You are on page 1of 14

Developmental Psychology Copyright 1981 by the American Psychological Association, Inc.

1981, Vol. 17, No. 3, 275-288 0012-1649/81 /1703-0275 J00.75

Attention Structure, Sociometric Status, and Dominance:


Interrelations, Behavioral Correlates, and
Relationships to Social Competence
Brian E. Vaughn Everett Waters
Department of Education State University of New York
University of California, Los Angeles at Stony Brook

To evaluate the interrelationships and external correlates of three potential mea-


sures of social competence, children in a preschool class were observed during,
free play throughout an entire school year. Ranks based on distribution of visual
regard among peers, sociometric scores, and competitive dominance interactions
were derived from data collected during each term. In addition, observations of
play behavior and of aggression were collected throughout the year. Attention
rank and sociometric preference were highly correlated and were substantially
stable across terms. Dominance rank was not strongly related to attention or to
sociometric rank and was less stable across time. The stability of the play and
aggression data varied from category to category as did relationships between
these categories and the status measures. Evidence demonstrates that attention
rank is related to social competence and is not a function of disruptive behavior,
activity level, or proximity to adults. Furthermore, a relationship between visual
regard among peers and the social structure of the preschool play group is dem-
onstrated. It is suggested that social competence, broadly conceived, is a more
salient basis for social organization among preschool peers than either the dis-
tribution of attention per se or competition/dominance relationships.

Social competence constructs play an im- Despite considerable agreement as to the im-
portant role in developmental theory and portance of social competence in theory, re-
research. In addition, assessment and en- search, and clinical interventions, assess-
hancement of social competence are gener- ment remains problematic. In part, this is
ally given high priority in the design and because the domain of social competence is
evaluation of early education and interven- difficult to define. For example, Anderson
tion programs, especially for socially de- and Messick (1974) listed 29 different cat-
prived or cognitively handicapped popula- egories of variables that could be relevant
tions (e.g., Anderson & Messick, 1974; to social competence in young children. Each
Greenspan, 1979; Kleck, 1976; O'Malley, of these categories could be assessed in a
1977; Simeonsson, 1978; White & Watts, number of ways and at various levels of anal-
1973; Zigler & Trickett, 1978; Lee, Note 1). ysis. Valid and economical assessments of
the construct are difficult to settle upon.
Portions of the data presented here were first reported
Moreover, analysis of specific components
at the meeting of the Animal Behavior Society, Seattle, of social competence such as attention span,
Washington, June 1980. creativity, and knowledge of social norms
The authors would like to thank Susan Merrick, Dan- does not offer the broadband type of assess-
ise Beale, and Jill Hasselquist who aided in the collection ment necessary in exploratory longitudinal
and tabulation of these data. The cooperation and pa-
tience of the director, teachers, and staff of the nursery research and in many program evaluation
school at the Institute of Child Development are grate- designs. ->
fully acknowledged. In addition, thanks are extended Global ratings, sociometric measures, and I
to F. F. Strayer and W. W. Hartup who read and com- observations of social dominance relation- /
mented upon previous drafts of this manuscript. ships are the most common approaches to
Requests for reprints should be sent to Everett
Waters, Department of Psychology, State University of assessment of a broadly defined social com-
New York at Stony Brook, Stony Brook, New York petence construct. Each has its advantages
11794. and its limitations. Generally, ratings by
275
276 BRIAN E. VAUGHN AND EVERETT WATERS

teachers or observers have been easy tn oh- sponses to various questions, and, although
tain and agreement among raters has been research exploring such relationships will
ejsy to acmevej,e.g., Behar & Stringfield, prove extremely valuable, this makes gen-
1974; Kohn (jTRosman, 1972). Unfortu- eralizations across studies employing differ-
nately, ratings of social competence are often ent techniques extremely tenuous at this
influenced by a variety of conceptually in- time. Even closely related questions ("Who
dependent adult-valued variables. These in- do you like?" vs. "Who do you dislike?")
clude intelligence, social class, sex, activity can result in scores with different levels of
level, and manageability. Although it is pos- stability and different patterns of external
sible to control for such halo effects using correlates (e.g., Gottlieb, Semmel, & Veld-
partial correlation or regression techniques, man, 1978; Hartup, Glazer, & Charles-
this requires a considerable amount of ad- worth, 1967). Finally, the more economical
ditional assessment of both the children and a sociometric task, the more cognitive so^
the adjihX Consequently, such procedures pmstication it tends to assume or demand
ate-"fTot oftenmrplemented. , at the subject. Nominations are faster than
Sociometric measures avoid the problem are picture array tasks, but they assume that
of assessing social competence from the the child knows and searches the names of
adult's point of view by eliciting playmate each classmate before nominating them. Pic-
preferences or nominations for each subject. ture tasks allow name recognition to be
The variable measured by these procedures tested, but they still assume that the subject
is more literally popularity/unpopularity searchesthe array thoroughly and that rea-
than it is social competence, but it is rea- sonaoie comparisons are, made among tEe
sonable to assume that social, competence many stimuli. Complete paired comparison
underlies a peer's popularity/unpopularity tasks ensure that each child is compared
to some extent (see Asher & Hymel, in press, with each other child, and because they obr
for a recent review). The primary difficulties tain extensive information from each sub-
that arise in sociometric assessments concern ject, tney are mghly reliable. As a result,
(a) the choice of sociometric instruments paired comparison data tend to have stronger
(e.g., peer nominations, picture-sociometric stability and stronger external correlates
techniques), (b) the choice of the sociometric than do abbreviated tasks (see Hartup,
question ("Who do you like?" vs. "Who do 1970). Unfortunately, they require a great
you dislike?") or nominations for roles in deal of time even in small classes (Vaughn
games, and (c) the social and cognitive skills & Waters, Note 2).
necessary to complete various sociometric Recent applications of ethological theory
tasks reliably and meaningfully. Most socio- and methods to the study of children's
groups have suggested that assessments of
metric tasks used with preschoolers involve dominance hierarchies may offer global,
individual testing of subjects outside the nonintrusive child-based competence assess-
classroom. This involves at least some dis- ments (e.g., Savin-Williams, 1976). The pri-
ruption of class activities or of ongoing play; mary problems with the ethological ap-
furthermore, not all children are attentive proaches concern the extensive data nec-
or willing to participate if the task keeps essary to construct transitive dominance
them away for very long. There is also a hierarchies from observations of dyadic in-
trade-off between the level of participation teractions. In a recent study, Straver and
and the amount (and quality) of information Straver (1976) collected data on agonistic^
that can be collected from each child. If the interactions for over 6 weeks without ob-"
amount of time and/or the level of partici- serving dyadic interactions of this type among!
pation is limited, then only one or a few so- many ot the possible pairs of children. Even
ciometric questions can be employed, re- given the necessary data, however, the re-
sulting in a limit in the amount of information lationship between status based on domi-
collected from each child. nance in agonistic interactions and the con-
Unfortunately, little is known about the cept of social competence remains an
meaning of, or relationships among, re- empirical question. In nonhuman primates,
ATTENTION STRUCTURE 277

dominance relationships primarily concern being isolated from peers. The finding that
adult interactions and are known to vary attention rank is positively correlated with
depending upon the criterion used to assess imitation by peers has recently been repli-
the pairwise interactions of the individuals cated by Abramovitch and Grusec (1978).
involved (e.g., access to food, social partners, Furthermore, we (Vaughn & Waters, 1980)
sleeping places or other resources, direction have reported that children high in the ajt-
of submissive gestures; Bernstein, 1970; Loy, tention rank also facilitate or maintain play-.
1975; Rowell, 1974). In humans, the sub- activities among peers more often than do.
jects of such ethological studies are generally lower ranked children These correlates
children (e.g., Abramovitch, 1976; Sluckin overlap considerably the domain of social
& Smith, 1977; Strayer & Strayer, 1976), competence and suggest possibilities for the
not adults, and the emphasis is often upon assessment of social competence that were
a composite of all competitive/conflict in- investigated in the present research.
teractions (Savin-Willians, 1976). Unfortu- _ Thejjurppses of this report are (a) to pre-
nately, there are very few data upon which sent the methodology for assessing visual
to evaluate the empirical, competence-re- regard among preschbolxpeers, (b) to eval-
lated correlates of dominance assessments uate the temporal and situational stability
in human children. In view of the relatively of this measure, (c) to evaluate the conver-
low frequency of actual aggressive interac- gent and discriminant validity of this mea-
tions in most preschool groups, however, it sure in relation to sociometric and compet-
seems unlikely that dominance in agonistic itive/dominance assessments and to a variety
interactions is as salient a basis for social of competence irrelevant behavior, (d) to
organization among preschool children as it compare the play and aggressive behavior
may be among adult nonhuman primates correlates of attention rank, sociometric sta-
(Vaughn & Waters, 1980). tus, and dominance status, and (e) to test
A new and potentially useful approach to the hypothesis that visual regard among pre-
the assessment of social dominance has been school peers reflects social organization and
developed in a series of reports by Chance is more than merely a consequence of inter-
(1967), Chance and Jolly (1970), and active and play contacts. Each of these goals
Chance and Larsen (1976). In this approach, contributes to the validation of the attention
social organization is described in terms of rank measure as a useful index of social com-
"attention structures." Chance and Jolly petence.
(1970) argue that actual aggressive encoun-
ters are rare even among nonhuman pri-
mates and would afford a costly and inef- Method
ficient basis for social organization. JQjey^
Subjects and Setting
propose instead that social organization
within primate groups can be characterized Children itLajsinglejrescliool class served as subjects
m terms of the deployment of attention^ in this study. These children were observed over the
.among individuals. In an early application three terms of the school year. At the beginning of the
of this approach to the study of preschool preschool year there were 22 children in the class rang-
ing in age from 4 years 0 months to 4 years 9 months
children's play groups, Hold (Note 3) dem- (14 male and 8 female). The group size fluctuated over
onstrated that children can be ranked reli- the course of the year as children dropped from the class
ably in terms of the amount of visual regard
iga and were replaced; no more than two children, however,
direcTedToWard them by peers andjthaLthese~
fc were dropped from or added to the class in any one
ranks have a variety of behavioral correlates term. Twenty children were class members for all three
terms. The class met five mornings a week for 3 to 4
often attributed to dominance status. In ad^ hours. The socioeconomic status (SES) of the children
dltion, Hold reported that attention rank is was heterogeneous, though over half of them came from
significantly correlated with a wide range of professional status families. For each given term, all
interactive behaviorincluding initiation of children in the class were present for a minimum of 60%
social interactions, organizing and maintain- of the behavioral observations and for a minimum of
75% of the class sessions.
ing" play activities among peers, and being Observations were made in a variety of physical set-
imitated by peersand with low scores on tings: inside the classrooms, on the playground, and in
278 BRIAN E. VAUGHN AND EVERETT WATERS

a gymnasium. The classroom was organized in a tra- Therefore, these totals were combined for all analyses
ditional manner with a variety of play areas. The outside reported here. A minimum of 100 observational rounds
play area contained large muscle play equipment, sand- of the class were made during each of the terms. In the
boxes, grassy spaces, and a paved path for riding toys. spring term, 100 observations were made both indoors
The gymnasium was equipped with mats, ladders, and and outside on the playground for purposes of compar-
a variety of other sports apparatus. jison across settings. Rank was determined from the ab-
Children were observed during_free play periods "solute number of looks and glances received from peers
which varied in duration from 45 minutes to 2 hour's, during each term.
depending on the weather, the proposed schedule ^f Sociometric measures. Two different sociometric
events (e.g., field trips), and the inclination of tty- measures of interpersonal preference were administered
teacher. Opportunities for play outdoors were largely at the end of each term: a paired comparison positive
dependent upon the weather. Consequently, the bulk of preference measure and a picture array negative preF
the data was collected indoors, except during the spring erence measure. For the paired comparison technique,
term, when an equal number of observations were made head and torso photographs of each child were made f
both indoors and outdoors for an analysis of situational and duplicated and cards were prepared for each pos-
stability. sible pairing of children in the class [n pairs = (n chil-
dren) X (n children - l)/2]. Children were presented
Measures of Social Status with each pair, one at a time, and asked, "Which of 0
these two children do you especially like?" Each child's
Attention measure. Observations of the frequency
of visual regard among peers were collected throughout picture appeared an equal number of times on the left
the school year. Each child was observed for 10 sec and and on the right sides of the cards, and the order of
the name of each child looked or glanced at was re- presentation was such that a given child would not be
seen twice before all of the other children were seen
corded. A look was defined as an orientation of the face
and eyes toward another child for 2 sec or more. A once (Torgerson, 1958). This task was completed in two
glance was defined as a similar orientation of the face or three 15-minute sessions per child. Care was taken
and eyes toward another child for lessjhan 2 sec. A to make sure that the child knew the names of each
target child was only scored as contributing a single unit classmate prior to the administration of the cards. If a
of visual regard to any other child during a given 10- child's interest and attention appeared to wander from
the task, the session was terminated and resumed an-
sec interval, even though several looks or glances might
other day. One week after the completion of this task,
have been directed toward a single child. Target children
each child was shown a randomly chosen subset of 20,
were scored as contributing looks to as many different
pairs from the complete paired comparison deck. The
children per observation period as they actually looked
average test-retest agreement for these pairs was 75%
or glanced at. Observers were instructed to count only
(range 45% to 100%). A child's rank on this measure^
those looks and glances that were clearly directed toward
was determined from the total number of positive nom-^
another child and not toward a toy or other object held
inations received across peers. -r
or looked at by the other child. Instances when the ob-
server was not confident of the recipient of the target The second sociometric task was a standard negative
child's looks and glances, and instances when the ob- preference task (see Hartup et al., 1967; Moore, Note
server was not sure whether a look or glance was to 4). Each child was presented with an array of head and
another child rather than to a toy or object held by the torso photographs of his/her classmates and asked to
child, were recorded but were not included as data in find a child whom he/she "did not especially like." After Cr
this report. Analyses of these ambiguous instances did a child was nominated, her/his photograph was turned
not reveal significant differences among the children face down and the same question was repeated for a
(i.e., no individual children were more ambiguous in total of three negative nominations. This sociometric
their looking patterns than were their peers) nor were task was usually administered after completion of the 0
there differences among the observers with respect to paired comparison task. Rank status on this measure
the frequency of recording looks and glances as ambig- was determined in the manner suggested by Hartup et
uous. al. (1967). The nominations were weighted according
The order in which children were observed was ran- to order of choice; initial choices were assigned a weight
domly determined with the restriction that each child ot nve, second choices a weight of three, and third
present in class on a given day was observed once before choices a weight of one. Sums were computed for each
any child was observed twice by the same observer. Two child and the totals rank ordered.
observers were responsible for collecting the attention Competition status rank. Following Abramovitch
data throughout the year. Rater agreement as to the (1976) and Strayer and Strayer (1976), competition
recipient of a look or a glance was established and main- status rank was determined in three steps. First, dyadic
tained at 80% throughout the course of the study (based interactions involving competition for objects and lo-
on six rounds of tne class [12U to Ii2 observations] per cations (see abject struggles and displacement in Table
term, and based on the formula, total number of agree- 4) were tabulated and a winner was designated for each
ments/total number of agreements + total number of interaction. Second, a table of all possible pairs of sub-
disagreements). Rater agreement on coding a given in- jects was constructed and the proportion of interactions
stance of visual regard as a look or a glance was also in which each child dominated each other child was
80% (using the same formula). Rank order correlations calculated. Finally, a rank order was derived that max-
between total looks and total glances received were pos- imized transitive relationships. This was done for the
itive_and significant for each^erm (median r = .65). data within each separate term as well as for data pooled
ATTENTION STRUCTURE 279

from the indoor observations for the entire year. The Results
matrix for the pooled data is presented in Table 1.
Linearity and rigidity indices were computed as sug- Temporal Stability
gested by Strayer and Strayer (1976). Coefficients of
.81 and .80, respectively, indicate that the data sub- The temporal stability of attention rank,
stantially approximate a transitive linear hierarchy and positive and negative sociometric status, and
account for a large proportion of the observed outcomes competition rank from the indoor assess-
of competitive encounters. Children were most likely tq^
lose in competitive interactions with children rankedl ments was evaluated across terms of the
above them and to win in competitions with children school year (Fall-Winter, Winter-Spring,
ranked below mem' ~ Fall-Spring) using Spearman rank-order
Behavioral observations during free play. Obser- correlations. As indicated in Table 2, the
vations of free play behavior were made throughout the attention and positive sociometric ranks i j
school year. Individual children were observed for a 30-
sec interval (focal individual sampling using modified
were highly stable across adjacent terms_and \\
Hanson frequencies; see Altmann, 1974)._after which also rronTjhe first to the third term? The
the observer recorded the presence or abseiKC, of ,be- negative preference sociometric and the
liavior in 16 categories of events, interactions, and types , competition ranks were less stable across
oTpenavioTniieSe observations were made by observers adjacent terms. The negative preference so-
not simultaneously collecting visual regard observations.
Special emphasis was placed on the quality of pla^L ciometric, however, showed a moderate de-
(seven subcategories) and on aggressive or agonistic in- gree of stability from the Fall to Spring
teractions (five subcategories, with four of these scored terms (rs = .54), whereas the competition
separafely for initiation or outcome). In the present re- rank order was quite unstable over this time.
port, these play and aggression variables are used to
assess convergent validity of the various status rank
These results suggest that both distribu-
measures as indices of social competence. tion of attention among peers and positive
Three other categories (activity level, visibility, and sociometric preferences are established early
proximity to the teacher) are used to assess the dis- in the school year and remain stable. Out-
criminant validity of the visual regard measure. Activity^ comes of competitive interactions also sta-
level relative to children in the immediate vicinity of the bilize to some degree during the early part
target child was scored as greater (+1), the same (0),
or lower (1) for each 30-sec time sample in the ob- of the year and are moderately stable across
servations of free play behavior during the Spring term. adjacent terms. The ranks vary too much
Visibility was scored as the number of 30-sec intervals from term to term, however, to yield signif-
m whicB a child's behavior might have attracted atten- icant correlations from the first term to the
tion from peers by virtue of its topography alone. Events
scored in this category included active attempts to at- third. Finally, negative sociometric prefer-
tract attention (e.g., standing on play equipment and ences were especially unstable from the first
shouting), any episode of crying, being taken from the to the second term. Since positive prefer-
classroom by an adult for any reason, and other behavior ences were stable across this period, this lack
that observers judged might create a disturbance. Time of stability is not due to unfamiliarity with ^
near the teacher was scored as the number of 30-sec
intervals in which the child spoke to, was spoken to by, the children in the class. The instability of
was in contact with, or was within 5 feet (1.5 m) of the the negative preference during the early
teacher. terms probably reflects the long time nec-
The children were observed in predetermined random essary to associate relatively low frequency
orders with the restriction that no child was observed aversive interactions with particular peers.
twice by the same observer until all of the other children
present on that day had been watched once. The number Negative sociometric preference is moder-
of observers ranged from three to five over the course ately stable from the Winter to Spring terms
of the study. At the outset of the observations, interrater (and from Fall to Spring) and would be
agreement over the entire 16-category list averaged 929k somewhat more stable if the more reliable \
(range 70% to 100% for single categories). For the seven
subcategories of play, the average interrater agreement
complete paired comparison technique were
was also 92% (range from 80% to 100%); for the aggres- employed.
sion subcategories, the average interrater agreement was
85% (range from 75% to 100%). For the three categories Situational Stability
of discriminant criteria, rater agreements were 81%,
70%, and 95% for activity level, visibility, and proximity The situational stability of attention and
to the teacher, respectively. Capsule definitions for each
subcategory of play and aggression are presented below the competition rank orders was evaluated
in Table 4. A total of 390 observational rounds of the across indoor and outdoor assessments dur-
class were made over the school year. ing the Spring term using Spearman rank-
K)
oo
0

Table 1
Competition Status Matrix Based on Objective Struggles and on Displacement for the Pooled Indoor Data From the Three Terms
NN" JN BY AM WY VC ZY TR RL TA GE MS JD DK KA JO RY GN HL AN m

NN X +
JN + + + x . + z
BY X + X X X m
AM X X X X X X <
WY X X X X + X >
VC
ZY + +
X
X
X
X
X
X
++ X+ x + ga
TR X X X Z
RL + X X X X + + + X >
TA X X X + + X X X
0
GE X X X X + + x
M
MS + X X X X <
JD X X X + X X X X + g
DK X X X X X X X X X X X W
KA X X X X X X X X X H
JO X X X X X X X X X ^
RY X X + .x x >
GN X X X X X X X X X X g
HL X + X X X X X + X X X ?o
W
AN X X X X X X
Note. X = Row is dominant over column. + = An equal number of competitions won by each contestant.
' The pairs of letters across the top and down the left side of the table are the initials of the contestants.
ATTENTION STRUCTURE 281

Table 2 with negative sociometric preference tended


Temporal Stability of Status Rank Orders to be negative but was highly variable from
term to term (rs = -.03, -.62, -.24). At-
Fall- Winter- Fall- tention rank and competition rank were not
Measure Winter Spring Spring
significantly correlated during any term;
Attention .81** .86** .84** they were moderately correlated, however,
Positive for the pooled data from the entire year
sociometric .92** .92** .90** (r = .48, p < .05).
Negative
sociometric .16 .46* .54* Positive and negative sociometric scores
Competition .51* .52* .11 were negatively correlated during each term
and in the pooled data. The correlationsjsmLg-
* p < .05. **p< .01. gest betayn_2fl%-and 50% cjjmmonjari-
ance and do noi_nile out the possibility of
order correlations. The correlation between different patterns (as well as direction)^
attention ranks from indoor and outdoor ob- behavioral correlates for the two measures.
servations was .90 (p < .01). The correlation Neither sociometric measure was consis-
between the competition ranks from indoor tently correlated with competition rank.
and outdoor observations was .07 (ns). _

Correlations Among the Rank Measures Temporal and Situational Stability of


Play and Aggression Measures
Rank order correlations among attention
rank, positive and negative sociometric sta- The six play behavior categories and the
tus, and the competition rank from the in- nine aggression and competition categories
door observations during each term and for are defined in Table 4. The total number of
the entire year are presented in Table 3. In occurrences for each category for the entire
brief, attention rank was strongly correlated year (indoors + outdoors) is presented along
with positive sociometric scores throughout with percent agreement among independent
the year (rs = .63, .85, .71). The correlation observers. In addition, the table summarizes
the temporal stability of each behavior cat-
Table 3 egory from the Fall to Winter and from the
Correlations Among Rank Measures by Term Winter to Spring terms (indoor data only),
and Pooled Indoor Data for Entire Year* and the situational stability of each behavior
category across the indoor and outdoor ob-
Positive Negative servations for the Spring term.
Term Attention sociometric sociometric
Play behavior (except for the category
Fall isolate) occurred more often than did ag-
,*
Positive soc. .63** " gressive behavior. As a result, the mean
Negative soc. -.03 -.44* scores for individuals on the play categories
Competition .33 .06 .47*
may be somewhat more dependable than
Winter
<
those for aggression. This implies that cor-
Positive soc. .85** " relations of the aggressive categories across
Negative soc. -.62** -.65** v
Competition .17 -.14 .23 time and situation and with the rank mea-
sures will be somewhat attenuated relative
Spring to the correlations involving the play behav-
Positive soc. .71**"
Negative soc. -.24 -.68** V ior categories (see Waters, 1978, 1980, for
Competition .22 .18 .26 related discussions).
Pooled data
Positive soc. .73** v / Behavioral Correlates of Rank Measures
Negative soc. -.47* -.75** "
Competition .48* .11 .22
Correlations of attention rank, positive
' All rs are Spearman rho. and negative sociometric scores, and com-
* p < .05. ** p < .01. petition rank with the play and aggression
282 BRIAN E. VAUGHN AND EVERETT WATERS

Table 4
Temporal and Situational Stability of Play and Aggression Measures
Rater In- Fall- Winter-
Category Frequency agreement out Winter Spring

Play
Interactive play. Play with peers
initiated and/or directed by
children. 1,030 91% .83** .26 .69**
Talk. Conversation with a peer in the
absence of play or other activity. 919 .80** .63** .74**
Parallel. Playing near a peer who is
involved in a similar activity, but
without interaction. 2,725 89% .64** .07 .32
Independent parallel. Playing near a
peer who is involved in a different
activity, but without interaction. 582 .78** .58** .72**
Independent alone. Engaged in an
activity or game, but neither near
nor interacting with a peer. 693 85% .48* .62** .65**
Isolated. Apart from peers and no
constructive activity. 274 100% .88** .73** .65**
Aggression
Hazing. Teasing and physical, vocal, or
gestural threats, often unprovoked,
seem to be used to "get a rise" out
of the recipient. 362 85% .57** .62** .62**
Initiates hazing. 362 78% .54* .73** .63**
Specific hostility. Physical attacks or
fights leading to submission. May
also include high intensity
arguments not included under
hazing (above). 100 .51* .17 .49*
Initiates specific hostility. 100 .34 .32 .83**
Object struggles. Fights (physical or
verbal) over possession of a desired
object. 219 100% .48* .46* .21
Wins object struggle. Child possessing
object at the end of an object
struggle. 219 100% .29 .43 .48*
Displacement. Removal of one child
by another from a position, either
by threat or physical removal. 153 75% .56** .53* .14
Wins displacement. 153 80% .64** .50* .73**
Game hostility. Aggression or hostile
behavior in the context of a game
(e.g., billy goat gruff pushes a troll
from a bridge, laughs as troll falls). 163 .43 .22 .49*

*p < .05. **/>< .01.

categories for the indoor observations during not replicated in data from the first or second
the Spring term are presented in Table 5.1 term.
The principal correlates of the attention
measure are interactive play, conversation, 1
Because the stability of the rank orders and the be-
success in competition for location, and havior categories tended to increase over the three terms,
game hostility. Attention rank tends to be and because the situational stability of the competition
negatively correlated with the noninteractive rank was low, only the correlations for the Spring-in-
door data are presented here. Complete tables of the
play categories (independent parallel play, correlations between the behavioral categories and the
independent play alone, isolate). The mod- status rank orders for each term are available from the
erate correlation with initiating hazing was authors upon request.
ATTENTION STRUCTURE 283

Table 5
Correlations Between Behavior Categories and Rank Measures for Indoor Data From the
Spring Term
Positive Negative
Category Attention sociometric sociometric Competition

Interactive play .77** .70** ^ -.44* .07


Talk .80** .51* ^ -.13 .32
Parallel play .43 .67** ^ -.12 -.07
Independent parallel play -.33 .03 -.21 -.62**
Independent alone -.29 .05 .15 -.24
Isolated -.33 .06 -.17 -.36
Hazing .44* .21 .57** .33
Initiates hazing .44* .06 .64** .58**
Specific hostility .13 -.16 .69** .13
Initiates specific hostility -.11 -.33 .67** .05
Object struggles .41 .31 -.01 .02
Wins object struggle .20 .03 .16 .15
Displacement .31 .06 .37 .04
Wins displacement .49* .12 .39 .53*
Game hostility .58** .31 -.11 .14

* p < .05. ** p< .01.

The positive sociometric measure had sim- ficiently well established for direct conflicts
Iilar Interactive play correlates, but no cor- to give way to yielding, displacement, and
I relation with the noninteractive play cate- possibly submission2 (see Omark, Strayer,
l gories. The correlations with winning' & Freedman, 1980, for extended discussions
displacement competition and game hostility of the development and functions of domi-
were lower than for the attention measure, nance hierarchies).
and there were no significant correlations
with any other aggression category. The neg- Discriminant Validation of Attention
ative sociometric measure correlated most Rank
strongly with hazing, initiating hazing, spe-
cific hostility, and initiating specific hostility. To evaluate the attention measure as an
These categories of aggressive interaction index of social competence, it is necessary
have been found previously to be related to to demonstrate that the measure is not sub-
socialization and adjustment (Manning, Note stantially correlated with behavior that might
5) and are less related to social competence influence attention among preschool peers
than are the other aggression categories. but that is unrelated to the construct of so-
The competition rank measure was neg- cial competence. Much behavior could play
atively correlated with parallel play and had a role in this process of discriminant vali-
small negative correlations with playing dation. Activity level, visibility, and time
alone or in isolation. These results suggest near the teacher were selected as being es-
a correlation between competition and social pecially important because of their potential
contacts, though not necessarily with inter- influence on visual regard. The rank order
active play. Larger correlations with initi- correlations of attention rank with high ac-
ating hazing and winning displacement in- tivity level, visibility, and time near the
teractions reflect the relationship between teacher were .41, .36, and .25, respectively.
competition rank and dominance. The cor- These results indicate that visual regard re-
relation with winning displacement inter- flected in the attention measure was not pri-
actions and the lack of any correlation with 2
Tables of the correlations between the aggression
object struggles or winning object struggles behavior categories and the competition rank order for
suggests that by the end of the school year, the Fall and Winter terms are available from the authors
dominance relationships may have been suf- upon request.
284 BRIAN E. VAUGHN AND EVERETT WATERS

marily received in the context of these non- of each child to the rank of the children to-
competence-related activities.3 ward whom he/she directed visual regard.
Additional discriminant validation of at- This was done in three steps. First, every
tention rank as an index of social compe- look or glance from one child to another was
tence arises from further analysis of the be- given a score equal to the attention rank of
havioral context in which looks and glances the child receiving the visual regard. Second,
are received. It might be hypothesized that the mean of these scores was computed for
visual regard is primarily received from play each child by summing the scores and di-
partners as a necessary consequence of the viding by the total number of looks and
interactive contacts. If this were so, then glances given for each child. If the child
without regard to the quality of play or to looked predominantly at high-ranking chil-
the social competence of the interactants, dren, the looking scores would have a higher
children who played more (or were more mean than if she/he paid more attention to
sociable) would receive more looks. If most lower ranking peers. Third, each subject's
visual regard were scored in such a context, mean looking score was compared to the
then its value as an index of social compe- mean that would have been expected if the
tence would be uncertain, though not dis- child looked equally often at each of her/his
proved. In the present study, 7,800 30-sec peers. If children's attention is equally (ran-
observations of free play behavior yielded domly) distributed among peers, then the
1,949 instances of interactive play and con- looking scores will not differ from the ex-
versation (see Table 4). Since our coding pected value (i.e., the values for the class
scheme allowed the distribution of only one will be distributed around the expected
glance to a given individual during any one mean, some above and some below the ex-
10-sec observation period, 1,949 episodes of pected value). If more visual regard is paid
interactive play and conversations could to high-ranking children, as predicted from
yield a maximum of 5,847 (3 X 1,949) looks theories of attention structure, then the
or glances between partners. On the average, mean looking scores should be consistently
one look or glance was coded every 10 sec above the expected value. By plotting these
in our observations of visual regard; thus the scores along the rank order of the subjects,
total number of looks and glances distributed we can answer two important questions.
during 7,800 30-sec intervals would be First, is more attention directed at high
. 23,400. Yisual regard received from play- ranked children than would be expected by
* mates during interactive play or conversa- chance? Second, is rank determined pri-
tion could account for only 25% of the total marily by looks and glances received from
looks and glances given.4 one or two individuals, or is there some con-
sensus of attention in the group as a whole?
Attention Rank and Attention Structure The data illustrating the relationship be-
As noted above, rank in the attention or-
3
der was determined on the basis of the fre- A recent re-analysis of the sociometric and attention
quency of looks and glances received from data presented in this paper by Langlois and Vaughn
peers. Rank ordering on a frequency mea- (Note 6) provides further evidence of the discriminant
validity of the attention measure. In their study, Lan-
sure, however, does not necessarily imply glois and Vaughn found that although physical attrac-
that there is an attention structure as pos- tiveness was a significant predictor of sociometric status
tulated by Chance (1967). For example, it in this class, it did not correlate significantly with at-
would be possible to rank order the individ- tention received from peers.
Further, although it is not directly relevant to the
uals in the class even if visual regard were issue of discriminant validation for the attention mea-
distributed on an essentially random basis. sure, it is of interest to note that the correlation between
Such a rank order would, of necessity, have the frequency of being coded as less active than other
many tied ranks and restricted variance, but children
4
and attention rank was .67.
the rank order could still be derived. Although a large number of looks and glances are
received during play from onlookers (as opposed to play
To determine the relationship between an partners), these looks are not necessarily unrelated to
individual's visual regard of peers and his/ the quality of play or to the competence of the play
her attention rank, we have related the rank partners who are being observed.
ATTENTION STRUCTURE 285

tween each child's rank in the attention order Discussion


and the distribution of visual regard to peers,
using the data for all 400 rounds of obser- The results of this study provide useful
vation collected over the entire year, are pre- replications of previous research on socio-
sented in Figure 1. The horizontal axis is the metric status, dominance relationships, and
value of looks and glances that would be visual regard among preschool peers. In ad-
expected if a child's visual regard were dition, by relating these measures to one
equally distributed among peers. The mean another and to a range of external correlates,
looking score for the group is 2.85 (expressed we have helped to place each measure in a
as a deviation from the expected mean), in- useful context and have taken steps to clarify
dicating that on the average children do di- the interpretation of individual differences.
rect visual regard toward higher ranking
peers more often than would be expected by The evidence that both positive and neg-
chance, r(19) = 3.78, p < .01. In addition, ative sociometric status can be stable over
all of the values for this class are above the several months replicates and extends im-
expected value, suggesting that there is a portant studies by Asher, Singleton, Tinsley,
consensus concerning attention distribution and Hymel (1979) and by Hartup et al.
in the group. These results confirm the ex- (1967). The improved stability, especially
istence of an attention structure within this for the positive sociometric preference scores,
class and indicate that the attention measure suggests that the paired comparison tech-
reflects more than a simple rank ordering of nique is likely to be psychometrically more
the sums of looks and glances received from reliable than are traditional picture-board^
peers. To the extent that attention rank is sociometric instruments (e.g., McCandless
validated as an index of social competence, & Marshall, 1957). The pattern of correlates
these results are important for the hypothesis of positive sociometric choices reported here
that social competence is an important basis is consistent with those found by Hartup et
for social organization among preschool chil- al. (1967). Furthermore, the present evi-
dren. dencethat certain categories of aggression

I I I ~r ~r i i i i i i i i i . ri ~iiii
ceo 6
n 5 - -
\
W 4

S . ~~
o 3 -
n
ZQ_ ?
>x
G> 1 -
,1 1 f i il i 1 l l l l t l l i I l l i t i i j f . s i f i j l

fV _i
C/) I
UJ<j ~2 - -

0-4 - -

Sin -6
Z^-7 1 1 1 i i 1 1 1 1 1 1 1 1 1 1 1 1 1 1 !
I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 19 20
ATTENTION RANK OF CHILD DISTRIBUTING VISUAL REGARD
Figure 1. Relationship between attention rank and rank of peers toward whom attention is directed.
286 BRIAN E. VAUGHN AND EVERETT WATERS

are more salient than others in eliciting neg- play and conversational skills are often used
ative preferences-elaborates upon Hartup as indices of social competence (e.g., Fur-
et al.'s demonstration that negative socio- man, Rahe, & Hartup, 1979). The pattern
metric nominations are related to delivering of correlations in Table 5 suggests, however,
"negative reinforcers." that the correlates of social competence need
Our analyses of the competition data, fol- not beTimited to the realm of prosocial be-
lowing the procedures suggested by Strayer havior. Children in this class who received
and Strayer (1976) and Strayer (1980), sug- disproportionate amounts of their peers' at-
gest that dyadic interactions concerning ob- tention were involved in many aggressive and
jects and space can yield rank orders that competitive encounters; they did not, how-
are sufficiently linear and rigid to qualify as ever, necessarily initiate these encounters.
"dominance" hierarchies. These relation- The data suggest that these children were
ships are, however, less stable in groups of skilled in meeting aggression in kind, whether
young children than in analogous groups of in playful contexts or when objects or posi-
juvenile primates. tions were in dispute (thus the correlation
The correlates of attention rank reported between attention rank and rank in the com-
here replicate findings by Abramovitch petitive hierarchy for the pooled data).
(1976), Hold (1976; Note 3), and Vaughn These results do more than suggest the va-
and Waters (1980). Rank in an attention lidity of the attention measure as an index
structure is at least moderately related to
of social competence. They also help us de-
rank in a competition hierarchy. Nonethe-
less, the behavioral correlates of attention fine the competence construct along lines
rank are substantially different from the cor- that fit the actual behavior of preschool chil-
relates of the competition hierarchy. In ad- dren. Further, these results prompt us to
dition, attention rank is a strong correlate offer the speculation that one of the reasons
of positive sociometric status; the competi- these children are watched by their peers,
tion hierarchy is not. The finding that at- even those not interacting directly with
tention rank is a strong correlate of positive them, ig__that_ there is much to be learned
sociometric preference has now been repli- from these children. The finding reported by~
cated in four different classes in two coun- Xbramovitch and Grusec (1978) that chil-
tries (Vaughn & Waters, Note 2), suggest- dren ranked high in the attention order are
ing that the result is quite robust. The imitated more often than are other children
evidence that attention rank is established lends credence to this speculation.
early in the year and is extremely stable over Over a decade ago, Chance (1967) and
the entire school year extends previous find- Chance and Jolly (1970) proposed that at-
ings and suggests that attention rank can be tention is an important basis for social or-
used as an index of social organization from ganization in monkeys, apes, and humans.
very early stages of group formation. Hinde (1974) has challenged this view and
The findings concerning the consistency argued that attention is at best a conse-
of behavioral categories over time and con- quence of existing social organization. We
text replicate similar results reported by are inclined to agree with Hinde. Our data
Roper and Hinde (1978), especially with indicate, however, that the distribution of
respect to play categories. Evidence that cat- attention among preschool peers is a pow-
egories of play behavior are important cor- erful reflection of social organization. The
relates of group structure suggests that present evidence of competence-related be-
broad generalizations about social organi- havioral correlates for the attention struc-
zation may be possible from observations of ture measure lends support to our hypothesis
social interactions in children's groups. that social competence is a more important
Over and above the replications of pre- basis for social organization than is either
vious data, we have taken steps to provide attention or dominance. This hypothesis ad-
convergent and discriminant validation of dresses a basic issue in the study of social
attention structure rank as a measure of so- organization and now deserves high priority
cial competence. Frequencies of interactive in developmental and comparative research.
ATTENTION STRUCTURE 287

Reference Notes Gottlieb, J., Semmel, M. L, & Veldman, D. J. Corre-


lates of sociometric status among mainstreamed men-
1. Lee, L. C. Assessing communicative competence via tally retarded children. Journal of Educational Psy-
the repertoire model. Paper presented for the Na- chology, 1978, 70, 396^t05.
tional Panel on Head Start Profiles of Program Ef- Greenspan, S. Social intelligence in the retarded. In
fects on Children, funded by Administration for Chil- N. R. Ellis (Ed.), Handbook of mental deficiency:
dren, Youth, and Families, 1978. Theory and research (2nd ed.). Hillsdale, N.J.: Erl-
2. Vaughn, B. E., & Waters, E. Social structures in baum, 1979.
a preschool playgroup: Reliability, stability, and Hartup, W. W. Peer interaction and social organization.
behavioral correlates. Paper presented at the meet- In P. Mussen (Ed.), Carmichael's manual of child
ing of the Animal Behavior Society, Seattle, Wash-- psychology (3rd ed.). New York: Wiley, 1970.
ington, June 1978. Hartup, W. W., Glazer, J., & Charlesworth, R. Peer
3. Hold, B. C. L. Soziale Rangordnung in Kinder- reinforcement and sociometric status. Child Devel-
gruppen unter beonderer Berucksichtigung der rang- opment, 1967, 38, 1017-1024.
spezifishen Verhaltensweisen. Unpublished doctoral Hinde, R. A. The biological basis of human social be-
dissertation, Ludwig-Maximillian-Universitat, Mu- havior. New York: McGraw-Hill, 1974.
nich, West Germany, 1975. Hold, B. C. L. Attention structure and rank specific
4. Moore, S. A sociometric status test for young chil- behavior in preschool children. In M. R. A. Chance
dren: Manual of instructions. Unpublished manu- & R. R. Larsen (Eds.), The social structure of at-
script, Institute of Child Development, University of tention. London: Wiley, 1976.
Minnesota, 1973. Kleck, R. E. Issues in social effectiveness: The case of
5. Manning, M. Personal communication. June 1975. the mentally retarded. In M. J. Begab & S. H. Rich-
6. Langlois, J., & Vaughn, B. E. Physical attractiveness ardson (Eds.), The mentally retarded and society: A
as a predictor of sociometric choice and attention social science perspective. Baltimore, Md.: University
structure status in preschool children. Unpublished Park Press, 1976.
manuscript, Psychology Department, University of Kohn, M., & Rosman, L. A. A social competence scale
Texas at Austin, 1980. and symptom checklist for the preschool child: Factor
dimensions, their cross instrument generality, and lon-
References gitudinal persistence. Developmental Psychology,
1972, 6, 430-444.
Abramovitch, R. The relation of attention and proximity
to rank in preschool children. In M. R. A. Chance Loy, J. The descent of dominance in Macaco: Insights
& R. R. Larson (Eds.), The social structure of at- into the structures of human societies. In R. Tuttle
tention. London: Wiley, 1976. (Ed.), Socioecology and psychology of primates. The
Abramovitch, R., & Grusec, J. Peer imitation in a nat- Hague, The Netherlands: Mouton, 1975.
ural setting. Child Development, 1978, 49, 60-65. " McCandless, B. R., & Marshall, H. R. A picture so-
Altmann, J. Observational study of behavior: Sampling ciometric technique for preschool children and its re-
methods. Behavior, 1974, 39, 227-261. lation to teacher judgements of friendship. Child De-
Anderson, S., & Messick, S. Social competence in young velopment, 1957, 28, 139-148.
children. Developmental Psychology, 1974, 10, 282- O'Malley, J. M. Research perspective on social com-
293. petence. Merrill-Palmer Quarterly, 1977, 23, 29-44.
Asher, S. R., & Hymel, S. Children's social competence Omark, D. R., Strayer, F. F., & Freedman, D. G. (Eds.).
in peer relations: Sociometric and behavioral assess- Dominance relations: An ethological view of human
ment. In J. D. Wine & M. D. Smye (Eds.), Social- conflict and social interaction. New York: Garland,
competence. New York: Guilford Press, in press. 1980.
Asher, S. R., Singleton, L. C., Tinsley, B. R., & Hymel, Roper, R., & Hinde, R. A. Social behavior in a play
S. A reliable sociometric measure for preschool chil- group: Consistency and complexity. Child Develop-
dren. Developmental Psychology, 1979,15, 443-444. ment, 1978, 49, 570-579.
Behar, L., & Stringfield, S. A behavior rating scale for Rowell, T. E. The concept of social dominance. Behav-
the preschool child. Developmental Psychology, 1974, ioral Biology, 1974, //, 131-154.
10, 601-610. Savin-Williams, R. C. An ethological study of domi-
Bernstein, I. S. Primate status hierarchies. In L. A. nance formation and maintenance in a group of hu-
Rosenblum (Ed.), Primate behavior (Vol. 1). New man adolescents. Child Development, 1976, 47, 972-
York: Academic Press, 1970. 979.
Chance, M. R. A. Attention structure as the basis of Simeonsson, R. J. Social competence. In J. Wortis (Ed.),
primate rank orders. Man (new series), 1967, 2, 503- Mental retardation and developmental disabilities:
518. An annual review. Vol. X. New York: Mazel, 1978.
Chance, M. R. A., & Jolly, C. Social groups of mon- Sluckin, A., & Smith, P. Two approaches to the concept
keys, apes, and men. New York: Dutton, 1970. of dominance in preschool children. Child Develop-
Chance, M. R. A., & Larsen, R. R. (Eds.). The social ment, 1977,45, 917-923.
structure of attention. London: Wiley, 1976. Strayer, F. F. An ethological analysis of preschool social
Furman, W., Rahe, D. F., & Hartup, W. W. Rehabil- ecology. In W. A. Collins (Ed.), Minnesota Sympo-
itation of socially withdrawn preschool children sium on Child Psychology (Vol. 12). Hillsdale, N.J.:
through mixed-age and same-age socialization. Child Erlbaum, 1980.
Development, 1979, 50, 915-922. Strayer, F. F., & Strayer, J. An ethological analysis of
288 BRIAN E. VAUGHN AND EVERETT WATERS

social agonism and dominance relations among pre- The attachment construct and the organization of
school children. Child Development, 1976, 47, 980- behavior and development. In K. Immelman, G. Bar-
989. low, M. Main, & L. Petrinovich (Eds.), Development
Torgerson, W. S. Theory and methods of scaling. New of behavior. New York: Cambridge University Press,
York: Wiley, 1958. 1980.
Vaughn, B. E., & Waters, E. Social organization among White, B. L., & Watts, J. C. Experience and environ-
preschool peers: Dominance, attention, and sociomet- ment: Major influences on the development of the
ric correlates. In D. R. Omark, F. F. Strayer, & young child. Englewood Cliffs, N.J.: Prentice-Hall,
D. G. Freedman (Eds.), Dominance relations: Etho- 1973.
logical perspectives on human conflict. New York: Zigler, E., & Trickett, P. K. IQ, social competence, and
Garland, 1980. evaluation of early childhood intervention programs.
Waters, E. The reliability and stability of individual American Psychologist, 1978, 33, 789-798.
differences in infant-mother attachment. Child De-
velopment, 1978, 49, 483-494.
Waters, E. Traits, relationships, and behavioral systems: Received June 10, 1980

Manuscripts Accepted for Publication


Age Segregation in Children's Social Interactions. Barbara Rogoff (Department of Psychology, Uni-
versity of Utah, Salt Lake City, Utah 84112), Shari Ellis, and Cindy C. Cromer.
Age Changes and Changes Over Time in Prosocial Intentions and Behavior Between Friends. Thomas
J. Berndt (Department of Psychology, Box 11A Yale Station, Yale University, New Haven, Con-
necticut 06520).
Young Children's Understanding of a Speaker's Intentional Use of a False Utterance. Brian P. Ackerman
(Department of Psychology, 220 Wolf Hall, University of Delaware, Newark, Delaware 19711).
Apparent Sex Differences in Cooperation/Competition: A Function of Individualism. George P. Knight
(Department of Psychology, University of Arizona, Tucson, Arizona 85712) and Spencer Kagan.
Illusory Contingency in Children at the State Fair. John R. Weisz (Department of Psychology, University
of North Carolina, Chapel Hill, North Carolina 27514).
Parental Discipline as Affected by the Sex of the Parent, Sex of the Child, and the Child's Apparent
Responsiveness to Discipline. Richard H. Passman (Department of Psychology, P.O. Box 413, Uni-
versity of Wisconsin, Milwaukee, Wisconsin 53201) and Raymond K. Mulhern.
Separation of Home Intellectual Environment and Parent IQ as Determinants of Child IQ. Langdon
E. Longstreth (Department of Psychology, University of Southern California, University Park, Los
Angeles, California 90007), Beryl Davis, Linda Carter, Debbi Flint, Jeffrey Owen, Marie Rickert,
and Ed Taylor.
Syntactic Form, Semantic Complexity, and Short-Term Memory as Factors Influencing Children's
Ability to Acquire New Linguistic Structures. Robbie Case (Department of Applied Psychology,
Ontario Institute for Studies in Education, 252 Bloor Street West, Toronto, Ontario M5S 1V6, Canada)
and Meredyth Daneman.
Mean Scores for Kohn Social Competence Found Markedly Different in Two Samples. Raymond Y.
Demers (Department of Family Medicine, 4201 St. Antoine, Wayne State University, Detroit, Mich-
igan 48201) and Rose Skell.
The Importance of Nonshred (E,) Environmental Influences in Behavioral Development. David C. Rowe
(Department of Psychology, Oberlin College, Oberlin, Ohio 44074) and Robert Plomin.
Changes in Interpersonal Distances and Categories of Play Behavior in the Early Weeks of Preschool.
John D. C. Shea (Department of Psychology, University of Newcastle, New South Wales 2308,
Australia).

(Continued on page 312)

You might also like