nexus is mediated through an intricate set of self-attitudes and emotions. To
date, studies of self-esteem, self-conceptions, and social acceptance have not been
convincing. Several investigators have noted small correlations between self-esteem and
popularity (Horowitz, 1962; Sears and Sherman, 1964), but other results suggest that any
correlation is curvilinear. Reese (1961) found that children with moderately high self-esteem
were better accepted by their peers than were children with either low or very high self-
esteem. Sixth graders with high self-esteem have been shown to make more extreme
statements about the likability of others than their low self-esteem counterparts; the extent to
which these children believe their evaluations of others are reciprocal is also positively
related to self-esteem (Cook et al., 1978). But the self-system is involved in social
relationships in very complex ways. Dodge and Frame (1982), for example, found that
aggressive boys were characterized by hostile attribution biases only when the provocation
was directed toward themselves. These biases were not evident in situations involving
provocations directed at someone else.
Developmental changes in social attraction have been examined relatively rarely. For one
thing, the characteristics associated with sociometric status do not appear to change with age.
Friendliness is as strongly associated with popularity among older children as among
younger children; indirect aggression is as strongly associated with rejection. Nevertheless,
recent investigations reveal that more of the variance in social preferences can be predicted
by fewer variables among young children than among older children. Social impact, too,
rests on fewer attributes among younger children than among older ones (Cole et al., 1982).
The results thus seem to indicate that person perception becomes more differentiated as
children grow older—a conclusion that is consistent with the results of other investigations in
which children's descriptions of one another were examined in relation to chronological age
(Livesley and Bromley, 1973). Also, certain sociometric dimensions may become
increasingly stable in the years between 6 and 12. For example, rejection status was observed
to be stable over a 5-year span when assessment was initiated in the fifth grade but only over
3 years when initiated in the third grade (Coie and Dodge, 1983). These results suggest that a
"crystallization" may occur in social relationships toward the end of middle childhood.

Friendship Selection
Children and their friends usually live in the same neighborhood, a condition that prevails in
both early and middle childhood (Epstein, in press; Fine, 1980). Young children depend on
their caretakers to put them in contact with other children more than school-age children do,
and classroom proximity becomes salient in friendship selection in middle childhood.
Conditions within classrooms, including seating arrangements and classroom organization,
are also reflected in friendship selection.
Children select their friends mainly from among children their own age. When classroom
conditions favor mixed-age choices (as in a one-room school), more than 67 percent of
children in the first 6 grades have one or more friends of some other age (Allen and Devin-
Sheehan, 1976). Nevertheless, the tendency for children and their friends to be similar in age
is very strong. Whether this concordance derives from the age segregation that marks most
schools and children's institutions or from children's own preferences is not certain.
Moreover, there may be no way to resolve this issue, since age grading is pervasive in
Western cultures.
Children and their friends are most commonly of the same sex (Tuma and Hallinan, 1977).
This concordance peaks between ages 6 and 12, even though same-sex choices are more
common than other-sex choices from the preschool years through adolescence. Since sex
segregation is not common in schools except in sports activities, the concordance may derive
from norms generated by children themselves rather than from the normative expectations of
adults.
Fewer cross-race friendship selections occur in integrated classrooms than would be
expected on the basis of chance, and this cleavage increases between ages 6 and 12—as
determined by longitudinal studies (Singleton and Asher, 1979). Racial differentiations are
not as strong in children's selection of playmates or work companions as in friendship
choices, and age differences are not as evident (Asher et al., 1982).
Behavioral similarities and their role in mutual attraction in middle childhood have not been
well studied. Kandel (1978), on the basis of one study of a large sample of adolescents,
concluded that these similarities are not especially important in the selection of associates,
except for similarities in significant nonnormative attitudes (e.g., about drug use). Given the
importance to children of "doing things together" with their friends, it is difficult to believe
that behavioral concordances are irrelevant in these selections (Smollar and Youniss, 1982).
Nevertheless, except for a small number of investigations using global measures such as IQ,
school achievement, or sociometric status, which show very modest concordance between
children and their friends, this issue has not been closely examined. This state of affairs is
unfortunate, since it has been known for some time that school-age children, like adults,
demonstrate greater attraction for peers with whom they share many attitudes than for
individuals with whom they share relatively few (Byrne and Griffitt, 1966).

Acquaintances
Friendship formation begins with acquaintanceship. As two individuals become familiar with
each other, attraction seems to increase (Berscheid and Walster, 1978). Mere exposure
(Zajonc, 1968) may account for these effects, and familiarization may also establish a secure
base for social interaction; moreover, as individuals become acquainted with one another,
their social repertoires become better meshed and more efficient. Various studies (mostly
with younger children) support these ideas. To date, however, these hypotheses have not
been used to any great extent in investigations with school-age children.
To investigate children's notions about the manner in which two individuals become friends,
Smollar and Youniss (1982) asked three questions of subjects between ages 6 and 13: "What
do you think might happen to make X and Y become friends?" "Not become friends" "To
become best friends?" The children's responses differed according to their ages. Younger
children indicated that strangers would become friends if they did something together or did
something special for one another. Children would become best friends, according to the
younger children, if they could spend increased amounts of time together, especially outside
school. In contrast, the older children emphasized getting to know each other ("talk and talk
and find out if they like the same things"); discovering similarities between themselves was
considered necessary to becoming best friends. Not becoming friends was associated among
the younger children with negative or inequitable interaction. This condition was identified
among the older children with the discovery that two individuals are different. What is
interesting in these findings is the revelation that concordance was viewed as essential in
friendship formation at all ages; it was mainly the expression of these concordances that
differed with age. Younger children emphasized concrete reciprocities, while older children
emphasized psychological similarities (e.g., in personality, likes, and attitudes).
Microanalytic studies of acquaintance interactions do not extend more than to the first few
encounters between children. Virtually no developmental studies have been executed. First
encounters differ with respect to a number of conditions, including the sociometric status of
the children involved. When both children are of high-status, information giving and seeking
are more frequent than when both children are of low status. Discussions about school,
sports, the children themselves, and their acquaintances are common. Pairs of third- and
fourth-grade children that include one high-status and one low-status child are virtually
identical in these respects to those of two high-ranking children, presumably because the
interaction is driven by the high-status member (Newcomb and Meister, 1982). A second
investigation revealed that "stars" and sociometrically "average" third-grade children
engaged in more introductory activity and information exchange, earlier onset of affective
communication, and game-playing than "isolate" or "rejected'' dyads. In contrast, isolates and
rejected children attempted to initiate games more frequently but engaged in more
inappropriate interactions than did stars or average dyads (Newcomb et al., 1982). These
analyses thus indicate that a major function of the early encounters between children is
assessment of interests and similarities.
Other studies indicate that synchronization is an outcome of the early encounters between
strangers. Brody et al. (in press) observed triads of first-and third-grade children from
different classrooms before and after a series of five play sessions. Postfamiliarization
measures revealed more verbal interaction and improved task performance than among
control subjects, indicating better meshing of individual contributions to task solution. Since
mere exposure seems to have variable effects on social attraction among children (Cantor
and Kubose, 1969), early social encounters seem mainly to provide a basis for interaction via
the sharing of information and behavioral coordination. Beyond this, no clear picture of
acquaintanceship processes emerges, and the features of acquaintance interaction that favor
the continuance of an association are unknown.

Friends
School-age children have, on the average, five best friends—a number that is a bit higher
than the number of friends acknowledged by preschool children and adolescents (Hallinan,
1980). In addition, relatively few (2 percent) have no friends when choosing in sociometric
interviews, although a somewhat larger number (between 6 and 11 percent) are not chosen
themselves. These frequencies do not change from age 6 to 12, although age differences have
never been studied adequately (Epstein, in press). School-age children and their friends tend
to be linked in twosomes rather than in the larger interlocking networks known as cliques or
crowds. Relatively few cliques are observed in most elementary school classrooms, in
contrast to junior and senior high school (Hallinan, 1976). Considerable interest is now
evident among investigators in the social interaction of friends. Most of the recent work
draws heavily from the theories of Harry Stack Sullivan (1953), who argued that friendships
are hallmarks of the "juvenile era," reflecting new needs for interpersonal intimacy and new
contexts for their expression. Recurrent themes in contemporary research are reciprocity,
equity, fairness, mutuality, and intimacy as these mark both children's conceptions of their
friends and their behavior with them. Presumably, these themes become more and more
important in middle childhood, so that the need for developmental studies is especially acute.
Cross-sectional studies confirm that friendship expectations among schoolchildren revolve
around these issues. Development does not involve a change from the absence of reciprocity
expectations among younger children to their presence. Reciprocity norms are evident among
kindergartners (Berndt, 1977). Interviews (Youniss, 1980) and written essays (Bigelow,
1977) confirm that "reward-cost" reciprocities figure prominently in children's expectations
of their friends at all ages. Bigelow's work suggests a progression from expectations among
second and third graders that are based on common activities to sharing of rewards and other
equities to mutual understanding, self-disclosure, and the sharing of interests among fifth and
sixth graders. Youniss's studies suggest that young friends "match" each other's contributions
to the interaction; older friends evidence equality and equal treatment in their relationships
with one another; and young adolescents stress interpersonal intimacy. Thus, the case can be
made that the major changes in children's friendship expectations occur in the way children
use notions about reciprocities in social relationships rather than in the emergence de novo of
generalized notions about fairness and mutuality. Individual differences in these
understandings are largely unexplored except for the work of Selman (1980), and a
connection between these social cognitive changes and behavior with friends has not been
established.
Observational studies confirm that friendships are based on reciprocity and mutuality, but
age changes in the behavioral manifestations of these reciprocities have been difficult to
document. (Presumably, this derives from the difficulty of measuring the subtle affective and
instrumental components of intimacy via direct observation.) Laboratory studies reveal that
in cooperative settings friends are more interactive, affective, attentive to equity rules,
mutually directive, and explore the materials more extensively than nonfriends (Foot et al.,,
1977; Newcomb and Brady, 1982; Newcomb et al., 1979). Changes in these differences with
age are not dramatic in middle childhood (Newcomb et al., 1979). Berndt (1981b) found that
fourth-grade friends assisted their partners and were more willing to share rewards with each
other than were first graders. These changes are not striking until early adolescence,
however, and then they occur most commonly when children have a choice between
cooperation or competition in the task.
Competitive settings change the interaction between friends. In these situations, males are
more competitive and less generous with their friends than with nonfriends (Berndt, 1981a).
Competition also occurs more commonly between friends than between nonfriends when
property rights are clearly understood (Staub and Noerenberg, 1981). Thus, under certain
circumstances, friendship may furnish a basis for competition, even though in others it
furnishes a basis for equity considerations and generosity. Studies of adults indicate that
close relationships may maintain a basis for hostile, aggressive interaction as well as
supportive, affectionate interaction. What evidence we have suggests that children also
manifest these complexities, even though these elements are not well understood. Future
investigations must concentrate not only on the documentation of fairness norms in
friendship interactions in middle childhood but also on a complex array of attributions that
make for distinctive interactions between children according to context.
More and more, the contemporary evidence stresses the importance of close relationships in
childhood. Much remains to be documented, however, about children and their friends.
Especially needed are studies addressing four issues: (1) the temporal course of friendships,
including the accommodations made by friends to stress that comes from both inside and
outside the relationship; (2) developmental vicissitudes, including the manner in which
expectations and attributions among friends are connected to social interactions and the
manner in which these relationships cycle through time; (3) individual differences among
friendship pairs in the structure of these relationships, the use of reciprocity rules, the content
of interactions, and their affective qualities; and (4) the socializing consequences of
friendships, i.e., their role in increasing similarities between children, the developmental
implications of having a best friend, and the value of friendships as protective factors in
times of stress. The research agenda is formidable.

Groups
Groups exist when social interaction occurs over time among three or more children, values
are shared, the members have a sense of belonging, and a structure exists to support the
activities of the collective. Empirical studies have concentrated on three issues: group
formation, norms, and structures.

Formation
Group formation has been understudied since the investigations 20 years ago by Sherif et al.
(1961). That work confirmed the ubiquity of social structures based on power relations, the
individual attributes associated with social power, and the emergence of group norms. Group
formation has not been studied with children in the early school years, so developmental
changes are undocumented. Only two studies have ever been conducted that chart changes in
group interaction as a function of the age of the group members (A. Smith, 1960; H. Smith,
1973), and no longitudinal studies document the connections between norms and structures
as they cycle through time. Processes in group formation among girls have been described
less well than those among boys. Little is known about group formation among children from
minority subcultures. And we know little about the impact of the macrostructure (e.g.,
schools) on the emergence of children's groups. Again, the research agenda is formidable.
Norms
Standards of conduct, called norms, govern the actions of group members. Certain norms
governing children's interactions with one another emanate from the core culture and
presumably derive from earlier socialization—e.g., sex-role stereotypes as well as attitudes
about authority, equity, and reciprocity. Other norms emanate from group interaction, and
these are the norms that adults associate with the peer culture. The existence of these
normative frameworks is not in doubt, but the processes through which they emerge are.
Knowing more about these processes is essential, however, since these norms may be salient
in the development of health behaviors, antisocial activities, and discordance in parent-child
relationships in middle childhood and adolescence.
On the basis of observations of little league baseball teams, Fine (1980) asserted that five
conditions must exist for norms to become established: (1) someone must ''know," i.e.,
possess a bit of social information and introduce it; (2) members must find the information
usable; (3) new norms must satisfy some common need; (4) norms must support group
structure and vice versa; and (5) circumstances must exist that trigger the normative activity.
This framework seems valuable for examining the introduction, maintenance, and cessation
of normative activities among children of various ages; these notions have not yet, however,
been used extensively in empirical investigations.
One's own appearance, drinking, drug use, or popularity may determine the selection of one's
associates (Dembo et al., 1979; Ladd, 1983). Longitudinal studies suggest that, in addition,
one's behavior influences the behavior of those individuals selected as associates (Britt and
Campbell, 1977; Kandel, 1978). Similarities among group members thus derive from both
normative selection and normative socialization. Once again, the connections between these
processes in middle childhood are not well understood, despite their obvious significance.
Especially understudied are normative constellations as they vary from one enclave to
another. Drug users, for example, tend to congregate with one another, although the weight
of the evidence suggests that children do not form subcultures that are distinguishable from
others except in drug use (Huba et al., 1979). Ordinarily, though, it is assumed that smoking
and drug use are especially common in groups of alienated or incompetent children. Which is
correct? What "mixtures" of core-culture and counterculture norms predict antisocial
behavior most successfully? So few intergroup comparisons have been made that it is
impossible to answer these questions.

Structures
Group members differentiate among themselves in terms of social power, i.e., their
effectiveness at directing, coordinating, and sanctioning the activities of other members.
These differentiations are the basis for the social structures that are visible in every group.
Group structures emerge in social interactions in early childhood, and these are relatively
well studied. Relatively few investigations, however, have documented the existence of
hierarchical social organizations in middle childhood (Sherif et al., 1961; Strayer and
Strayer, 1976).
In some instances these structures seem to be based on dominance interactions, although
social structures may also be based on being good at games, knowing how to organize
activities, and social competence. In general, individual differences in attributes that
facilitate the group's objectives are the basis for the social order; structures based on
dominance interactions do not always vary concordantly. Leaders are not necessarily tough
or mean, nor are they necessarily the most popular children. Leaders are the ones who know
what to do.
Our notions about group structures have been derived mainly from observations of children
in classrooms and summer camps. Classroom observations are notably constrained in
providing us with a clear picture of those structures that exist in informal groups. The social
structure of the classroom is dominated by an adult (the teacher), and normative activity
revolves around academics. Camp settings are not as heavily constrained but have their own
limitations. One of the most severe gaps in our knowledge of the social psychology of
middle childhood is information about the structure and functioning of informal groups.
Omitted from most studies is a consideration of the connections between norms and the
social organization. Sherif et al. (1961) documented the intimate relationship between
structure and function in social groups, but current investigations have been dominated by
the notion that social structures exist mainly to reduce the amount of aggression among
group members (Savin-Williams, 1979; Strayer and Strayer, 1976). Separate considerations
of normarive functions and social organization have not been wise, however, since this
strategy has reduced the interest of investigators in group-to-group variations. These
variations need to be better understood, not only to document the diversity of the social
environment but also to better understand the conditions under which children are attracted
to membership in certain groups.

Situational Components

Setting Conditions
The conditions of settings constrain child-child interactions (Barker and Wright, 1951).
Nevertheless, we have no clear idea about where children spend time when they are on their
own, let alone what social experience is like in these places. Enough has been accomplished
to demonstrate that child-child interactions vary according to conditions of the setting, but no
"ecology of the peer context" emerges from this material.
Playthings constrain the amount and maturity of child-child interactions, as examination of
the protocols from One Boy's Day (Barker and Wright, 1951) shows. Other than the
documentation that physical activities promote "robust" interactions and that nonphysical
activities are especially associated with helping behaviors (Gump et al., 1957), little is
known about playthings and play situations as constraints on social interactions in middle
childhood.
The availability of resources and the space with which to use them bear on social
interactions. Crowding effects have been studied mainly with younger children, and
methodological flaws mar many of the studies with school-age children. Space variations
apparently do not affect either positive social interaction or aggression in a linear mode. Only
when space per child is severely limited is positive interaction reduced and negative
interaction increased. With severe crowding, children experience emotional arousal and
increased competitiveness (Aiello et al., 1979). But there is also evidence that access to
resources may be more important in determining the nature of child-child interactions than
the amount of space available (Smith and Connolly, 1977).
The number of children congregated together is also salient even though, in the primary
grades, children tend to interact in dyads rather than in larger sets. Three-child (and larger)
enclaves become more common during middle childhood on playgrounds and in parks
(Eifermann, 1971), but dyadic interaction, with its concentration of social attention, remains
evident. Interaction is more intense and cohesive in smaller enclaves than in larger ones.
Consensus in group discussions is easier to reach and leaders exert more extensive influence,
even though the members of small enclaves have a sense of self-importance and coalitions
are less common (Hare, 1953). Given that children remain committed to dyadic interactions
throughout middle childhood, it is to be regretted that we do nor know whether the nature of
these exchanges varies according to the size of the groups in which they are embedded.

Problems
Every social situation contains some element of challenge, something to be done, something
to activate or energize the actors. Multiple challenges exist in most social situations, and
these may form a hierarchy according to their relative importance. Different challenges may
be important at different times. In some instances the main tasks are externally imposed and
evident to everyone; in others the tasks are not well defined, and the participants must
construct the task as they go along. Only recently has there been much interest in tasks and
their role in child-child relationships. Illustrations can be selected, however, that demonstrate
(a) the effects of a task on social interaction, (b) changes that occur with age in the tasks
children consider important, and (c) individual differences in the tactics children use in
certain situations.
The strategic demands of a social situation vary according to context. Children usually
understand, for example, that when a child is being teased by a number of children the most
appropriate way to assist the victim is through ordering and commanding the teasers (Ladd
and Oden, 1979). When encountering a solitary child who has been teased, however,
appropriate assistance is understood to include consolation, instruction, and suggestions of
alternative actions. Children who do not endorse these strategic norms but instead endorse
idiosyncratic social strategies turn out to be relatively low in the sociometric hierarchy. Thus,
a child's endorsement of normative strategies in certain task situations seems to be central in
social effectiveness.
When social situations are not well structured, children construct their own goals, and to
some extent these change between ages 6 and 12. Renshaw and Asher (1983) showed third-
and sixth-grade children four hypothetical social situations (social contact with strange
children, group entry, a friendship issue, and an instance of conflict) and conducted
interviews about the goals and strategies the children considered important. Both the third
and the sixth graders recognized "friendly" goals as most appropriate in each of these social
situations, but, when asked to mention their own goals, the older children mentioned friendly
ones more often and the younger children were more often concerned with defending their
rights. Endorsed social strategies also differed according to age. The older children, in
contrast to the younger ones, more often were outgoing and accommodative, indicating more
sophisticated adjustment to the social task.
One task that has received considerable recent attention is group entry. Results indicate that
children vary in the extent to which they view entry as important as well as in the tactics they
use to enter a group (Dodge et al., 1983; Putallaz and Gottman, 1981). As it turns out,
successful entry usually involves sequences of tactics that progress from the use of low-risk
ones (e.g., waiting and hovering on the edges of a group) to high-risk ones (e.g., statements
and requests). Moreover, the child must then use high-risk tactics that maintain the group's
frame of reference (e.g., calling attention to what the group is doing) rather than disrupting it
(e.g., calling attention to oneself). The effectiveness of these strategies has been
demonstrated among children interacting with familiar associates as well as unfamiliar ones
and in naturalistic as well as laboratory settings. Some evidence suggests that the high-risk
tactics that work are more consistently used by children toward the end of middle childhood
than at the beginning (Lubin and Forbes, 1981). Furthermore, their use differentiates popular,
neglected, and rejected children (Dodge et al., 1983; Putallaz and Gottman, 1981).
There is every reason to believe, then, that tasks interact with developmental and clinical
status in determining the social strategies that children use with their associates. Even a
child's understanding of what the social task is may index relevant individual differences in
social functioning. Clearly, additional work in this important new area should be encouraged.

Actors

Age
Children's associates vary widely in age, with as much as 65 percent of social contacts
occurring with others who are more than 12 months younger or older (Barker and Wright,
1955). Mixed-age contacts among adolescents occur more commonly in shopping malls and
parks than in schoolyards (Montemayor and Van Komen, 1980). Similar data are not
available, however, to show where mixed-age experiences are especially common among
school-age children. Nevertheless, two conclusions emerge from recent studies of mixed-age
interactions compared with same-age interactions: mixed-age interactions are less egalitarian
than same-age interactions, and social accommodations are made in mixed-age
circumstances that are not evident under same-age conditions.
Children are more nurturant and directive with younger children and more dependent with
older children (Graziano et al., 1976; Whiting and Whiting, 1975); similar role asymmetries
are evident between siblings who differ in age by 2-3 years (Brody et al., 1982). Same-age
interaction, in contrast, is marked by "sociable" and "aggressive" interactions (i.e., equal
exchanges) to a greater extent than mixed-age interactions (Whiting and Whiting, 1975).
These differences are concordant with differences in attributions made to same- and mixed-
age associates. Power attributions are more commonly made to older associates ("smart,"
"best," ''bossy'') and their reciprocals to younger associates ("weak," "dumb"). In these
comparisons, attributions to same-age associates are more likely to resemble those made to
younger associates than to older ones (Graziano et al., 1980).
Three omissions, however, mark our data base. First, the differences between mixed and
same-age interactions that emerge in early childhood and extend to the early school years
have not been explored among older schoolchildren. The developmental course of the
complementarities existing in mixed-age interactions is thus not well documented. Second,
little is known about these complementarities as a function of the age differences of children.
Most investigators have examined mixed-age interactions between children who differ by
two years in chronological age. One-year differences seem to affect children's efficacies as
role models (Thelen and Kirkland, 1976), but little else is known about these smaller
differences. No investigator has examined the complementarities that may exist across
greater differences in age. Third, little is known about the role of attributions in generating
the differences between mixed- and same-age interaction in various setting conditions. The
existing data derive mainly from observational studies in cooperative atmospheres.

Sex
Children's societies are segregated by sex. Within these male and female cultures, boys
interact in outdoor public places more commonly than girls do and come under the
supervision of adults less frequently. Boys interact in larger groups than do girls, and mixed-
age contacts are more frequent in male interactions. Play among boys is rougher and includes
more frequent instances of fighting than play among girls (Lever, 1976; Thorne, 1982). In
addition, social speech seems to serve different functions in male and female societies. Girls
use speech more extensively than boys to create and maintain relationships, to criticize
others in acceptable ways, and to clarify the speech of others. Boys use words to assert social
position, to attract and maintain an audience, and to assert themselves when other speakers
have the attention of the audience (Maltz and Barker, in press).
Mixed- and same-sex interactions have seldom been compared. Sgan and Pickert (1980)
examined assertive bids in a cooperative task with triads of kindergarten, first-grade, and
third-grade children. With age, boys made proportionally fewer cross-sex assertive bids in
mixed-sex triads and girls made more; concomitantly, same-sex assertions increased among
girls but decreased among boys. These trends result in mixed-sex interactions becoming
more egalitarian with age, but whether these trends are evident in competitive or
individualistic tasks is not known. Coalition formation differs according to gender
composition, with same-sex coalitions formed more frequently in mixed-sex triads than
cross-sex coalitions, especially when children of the same sex occupy positions of relatively
low social power (Leimbach and Hartup, 1981). Gender and power relationships thus may
interact in the emergence of cross-sex social organizations. Developmental trends in these
outcomes have not been charted.
Despite the sex cleavage in middle childhood, sex segregation is not complete. Rather, this
segregation seems to have a "with-then-apart" structure in which the sexes congregate
separately but also come together in many situations, especially schoolyards and city streets.
Using the techniques of participant observation in school hallways, cafeterias, and
playgrounds, Thorne (1982) established, first, that sex segregation derives from both
inclusion and exclusion and, second, that cross-sex exclusions are based mainly on sex
typing and the "riskiness" of romantic involvement. Even so, "borderwork" (i.e., ritualized
"invasions" resulting in cross-sex interaction) are common. Some of these, for example,
"chasing," and ''kiss and chase," have sexual overtones. Others involve stigmatization (boys
referring to the girls as cootie queens). Still others involve territorial invasions. But cross-sex
interactions also occur, in which the children merge easily into interactions with the opposite
sex. Common examples include the inclusion of girls in team sports, especially tomboys (see
Lever, 1976). As adolescence approaches, the taboo against romantic involvement begins to
break down, and occasionally boys and girls will consider themselves as going together. And
when tasks and resources are absorbing and adults legitimate cross-sex interactions, more
harmonious cross-sex contacts occur.
These observations indicate that much can be learned from naturalistic investigations about
the forces that support sex segregation in middle childhood and, more important, the forces
that instigate and maintain cross-sex interactions in these years. The developmental
implications of cross-sex interactions are especially significant. Indeed, it may be their
highly ritualized nature that represents their significance in development.

Race
Racial awareness increases in middle childhood, although the contemporary evidence is not
extensive concerning the foundations of the race cleavage that marks children's societies.
Prowhite/antiblack biases are evident among white children, but in many instances the
choices of black children do not depart from chance (Banks, 1976). "Eurocentric"
prowhite/antiblack choices decline among school-age black children (Spencer, 1981),
although there is considerable variation in the existence of problack biases. Associative
contacts are more likely to be of the same sex than of the same race (Asher et al., 1982), but
assortments in cafeterias, playgrounds, and hallways are notably segregated by race. School
integration has changed these patterns somewhat, although friendship interactions among
adolescents are strongly constrained by race (see above).
These segregations in child-child interactions are undoubtedly based on both inclusive and
exclusive processes. The salience of racial similarities has seldom been explored as a basis
for inclusion; in contrast, considerable thought has been given to racial biases as sources of
exclusion. The borderwork and other conditions that instigate mixed-race interactions are not
well established except that cooperative activities in the service of superordinate goals seem
to promote it (Aronson, 1978). Microanalytic studies of mixed-race interactions compared
with same-race interactions are virtually nonexistent. Among younger adolescents in mixed-
race conditions, white children are more likely to initiate social interaction than are black
children, and white children have stronger influence on group decisions. No evidence
establishes the precursors of these patterns in middle childhood or the conditions that might
modify them.

Comment
Presumably many other "actor attributes" determine the nature of child-child interactions in
middle childhood. The source of a leader's authority and his or her personal attributes
determine social effectiveness among adolescents, but the emergence of these conditions in
middle childhood is unstudied. The status of handicapped children in mainstreamed
classrooms is generally not good, and observational studies are now being addressed to the
attributions and attitudes that may be responsible (see Hartup, 1983). Overall, though, we
know relatively little about the implicit personality theories of schoolchildren and scarcely
more about these matters in early adolescents. Recent work indicates the significance of
these implicit theories in interactions among adults, however, so concentrated work with
children is urgently needed.

Peer Relationships And The Individual Child

Poor peer relationships in middle childhood are characteristic of children who are at risk for
emotional and behavioral disturbances in adolescence and adulthood. Early childhood
assessments are not strong predictors of social difficulties in middle childhood (Richman et
al., 1982), but individual differences during the school years are correlated with subsequent
adjustment. Negative reputations and social rejection among elementary school children, for
example, are prognostic indicators of continuing rejection by schoolmates (Coie and Dodge,
1983) as well as poor mental health and psychosexual difficulties in adolescence and young
adulthood (Cowen et al., 1973; Roff, 1963; Sundby and Kreyberg, 1968). Prediction of
schizophrenic breakdown from peer status in middle childhood has not been demonstrated
since social withdrawal and isolation are themselves not stable through this time (Coie and
Dodge, 1983). Beginning in early adolescence, however, irritability, aggressiveness, and
negativistic behavior in peer interactions are characteristic of premorbid individuals (Watt
and Lubensky, 1976).
Consistent as these results are, it is difficult to interpret them. Childhood indicators of later
maladjustment include somatic disturbances, family difficulties, and school failure as well as
difficulties with contemporaries. Peer problems, then, may simply reflect general difficulties
in development and not be direct determinants of emotional disturbance. Whatever the
childhood antecedents of behavior disorders, difficulties with contemporaries may
contribute, on their own, to negative self-attitudes, alienation, and reductions in social
effectiveness. Poor peer relationships are among the most consistently reported "problems"
in the referral of children to mental health clinics (Achenbach and Edelbrook, 1981).
Consequently, the connections between family and peer relationships across time, as these
involve self- and other attributions, social isolation, and social competence, need to be
examined (following the example of one investigation, in which social isolation among 5-
year-olds was found to be predictive of social cognitive difficulties a year later and which at
that point were linked to peer rejection and withdrawal—Rubin et al., in press). New
documentation connecting peer adjustment in middle childhood to later negative outcomes is
not needed. Rather, the studies needed should be multidimensional examinations over time
of social development and life events culminating in undesirable outcomes.
Similar comments can be made about peer relationships and crime. In middle childhood,
"delinquents to be" have difficulties getting along with others and in treating others
courteously, tactfully, and fairly. These children are also less well liked by their
contemporaries (Conger and Miller, 1966; West and Farrington, 1973). Concordantly, self-
evaluations in early adolescence indicate that delinquents to be do not enjoy close personal
relationships with others, are less interested in organized activities, and are immature. Middle
childhood, then, may be a time in which precursors of criminal behavior are established,
including negative self-attitudes and alienation. Delinquency is not a well-differentiated
construct, however. These notions deserve examination in carefully designed longitudinal
studies.
Studies centered on the processes of peer socialization have been conducted mainly with
preschool children. The evidence indicates that modeling and reinforcing events are used in
this context with increasing deliberation in middle childhood (Hartup, 1983). What children
learn and how much they learn through these contingencies are difficult to specify and may
remain so, although it is clear that peer interactions involve prosocial as well as aggressive
experiences. Since more aggression occurs in child-child interactions than in other contexts,
and since endorsement of antisocial norms countenanced by other children increases in
middle childhood (Berndt, 1979), fine-grained analyses should be extended to naturalistic
situations. These studies are needed to specify the conditions associated with the emergence
of individual differences in assertiveness and aggression. Children do not become uniformly
more "conforming" in middle childhood (see Hartup, 1983), but other conditions
encouraging the development of antisocial behavior within the peer context may exist in
these years—for example, the combination of a negative reputation among one's peers and
one's own disposition to attribute hostile intentions to others (Dodge and Frame, 1982).
Sexual socialization in middle childhood is understudied, mainly became of the difficulty of
conducting relevant studies. Early childhood studies make it clear that child-child
interactions extend the sex typing of the individual; playground observations of
schoolchildren support this thesis (see above). But sexual knowledge and experimentation
also derive from contacts with other children (Kinsey et al., 1948). The scarcity of
information on this subject, however, is disturbing—consider, for example, the need to
determine the circumstances contributing to sexual socialization as antecedents of adolescent
pregnancy.
Several investigators (see Thorne, 1982) have commented on the sexual character of
children's play, especially in the robust, physical interactions of boys. Sexual concomitants in
child-child interactions are common and explicit in conversations on the playground. It is
difficult to discount the normative significance of these events or to ignore their possible
contributions to sexual socialization. Close examination of sexuality in child-child
interactions could illuminate many critical issues. Same-sex contacts in middle childhood,
for example, may contribute both sexual knowledge and normative "stylistic" elements to the
child's repertoire, e.g., braggadocio among boys. Opposite-sex encounters may constitute
sexually neutralized introductions to the complementarities needed in heterosexual
communication.
Moral relativism was believed by Piaget (1932) to arise within child-child interactions in
middle childhood. Recent studies demonstrate that children's conversations modify their
moral judgments (Berndt et al., 1980), but it has been difficult to connect peer interaction in
the natural context with the maturity of the child's moral orientation. Preadolescents who
belong to clubs and social organizations receive higher scores on moral judgment than do
those who belong to relatively few organizations (Keasey, 1971), but the correlational nature
of this evidence reduces its significance. Children's conversations about moral issues need
closer examination, and content analyses centered on moral interaction in natural settings
should be encouraged. Whether the child's moral orientation is tied especially closely to peer
interactions may not be the most relevant issue. More important, we need to better
understand those processes in child-child interactions that have moral implications. Along
with knowing what children do when they are on their own, this information could contribute
greatly to knowledge about the role of child-child relationships in moral socialization.

Family And Peer Relationships

Familial Correlates Of Peer Competence

In early childhood, secure attachments between a child and his or her caretakers promote
exploration of the environment, including the other children who inhabit it. Mothers arrange
contacts between their young children and other youngsters, believing this to be desirable. In
addition, social interactions within the family promote individuation and the growth of self-
esteem—conditions that maximize the chances of success once peer interaction begins.
Consistent with these notions are empirical studies showing secure attachments in the first 2
years to be antecedents of sociability, empathy, and effectiveness in child-child relationships
at ages 3 and 4 (Waters et al., 1979).
The connections between family and peer relationships extending into middle childhood are
not well documented. One would expect parent-child relationships marked by emotional
support and appropriate demands for compliance to continue to be associated with positive
outcomes in child-child relationships. And, indeed, cross-sectional studies reveal that
mothers and fathers of well-liked children are emotionally supportive, infrequently
frustrating and punitive, and discouraging of antisocial behavior in their children (Winder
and Rau, 1962).
Other investigators (Hoffman, 1961) have observed the antecedents of self-confidence,
assertiveness, and effectiveness with other children to include (a) among boys, affection
from both mothers and fathers accompanied by dominance from fathers but not mothers, and
(b) among girls, affection from both parents accompanied by dominance from mothers but
not fathers. In addition, sociometric status is known to be positively correlated with parental
affection and the absence of family tension (Cox, 1966) as well as with parents' satisfaction
with their children (Elkins, 1958). No evidence suggests that the child-rearing correlates of
social competence in middle childhood differ in any substantive way from their correlates in
early childhood, even though specific exchanges between parents and their children that can
be called supportive or dominant undoubtedly change.
Assuming that family and peer relationships are interactive in individual development, one
would expect disturbance in one context (e.g., the family) to disturb the child in the other
(e.g., child-child relationships). For example, unemployment, terminal illness within the
family, and family conflict would be expected to affect the child's functioning with other
children. Scattered evidence is consistent with this notion. Girls ages 10-13 whose fathers
were unemployed for a substantial time during the depression reported more strain in their
relationships with age-mates, less self-confidence, and more concern about having friends
than did girls from less deprived families (Elder, 1974). Time spent with peers and
popularity, however, did not differ between girls in these two types of families. Perturbations
traced to the father's unemployment were not evident among boys. The data suggest that the
earlier maturing of the girls and their greater concern with grooming and attractiveness may
underlie these findings. In addition, boys may have been buffered from the impact of their
father's unemployment by their own increased work activities.
Divorce and its impact on socialization in the peer context have not been examined
extensively with children ages 6-12. Studies of younger children (e.g., Hetherington, 1979)
indicate that age-mate relationships suffer initially with the occurrence of divorce and that
"recovery" coincides with stabilization of interpersonal relationships within the family.
Young boys are more at risk for peer difficulties in the aftermath of divorce than young girls
are—the latter apparently receiving more frequent emotional support from their teachers and
mothers.
One interesting notion is that in middle childhood friends and other associates assist in the
amelioration of the anxiety associated with divorce, in the resolution of loyalty conflicts, and
in coping with the economic and practical exigencies deriving from the divorce. One
investigation suggests these dynamics among boys but not girls (Wallerstein and Kelly,
1981). In this instance, school-age boys were able to turn to friends, seemingly to put
distance between themselves and the troubled household. Girls, however, entered into
friendships only when their relationships with their mothers were supportive. Otherwise,
girls felt it necessary to abstain from interacting extensively with other children and were
constrained from entering friendships. This sex difference was not evident among preschool
children nor among adolescents in this investigation. Thus, the peer concomitants of divorce
in middle childhood need further examination. The extent to which children use their friends
instead of or in addition to their families for emotional and social support in middle
childhood is an interesting question, especially since those adaptations may be sex linked.

Parents Versus Peers: The Issue Of Cross-Pressures

The conventional wisdom stresses that the carryover from peer relationships to family
relationships in middle childhood is mainly in the form of increased opposition between
children and their parents. According to this argument, exposure to age-mates erodes the
child's orientation to the family and establishes nonnative opposition. The evidence,
however, does not suggest that these oppositions are especially intense before puberty; the
opposition revolves mainly around antisocial norms. From one investigation (Berndt, 1979)
with children in grades 3, 6, 9, and 11, the most striking results were (a) small decreases in
conformity on prosocial issues with both parents and friends, (b) a gradual decline in
conformity with parents in neutral situations but little change in conformity to peers, and (c)
an increase in peer conformity to antisocial norms between grades 3 and 9 but not beyond.
Children thus continue to use their parents as well as their friends as anchors for prosocial
activity, disengaging from their parent's mainly as normative anchors in antisocial activity.
Other investigators have observed that peer-endorsed standards of misconduct become
increasingly salient from grade 3 to 6 and from grade 6 to 8 but not beyond (Bixenstine et al.,
1976). The major age changes occurring in response to cross-pressures thus involve
antisocial norms; normarive opposition increases as puberty approaches.
Changes in the general attitudes of children toward parents and peers are similar: (a) attitudes
toward both parents and peers are more favorable than unfavorable at all ages, (b) the
number of children reporting positive attitudes toward parents declines somewhat during
middle childhood and an increase occurs again in middle adolescence, and (c) there is no
general increase in the favorability of attitudes toward peers (Harris and Tseng, 1957). Other
than a slight dip in the popularity of parents in preadolescence, there is thus no indication
that parents are increasingly rejected nor peers increasingly accepted during middle
childhood, although individual differences may be wide. By early adolescence, most
individuals are able to synthesize their understandings and expectations of their parents and
their peers.
Taken together, the literature connecting family and peer relationships is narrowly focused.
We know the dimensions in child-rearing that predict sociometric status in middle childhood
but little about changes in the family as these may bring about changes in child-child
relationships. We know that nonnative concordance and discordance change with age, but we
know little about the conditions that bring about these changes. We know little about the
strategies that parents use for knowing where their children are, arranging contacts with other
children, and coaching them in social skills. What attention is given by parents to children's
thinking about their companions? Do parents contribute to the child's increasingly
differentiated and "psychological" perceptions of their associates? Are there similarities
between the theories of personality and attributional conventions used by parents with their
children and those used by children with their associates? In what ways does intimacy within
family relationships carry forward into the close relationships emerging between children
and their friends? These issues cannot be addressed without concurrent studies in both the
family and the peer contexts.

The Peer Context And The School

Classrooms are social units of major significance in Western cultures. Teachers establish the
climate in these contexts, setting the conditions for social interactions and relationships.
Child-child relationships, however, may constitute the "social frontier" in the classroom
(Minuchin and Shapiro, 1983). In addition, child-child relationships within the classroom
and the school are learning contexts; children teach things to one another, and these
interactions contribute to children's growing understanding of the conditions under which
people work and achieve.

Classroom Conditions And Peer Interactions

Numerous classroom conditions influence child-child interactions. The number of students,
the physical arrangements, open versus traditional classroom structures, curriculum content,
and teaching style are known to be correlated with variations in children's interactions with
one another. Studies of these setting conditions are not easy to design, however, owing to the
common confounding of these conditions with one another. For example, friendships and
cliques are more numerous in large classes than in smaller ones (Hallinan, 1976) and group
activity is more frequent. And these effects depend on the extent to which teachers organize
small classes differently from larger ones (Smith and Glass, 1979).
Social interactions in open and traditional classrooms are not the same. Child-child contacts
are more numerous in open classrooms, involving both work-oriented and social matters
(Minuchin, 1976), and cross-sex and cross-age contacts are more frequent. Cooperative
interactions are more common in open classrooms, since cooperative work opportunities are
scheduled more often than in traditional settings. Cooperation in out-of-class situations,
however, is also more common among children enrolled in open than in traditional
classrooms. The induction of generalized cooperative expectations may thus be one outcome
of experience in these situations. Nevertheless, the extent to which children create a
cooperative ethic on their own in open classrooms is not known. Open classrooms provide a
basis for friendship selection that differs from conditions in traditional classrooms. Hallinan
(1976) found relatively rigid sociometric hierarchies in traditional classrooms along with
clear-cut consensus concerning the identities of popular and isolated children. More diffuse
social organizations were observed in open classrooms, with unreciprocated choices
occurring less commonly and persisting over a shorter time than in traditional settings. Other
investigators (Epstein, 1983) report that more students are selected and fewer are neglected
as best friends in open than in traditional situations, with sociometric choices that are more
commonly reciprocal. Open settings thus seem to encourage the continuing reorganization of
close relationships to a greater extent than traditional settings do.

Curriculum Content
Curricular interventions centered on socialization consist of four main types: moral
education, affective education, cooperative learning, and social skills training. Moral
education has been studied in numerous variations ranging from the use of lessons that
emphasize appropriate moral attitudes and behavior to the creation of ''moral schools."
Sometimes moral education consists of discussions about moral principles among the
children themselves; other times it consists of the incorporation of moral issues into the
curriculum in social studies or literature courses. Lockwood (1978) considered many of these
studies to be poorly executed, although well-designed investigations demonstrate that (a) the
direct discussion of moral dilemmas results in small advances in the maturity of moral
reasoning among children; (b) these advances are more common among younger children
than older ones, although individual differences are considerable; and (c) questions remain
about the persistence of these advances and their manifestations in behavior. There has been
little systematic evaluation of "moral schools" and virtually no effort to document the effects
of moral education on child-child interactions. Many unresolved issues remain in evaluating
moral education, including the effects of these interventions on children's interactions with
one another. Moreover, these issues have been recognized for some time. It is nevertheless
curious why we know so little about the effects of moral education on children's relationships
with one another.
Model programs of affective education have been used with elementary school children,
including the Human Development Program (Bessell and Palomares, 1970), the Affective
Education Program (Newberg, 1980), and-the Empathy Training Project (Feshbach, 1979).
These and other models emphasize group dynamics, social values, and personal adjustment.
Affective education has spread widely through U.S. schools, although definitive evaluation
of the impact of these programs on child-child relationships is scarce. Again, Lockwood
(1978) evaluated the evidence as showing positive effects on classroom behavior but
inconsistent effects on self-esteem, self-concept, personal adjustment, and social values. The
Empathy Training Project demonstrated decreases in rated aggression among children in the
program and cognitive gains among those children showing the greatest gains in prosocial
behavior and the greatest decreases in aggression. Further work with this program is needed,
since its effects in nonexperimental settings have not been documented.
Cooperative learning environments promote friendly conversation, sharing, and helping
among children, with the reverse being the case in competitive settings (Stendler et al.,
1951). Peer tutoring occurs more frequently under cooperative than competitive conditions
(DeVries and Edwards, 1972), and altruism occurs more frequently following cooperative
than competitive experience (Johnson et al., 1976). In addition, attitudes toward oneself and
one's coworkers are more positive as a consequence of cooperative rather than competitive
experience. Cooperative classrooms are also more cohesive social units than competitive
ones. Cohesiveness in racially integrated classrooms is evident when cooperative experiences
prevail, although it is necessary for the contributions of minority children to be recognized as
essential to class success in cooperative tasks in order for this to occur (Aronson, 1978).
Social skills training has been used with schoolchildren mainly in an effort to improve the
status of isolated and withdrawn children. Numerous interventions have been tried, most
based on the hypothesis that such children have difficulties in peer relationships because of
their inadequate social skills, e.g., communication skills. Some of these interventions have
been based on modeling; others have involved "unprogrammed" opportunities for withdrawn
children to interact with better skilled companions. Coaching, which is an intervention that
combines direct instruction, opportunities for rehearsal, and corrective feedback, has been
used extensively.
The efficacy of modeling and unprogrammed socialization strategies has been demonstrated
most thoroughly with preschool children. One investigation revealed that modeling
techniques are effective in improving the social status of third- and fourth-grade children
(Gresham and Nagle, 1980). Coaching studies have been variable in their outcomes (Combs
and Slaby, 1977; Conger and Keane, 1981), but these techniques do seem to be effective in
improving the sociometric status of isolated children and in some cases increase the
frequency of the child's social contacts. Long-term maintenance of these outcomes has been
assessed (Oden and Asher, 1977), although effects outside the school are unknown and
effects on measures other than sociometric tests and classroom observations are not well
documented. More serious, however, is the scarcity of developmental studies in this area. We
know that training in social skills can be effective, but the extent to which outcomes are
generalized outside the classroom and the extent to which developmental modifications need
to be made in the interventions themselves remain to be evaluated.

Children As Teachers
Believed to benefit both tutor and tutee, peer tutoring has been viewed as a cost-effective
instructional supplement in classrooms and as a basic element of socialization in certain
cultures—e.g., the USSR. Empirical studies have mostly concerned the outcomes of the
tutoring experiences—either for the child doing the teaching or for the child being taught.
Benefits to the tutor are believed to include increases in motivation and task involvement that
lead to gains in school achievement. Enhancement of self-esteem, prosocial behavior, and
attitudes toward school are also cited as tutor benefits. The evidence is not entirely consistent
in relation to these outcomes, and there is no obvious reason for the inconsistencies (Hartup,
1983). Tutee benefits are more clear-cut. The training of tutors must be carefully
accomplished to maximize tutee outcomes, and maintenance regimes must be closely
monitored (see Allen, 1976). Nevertheless, children clearly can teach one another a variety of
subjects.
Very little effort has been made to determine the techniques that children use to teach one
another. We know that children prefer to teach younger children and, conversely, to be
taught by older children. Same-sex tutors and nonevaluative instructional conditions are also
preferred (Lohman, 1969). Little is known about the strategies that children use in teaching
one another or how strategies vary according to setting. The weight of the evidence suggests
that peer teaching resembles adult pedagogics. Cooper et al. (1982) observed that issuing
directives, describing the task, and making evaluative comments were the techniques most
commonly used in classrooms; demonstration, labeling, pointing, questioning, praise, and
criticism were common, too. Kindergarten children were more directive and intrusive than
were second graders (especially when the children were asked to assumed a tutor role), but,
since these observations were conducted in same-age situations, it is not clear whether the
age differences were a function of the developmental status of the tutors, the tutees, or a
combination of the two.
As it turns out, school-age children make a variety of instructional accommodations to the
age of their tutees. Children instructing younger children use repetitions, strategic advice,
progress checkups, direct assistance, and praise more frequently than children who instruct
same-age tutees (Ludeke and Hartup, 1983). Children seem to possess ''implicit theories" of
teaching that assume younger children to require more cognitive structuring and more
supportive and corrective feedback than same-age children. These theories have been studied
only in relation to the actions of older children with younger associates, not vice versa;
nothing is known about "upward" accommodations. Again, information is not available
concerning these accommodations in relation to the magnitude of the age difference between
children. Tutoring strategies are more elaborate when the difference between tutor and tutee
is 4 years rather than 2 years, but nothing else is known.

Peer And Teacher Norms

Some investigators (Coleman, 1961) have regarded cross-pressures between peer and teacher
norms as major sources of tension and dysfunction in schools. Certain evidence is consistent
with this notion. Peer standards of misconduct are more readily endorsed by children when
these endorsements will be secret than when they will be revealed to parents and teachers
(Devereux, 1970). It is also clear that friends are sources of significant variance in the use of
leisure time and decisions about whether to smoke or use drugs. But the issue is more
complex. Surveys and questionnaires do not reveal that either middle childhood or
adolescence is a stormy period of normative dissonance. The notion that adult-child
relationships are understood by children to require self-control while peer relationships are
based on self-indulgence and unbridled instinctual activity is not substantiated (Emmerich et
al., 1971).
A small number of studies tell us about the tacit rules that children use to govern their
behavior in the classroom and their notions about the manner in which classrooms work.
These studies are largely normative and not addressed to the manner in which children
acquire these rules. Nucci and Turiel (1978) observed social transgressions in nursery
schools, interviewing the children about these incidents. Their distinctions between
conventional transgressions (e.g., playing in the wrong place) and moral transgressions (e.g.,
taking something that belongs to someone else) agreed most of the time with the distinctions
made by adults. Second, fifth, and seventh graders also make these distinctions (Nucci and
Nucci, 1979), reacting to conventional transgressions with comments about the rules and to
moral transgressions with arguments about the intrinsic implications of these events.
Moreover, school rules are seen by children according to these same distinctions; for
example, rules about harming others are distinguished from conventions about dress. Most
children believed that rules about doing harm are necessary.
Teachers do not always react consistently to conventional and moral transgressions;
conventional transgressions are sometimes treated as moral issues and vice versa (Nucci,
1979). The extent of these incongruities and their effects on children (especially their effects
on the teacher's credibility) are unknown. And, more broadly, we know little about the
effects of different patterns of school authority and organization on children's understanding
of social rules and the internalization of responsibility norms.
Missing, too, is information about the child's distinctions between parent-endorsed norms
and teacher-endorsed norms. Most of our attention has been given to the differentiation
occurring in middle childhood between adult-endorsed (i.e., parents and teachers combined)
and peer-endorsed norms. Nevertheless, parents are not teachers. Compliance demands,
conduct expectations, and the contingencies involved in the expression of affection and
support differ in families and classrooms. Consequently, a differentiated examination of
children's reactions to adult authorities needs to be undertaken in order to understand the
connections between the family and the peer system, on one hand, and between the peer
system and the school, on the other. Investigations focused explicitly on concordance and
discordance between teacher-child and parent-child interactions are essential.

Methodological Issues
Socialization in the peer context confronts the investigator with numerous difficulties in data
gathering. Neither as open to surveillance as preschool children nor as articulate as
adolescents, school-age children are elusive quarry. Trained observers are an alien presence
in the peer context; children bar them from access to activities with their companions, and
observers respect the child's rights to privacy. Moreover, the intrusion of observers into the
peer context unquestionably alters the events that occur there.
Nevertheless, one can argue that we have not been as creative as we might be in examining
child-child relationships outside the school. First, more effective use can be made of those
individuals whom we select as informants. Children themselves can be involved in many
ways other than to complete Guess Who tests, sociometric nominations, checklists, and
questionnaires. Recent studies (Youniss, 1980) suggest that the child interview has been
underused as a means of gathering data on a variety of timely and theoretically relevant
issues; children between ages 6 and 12 can be articulate about many issues. The nuances of
sexual socialization may never be revealed in response to questioning by adult examiners,
but the structure of the child's theories of interpersonal relationships might be.
Children can be used as observers of their own actions and the actions of their companions.
One cannot expect children to carry clipboards and stopwatches to their hideouts or their
playgrounds, but child observations can be accumulated in other ways. For example, the
telephone can be used to obtain information about recent events, the circumstances under
which the events occurred, their content, and their outcomes. One would expect these
observations not to be as "clean" as those of trained observers, but no one knows the exact
strengths and weaknesses of this strategy. Telemetric techniques can be used, too—both to
gather time-use information and to gather information about the attributions and affects
experienced in social interactions. To be sure, these technologies do not solve the issues of
access and privacy that were mentioned, but their use would extend the range of settings in
which we work, thereby justifying an increased effort to use them.
Parents are underused observers of child-child relationships. Restricted to the events that
they can observe and to what their children tell them, parents nevertheless accumulate a
considerable fund of information about the activities of their children and their companions.
Diary records, an ancient and underused technique, are once again being utilized in studies of
social development (see Radke-Yarrow et al., 1983). Electronic modes of data collection can
supplement the written record in these efforts. Also, interviews should not be written off as
data-gathering devices.
What about the scientist as observer or experimenter? Participant observation may be
feasible in studying informal groups of adolescents, especially if the observer is sufficiently
youthful (see Sherif and Sherif, 1964). No 20-year-old graduate student, however, can pass
as a 10-year-old. Only more creative (and ethical) uses of "lurking" can be encouraged. New
work suggests that we have not exhausted the possibilities (see, for example, Thorne's 1982
ethnographic observations centered on cross-sex borderwork occurring in playgrounds,
hallways, and school cafeterias). Shopping malls and other sites have been used for
observations of adolescents. Why not use similar observational settings to capture certain
aspects of peer interaction among school-age children? These strategies are labor-intensive,
but there is little choice. "Quick-test" classroom assessments must give way to more complex
and time-consuming assessments of child-child interactions outside the classroom.
A recurrent theme throughout this chapter is the need for developmental studies, either
through cross-sectional or longitudinal analysis. Unfortunately, more is involved in this
effort than the assessment of children at different ages or tracking the necessary cohorts over
time. The construction of age-appropriate measures is a continuing need and a complicated
business. Sufficient attention is almost never given to psychometric issues and the
appropriateness of research designs for conducting developmental work in this area (see
Fischer and Bullock, in this volume). Investigators cannot avoid these issues, however, any
more than they can avoid the other complexities inherent in developmental research.

Conclusion
Middle childhood is a time of consolidation and extension of peer relationships rather than a
time of beginnings. Children make their initial contacts with other children in early
childhood; commerce with them, however, increases dramatically between ages 6 and 12.
Younger children understand certain things about the intentions and motives of other
children, but these are elaborated and used with increasing effectiveness in middle
childhood. Similarly, communication and the coordination necessary for engaging in
cooperation and competition are established in the preschool years, but new integrations
emerge among schoolchildren. Within the peer context, new content (e.g., sex) enters into
child-child interactions, but these issues are integrated into normative structures whose
precursors trace back to early childhood. Preschool children possess nascent notions about
friendships and their implications, whereas the capacities for engaging in intimate
interactions seem to emerge between 6 and 12. Younger children interact distinctively with
adults as contrasted with age-mates, but more elaborate differentiations emerge in middle
childhood within the social networks of the family, the peer context, and the school. Parent-
child interactions change to some extent as children increase their activities with other
children. Certain normative oppositions arise between parents and their children; issues
connected with supervision and compliance change. But parents and children work out
accommodations to these differences without changing the basic nature of their relationships
and usually without detachment from one another.
Middle childhood is a distinctive time. The years between 6 and 12 present new and insistent
demands for working out accommodations with other children—i.e., individuals who are
similar to the child in cognitive capacities, knowledge, and social experience. Children must
construct arrangements for working and playing with similar individuals governed by rules
that differ, in many ways, from the rules that govern their exchanges with dissimilar
individuals. Children must construct interactions with others on an equal basis and sustain
them across situations and across time. No theme, issue, or comer to be turned may thus be
evident in child-child relationships during middle childhood, but children must construct a
wider and more varied range of accommodations that "work" with age-mates. In short,
coming to terms with the peer context is itself a major challenge in the years between 6 and
12.