Content uploaded by L. Alan Sroufe
Author content
All content in this area was uploaded by L. Alan Sroufe on Jan 06, 2019
Content may be subject to copyright.
The Significance of Gender Boundaries in Preadolescence: Contemporary Correlates and
Antecedents of Boundary Violation and Maintenance
Author(s): L. Alan Sroufe, Christopher Bennett, Michelle Englund, Joan Urban, Shmuel
Shulman
Reviewed work(s):
Source:
Child Development,
Vol. 64, No. 2 (Apr., 1993), pp. 455-466
Published by: Blackwell Publishing on behalf of the Society for Research in Child Development
Stable URL: http://www.jstor.org/stable/1131262 .
Accessed: 29/10/2011 13:25
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Blackwell Publishing and Society for Research in Child Development are collaborating with JSTOR to digitize,
preserve and extend access to Child Development.
http://www.jstor.org
The Significance of Gender Boundaries
in Preadolescence: Contemporary Correlates
and Antecedents of Boundary Violation
and Maintenance
L. Alan Sroufe, Christopher Bennett, Michelle Englund,
and Joan Urban
University of Minnesota
Shmuel Shulman
Tel Aviv University
SROUFE, L. ALAN; BENNETT, CHRISTOPHER; ENGLUND, MICHELLE; URBAN, JOAN; and SHULMAN,
SHMUEL.
The Significance of Gender Boundaries in Preadolescence: Contemporary
Correlates
and Antecedents of Boundary Violation and Maintenance.
CHILD
DEVELOPMENT,
1993, 64, 455-
466. Previous research has established the importance of gender boundaries as a normative
aspect of development in middle childhood. Here, the nature and importance
of gender bound-
aries as an individual differences construct
was explored. Ratings
of gender boundary
violation
and gender boundary maintenance were made of 47 10-11-year-old children participating
in a
series of summer day camps. These ratings
were supported by videotape-based behavior codings
of gender boundary
violating behaviors and by live observations of sheer number of associations
with members of the opposite gender. In addition, considerable external validation of these
individual differences was obtained. Children low on gender boundary
violation and (especially)
children high on boundary
maintenance were independently judged by camp counselors to be
socially competent. They also were found to be higher on a friendship
variable,
based on observa-
tion. Those who violated boundaries were especially unpopular with peers, based on a child
interview. Finally, boundary
violation and maintenance were related to attachment
history and
to early measures of parent-child generational boundary
distortions.
The literature on peers in middle child-
hood and preadolescence has been focused
largely on interactions and friendships
within gender groups (Hartup, in press).
This, no doubt, is due to the characteristic
and culturally universal segregation of the
genders that occurs prior to adolescence
(e.g., Maccoby & Jacklin, 1987; Schofield,
1981; Whiting & Edwards, 1988). However,
this relative neglect of cross-gender phe-
nomena has been unfortunate for a number
of reasons. What boys and girls do together,
as well as how they do it and what they do
not do, seem likely to be of developmental
significance (Maccoby, 1990).
As Thorne (1986) points out in her arti-
cle, "Girls and Boys Together... but Mostly
Apart," gender cleavage is not total even in
the preadolescent period. Moreover, al-
though direct interactions are rare, members
of the opposite gender are often of great in-
terest to children during this period. Ten-
and eleven-year-old boys and girls talk about
each other a great deal (including assertions
about who likes whom), and, at times, ob-
serve the actions of one another closely.
Also, when social commerce does cross gen-
der boundaries, the interactions are often
marked by high arousal and displays of
affect.
The salience of cross-gender interac-
tions also may be inferred from the elaborate
system of rules and rituals that surround
them (Thorne & Luria, 1986). At no other
age do boys and girls behave as though one
may be contaminated by mere interaction or
proximity with members of the other gender.
Claimed fear of "cooties" and "boy germs"
conveys in a playful way the seriousness of
the issues involved. So too do the swift and
certain consequences that occur when gen-
der boundaries are crossed. For example, a
This research has been supported by a grant from the National Institute of Mental Health
(MH 40864-05). Address reprint requests to: L. Alan Sroufe, Institute of Child Development,
University of Minnesota, 51 East River Road, Minneapolis, MN 55455.
[Child Development, 1993, 64, 455-466. ? 1993 by the Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/93/6402-0005$01.00]
456 Child Development
boy is seen leaving a girls' tent (where he
went to get his radio). All boys who witness
this begin to tease and taunt him ("Uuh, he's
with the girls!", "Did you kiss anyone Char-
lie?", etc.). He has to chase and hit each boy,
in turn, to reestablish his place in the group.
Like Thorne and Luria (1986) we observed
this and countless such examples in our
summer camp research (e.g., Elicker, En-
glund, & Sroufe, 1992).
In general, contact by children with
members of the other gender in public set-
tings is disallowed, yet there are numerous
exceptions to this "law," all of which are
widely understood by children (Maccoby,
1990). These include protection when ac-
companied by a same-gender peer, immu-
nity when there is active disavowal of inter-
est or when there is some form of "cover" or
excuse. Such hypothesized rules, abstracted
from our own experience and the work of
Thorne and Luria (1986), have been summa-
rized in Table 1. Were cross-gender interac-
tions of minimal importance or of little inter-
est to children, such an elaborate system of
rules and rituals would hardly be necessary.
It is our contention that preadolescents
face a complex developmental task. They
must harmonize two seemingly contradic-
tory goals. On the one hand, they must main-
tain a clear gender boundary, and, on the
other hand, they must find opportunities to
express their interest in, and learn about,
members of the other gender. We believe
both goals are important. Maintaining gen-
der boundaries likely functions to enhance
loyalty in same-gender friendships and to
promote acquisition of skills for group func-
tioning, as well as preventing premature ef-
forts at heterosexual intimacy. Thus, inti-
macy and capacity for functioning in groups
are first mastered in the less complex, same-
gender peer group, as suggested by Sullivan
(1953). Yet, exposure to, and learning about,
members of the other gender during this pe-
riod should reduce the challenges of hetero-
sexual friendship or intimacy in adolescence
(Thorne & Luria, 1986).
Herein lies the importance of the rule
system that governs cross-gender social con-
tacts. By adhering to this system preadoles-
cents can acquire experience with the other
gender, while still maintaining a distinct
boundary between genders; that is, they can
be together while still being apart. The rule
system serves a regulatory function. By pre-
scribing when contact may occur, in what
circumstances, and in what manner, it as-
sures that contact which serves development
may take place. It is for these reasons that
we emphasize assessments of gender bound-
ary violation and maintenance in the report
to follow.
The concept of a "gender boundary,"
while having physical manifestations, is a
psychological construct. Maintaining the
gender boundary does not prohibit mere
physical proximity or even interaction but,
rather, psychological intimacy or identifica-
tion with the other gender. Whether particu-
lar behavior violates or maintains the gender
boundary depends on context. Two children
may be right next to each other, but if sitting
back to back at separate lunch tables and tak-
TABLE 1
KNOWING THE RULES: UNDER WHAT CIRCUMSTANCES IS IT PERMISSIBLE TO HAVE CONTACT
WITH THE OTHER GENDER IN MIDDLE CHILDHOOD?
Rule: The contact is accidental.
Example: You're not looking where you are going and you bump into someone.
Rule: The contact is incidental.
Example: You go to get some lemonade and wait while two children of the other gender get some.
(There should be no conversation.)
Rule: The contact is in the guise of some clear and necessary purpose.
Example: You may say, "Pass the lemonade," to persons of the other gender at the next table. No
interest in them is expressed.
Rule: An adult compels you to have contact.
Example: "Go get that map from X and Y and bring it to me."
Rule: You are accompanied by someone of your own gender.
Example: Two girls may talk to two boys though physical closeness with your own partner
must
be maintained and intimacy with the others is disallowed.
Rule: The interaction or contact is accompanied by disavowal.
Example: You say someone is ugly or hurl some other insult or (more commonly for boys) push or
throw something at them as you pass by.
Sroufe et al. 457
ing no notice of one another, there is little
risk of boundary violation even though they
are only inches apart.
Even physical contact
may occur in certain circumstances; yet the
boundary may be maintained, if certain ac-
tions are taken. Risk of boundary violation
and boundary maintenance activities must
both be considered in seeking to understand
the boundary construct.
Prototypic of "engagement with protec-
tion" is what Thorne and Luria (1986) call
"borderwork." This term applies to the
ubiquitous taunting, teasing, name calling,
and chase games between boys' and girls'
groups seen on playgrounds. Such high en-
ergy contacts not only are generally brief,
hit-and-run
missions but are also accompa-
nied by outspoken disavowal of interest.
They occur at the boundary
between groups
and serve to proclaim the boundary through
verbal assertions and physical definition of
"sides." Yet, anyone who observes such
transactions cannot miss the high level of in-
terest and excitement that characterizes
them. Measures of mere frequency of con-
tact underestimate the importance of such
behavior.
Almost all past work on gender bound-
aries in childhood has been normative, ex-
amining behavior of boys and girls in gen-
eral. Maccoby and Jacklin (1987) did explore
the stability and correlates of individual
preference for same-sex playmates, finding
little support for this individual differences
construct. But these were 41/2-61/2-year-olds.
Perhaps gender boundary maintenance is
not yet a salient developmental issue at that
age. If maintenance of gender boundaries is,
indeed, a salient issue for the preadolescent
period, then it should be an important
arena
for assessing individual differences. Recip-
rocally, should individual differences with
regard to gender boundary maintenance (or
violation) have meaningful contemporary,
antecedent, and outcome correlates, this
would affirm
the importance of the construct
(Sroufe, 1991; Sroufe, Carlson, & Shulman,
in press). In our view, maintaining gender
boundaries is an important aspect of social
competence during preadolescence. There-
fore, we would predict that it would corre-
late positively with other contemporary as-
sessments of social competence, such as
counselor rankings, sociometric popularity,
and capacity to form friendships, and with
key antecedents of social competence (Wa-
ters & Sroufe, 1983). These predictions are
in sharp distinction to the alternative hy-
pothesis that preadolescents who frequently
cross gender boundaries are the precocious
leaders of the peer group.
The purpose of this paper is to present
procedures for assessing gender boundary
maintenance (and violation) in individual
children and to validate this construct
against antecedent and contemporary
corre-
lates, drawing on the data base of a longitu-
dinal study. The antecedents included as-
sessment of attachment in infancy and
assessments of parent-child boundary prob-
lems at 24 and 42 months. Attachment
classi-
fication is a robust predictor of later social
competence (e.g., Elicker et al., 1992) and
has been implicated in self-development
(e.g., Sroufe, 1990). The generational bound-
ary
assessments were specifically selected as
potential antecedents of gender boundary
violation. This prediction was based on a
presumed role of parent-child relationships
in promoting firm ego boundaries in the
child and the tendency for boundaries to be
violated across genders in families where
it occurs (Sroufe & Fleeson, 1988; Sroufe,
Jacobvitz, Mangelsdorf, DeAngelo, & Ward,
1985). Correlates of these individual differ-
ences in adolescent behavior await later data
collection, although this issue has been stud-
ied indirectly by Kagan and Moss (1962)
and, more recently, by Feldman, Rosenthal,
Brown, and Canning (in press).
Method
SUBJECTS
Subjects were selected from the Moth-
er-Child Project at the University of Minne-
sota, a 15-year
longitudinal study of children
at risk for developmental problems, due to
poverty at the time of the child's birth (Ege-
land & Brunnquell, 1979). Mothers of the
children were primiparous and were re-
cruited during the third trimester of their
pregnancy
from
public health clinics in Min-
neapolis. At the time of delivery, the moth-
ers were young (range = 12-37 years, M =
20.52, SD = 3.65) and mostly single (62%).
Forty percent had not completed high
school. Eighty percent of the mothers were
Caucasian, 14% were Black, and the re-
maining 6% were Native American or His-
panic. At the time of the 10-11-year-old as-
sessment 20% of the heads of household in
the subsample described below had profes-
sional, technical, or managerial jobs; 62%
were in clerical, crafts person, service or la-
bor jobs; 16% were unemployed; and 2%
were students.
458 Child Development
Forty-eight children were selected from
the larger sample for participation in three
4-week summer camps held in consecutive
years on the university campus. Sixteen chil-
dren participated each year; one child in the
third camp returned to his out-of-state home
after 5 days, leaving 47 subjects for the
study. Children for each camp were selected
on the basis of several criteria: attachment
classification (approximately equal numbers
of children with secure and anxious histories
and, as much as possible, children with re-
sistant and avoidant histories in each camp);
gender (equal numbers of boys and girls in
each camp and within secure and anxious
groups); race (approximately equal numbers
of Caucasian children in secure and anxious
groups); and age (to reduce age variation
across camps, the younger subjects in the
sample were selected for participation in
the last year). Mean age of the subjects at
the time of camp participation was 10 years,
11 months (range = 9 years, 7 months to 11
years, 8 months; SD = 7 months).
PROCEDURE
Setting and Camp Routine
Children attended the Minnesota Sum-
mer Camps 5 days a week, 41/2
hours each
day for 4 weeks. They were transported
daily by van or car between home and camp.
The attendance of the children was very
consistent (M = 18.3 days, SD = 1.8), and
there were few absences. The daily program
of activities was varied and interesting, in-
cluding group circle times, singing, snacks
and lunch, swimming, arts and crafts, and
outdoor games and sports. Weekly day trips
were taken to local recreation parks, and one
overnight camp-out was held at the camp
headquarters on the university campus.
Many of the camp activities involved all of
the children as a group, but there were also
opportunities for children to choose among
two or more activities and to select their own
companions. Observations took place during
all of these activities.
Each year's camp counselor staff was
composed of four or five master's level grad-
uate students and one or two advanced un-
dergraduate assistants, all experienced in
working with school-aged children.
Live Observational Data
The relative frequency of cross-gender
interaction was determined from live obser-
vations made during the camp. A child sam-
pling procedure was used, with observations
being made in all activities except for coun-
selor-structured activities. Frequency of ob-
servation of each child was approximately
equal in each activity within each camp. Be-
cause of some slight variations and occa-
sional larger variations (due to absences),
proportion scores were used. Each child was
observed, in turn, just long enough to deter-
mine whether the child was engaged with
anyone and the identity of the partner(s). An
average of 317 observations were made per
child over the course of the 4-week camps.
For each child several scores were derived.
The "cross-gender" score is the proportion
of times that each child, when observed, was
engaged with only members of the opposite
gender.
Videotape Data
A total of 138 hours of videotape were
available for this study, primarily high-
quality, close-in color video recordings.
These were edited to extract examples of in-
teraction or proximity between boys and
girls, excluding activities structured by
counselors, resulting in 438 events on ap-
proximately 7 hours of tape. An event sam-
pling method was used; therefore children
were not equally represented on the tapes.
Isolated children would be seen less fre-
quently, for example, though every child ap-
peared numerous times. These tapes were
the basis of behavioral codings and ratings
of gender boundary violation and boundary
maintenance (described below). Two inde-
pendent coders did all codings and ratings.
Coders viewed the tapes as often as they
wished before making their judgments. The
two ratings were made for children in all
camps; the behavioral codings, which are
time consuming and difficult, were made
only in the first camp. Although the ratings
represent the principal data, we believe that
effort in doing the behavioral coding helped
insure reliability on the ratings. Also, the rat-
ings are based on the behaviors in the cod-
ing scheme.
MEASURES
Behavioral Codes (Camp 1)
Nonviolating behavior.-Non-gender-
boundary violating codes included proxim-
ity or interaction which contained no ele-
ment of personal interest, was casual,
entailed active disavowal of interest, and/or
was in the context of some form of protec-
tion. Brief sketches of the specific codes are
as follows:
1. Proximity: The child is merely near
members of the opposite gender; no interest,
interaction or desire for interaction is ap-
parent.
Sroufe et al. 459
2. Onlooking: Here the child in fact
watches (but only watches) members of the
other gender. Interest in the activity may be
apparent,
but there is no expressed interest
in the group per se. Such episodes must be
brief, and the child does not follow the
group from place to place (in contrast, see
"hovering" below).
3. Relaxed interaction: This often en-
tails multiple members of each gender
("safety in numbers"). If there is one-on-one
interaction, it must be brief, casual, de-
manded by a counselor or have some clear
purpose, other than interaction per se (e.g.,
"Pass the lemonade"). It can only rarely be
one on more-than-one; then some form of
protection is needed. It is called "relaxed"
because tension is rare in such interaction.
4. Border work: This is interaction
where the apparent
motive is to disclaim in-
terest (see Thorne & Luria, 1986). It is often
brief and filled with tension (teasing, insults,
taunts, shoving, etc.). It may involve an indi-
vidual child but is often done in groups.
Gender boundary violating codes.-
These disparate codes have in common a
threat to the boundary, either because inter-
est in intimacy is expressed or one remains
too long in a vulnerable position. Codes are
as follows:
1. Hovering: The child stays oriented to
and near a group of the opposite gender
more than briefly, perhaps communicating
with them, and perhaps changing locations
with them. In some way interest in them per
se is conveyed and no protection (e.g., from
a same-gender counselor or peer) is present.
The child is not necessarily attempting to
join the group and rarely
appears
to be doing
SO.
2. Joining: A single child actually be-
comes part of an opposite gender group.
This would include moving into a certain
area (e.g., sitting at a lunch table with only
members of the other gender) or engaging in
interaction or an activity without protection.
3. Heterosexual meaning: This includes
intimate physical contact, flirting, sexual
gestures, or verbal sexual innuendoes ex-
pressed to members of the other gender. It
does not include statements made when in
the presence of only same-gender peers.
Boys and girls this age generally do not
touch, except as part
of border
work, and sex-
ual comments usually are made when no
members of the other gender are present.
This category does not include statements
about another party's interests (e.g., "Jenni-
fer likes you Greg") or even expressions of
liking deliberately sent through an emissary
(although we never observed this behavior).
One other category, "participating
with
members of the other gender" was consid-
ered ambiguous. Here the child is involved
with only members of the other gender in
some activity (e.g., waiting for a turn to dive)
but is not interacting with them. They may
be in "dangerous"
territory,
but the interest
seems to be in the activity. This was coded
but not included in the analyses. A complete
manual for these codings is available from
the authors.
Ratings of Violation and Maintenance
(Camps 1-3)
The seven-point Gender Boundary
Vio-
lation and Boundary
Maintenance Scales are
found in the Appendix. These scales are not
simply opposites, although one would ex-
pect a modest negative correlation. To get
the highest scores on maintenance, the child
must engage in behaviors that actively de-
fine the gender boundary, such as taking im-
mediate action to leave a threatening area,
calling in reinforcements when needed, as-
serting disinterest or dislike of opposite-
gender members, or assuming a protective
physical stance (as witnessed by one girl
who kept her elbows up when in close prox-
imity to boys). A child gets the highest score
on violation if he or she frequently violates
boundaries and expressed intimacy with the
other gender, regardless of whether also
showing maintenance behaviors at times.
On the other hand, a child may get a low
boundary violation score simply by having
no cross-gender contact. Some may have a
lot of contact but always maintain; an occa-
sional child may show a lot of each category
of behavior. Still, children showing high vio-
lation and low maintenance should be the
least competent children. The two scales to-
gether define the boundary construct.
Social Competence Indices
The rationale for selecting social com-
petence indices had two parts: (1) given the
limited sample size we limited the number
of variables used and (2) we wanted to sam-
ple the domain broadly. Therefore, rather
than using multiple counselor ratings,
obser-
vational indices, or child nomination mea-
sures, we drew one measure from each of
these sources and sought to tap a different
aspect of competence with each. We used a
counselor rating of social skill, an observa-
tional measure of friendship, and a peer-
interview-based measure of "popularity."
It
460 Child Development
was the case that each of these measures was
supported by data from the other domains;
for example, the observational friendship
measure (to be described) was strongly re-
lated to counselor judgments of friendship
and to reciprocated peer nominations of
friends (Elicker et al., 1992). We believed a
convergence of measures from different
sources would be the most adequate test of
our hypotheses.
Counselor judgments.-Each of four
counselors made ratings of each child on a
seven-point Social Skills with Peers Scale.
High scores on this scale involve more than
merely being socially active. Children re-
ceiving high scores skillfully modulate their
behavior to enhance interaction and to sus-
tain interest in activities. They enjoy social
relationships, establish close friendships,
and are comfortable in a wide variety of so-
cial settings. Data were composited across
counselors, yielding a highly reliable index.
Peer popularity.-A camp "exit" inter-
view was used to index the overall popular-
ity of each child. While the interview had as
primary foci the child's models of relation-
ships and social awareness, it also was possi-
ble to derive a within-gender popularity in-
dex (Elicker et al., 1992). In our procedure
each child was asked whom they especially
liked and disliked. Then, they also were
asked specifically about unnamed children
of the same gender. For each interview a
given child received a score of + 2, if sponta-
neously named as liked, +1 if described as
liked under questioning, 0 if regarded in
neutral terms, - 1 if described as not liked
under questioning, and - 2, if spontaneously
named as disliked. These are then summed
for each child across all interviews with chil-
dren of the same gender.
Observation "friendship" score.-We
had records of the proportion of times each
child was with every other child. The high-
est of these was taken as the friendship
score, under the assumption that associa-
tions would pile up between friends,
whereas children without close friends
would have more scattering of associations.
This measure is strongly related to both
counselor and peer judgments of friendship
(Elicker et al., 1992). We used this as our
behavioral measure of competence, rather
than total peer contact or total same gender
contact, because it is more independent of
the frequency of cross-gender association
score and controls for amount of sheer social
activity.
Early Childhood Measures
Attachment pattern.-At 12 and 18
months, all subjects were assessed in Ains-
worth's Strange Situation procedure with
their primary
caregivers. Cases were classi-
fied as anxious/avoidant (A), secure (B), or
anxious/resistant
(C) by trained pairs of cod-
ers following Ainsworth's procedures (Ains-
worth, Blehar, Waters, & Wall, 1978). In
those cases where the 12- and 18-month as-
sessments were not consistent (25%),
a pro-
file of behavior in a tool problem assessment
at 24 months, previously validated against
attachment classification, was used to re-
solve the case (Gove, 1983). In the end all
of these subjects were given a single classi-
fication, A, B, or C.
Seductive care (24 months).-During a
clean-up task at 24 months, approximately
10%
of the mothers in the total sample were
seen to exhibit some form of seductive be-
havior with the child. Behaviors included
sensual whispering or touching, touching of
genitals, and attempted bribes with af-
fection. These are described more fully in
Sroufe and Ward (1980). A total of seven
children across the three camps had mothers
who behaved seductively at 24 months.
Generational boundary dissolution (42
months).-In a series of teaching tasks, care-
givers were rated on two scales: (1) Nonre-
sponsive Physical Intimacy, which over-
lapped somewhat with the seductive
category at 24 months (including any efforts
at physical intimacy that was for the
mother's
gratification
rather than for support
of the child); (2) Generational
Boundary
Dis-
solution, which focused on blurring of
boundaries between parent and child (e.g.,
the child is pressed toward
the parental
role;
parent engages in provocative teasing and
taunting or otherwise behaves as though a
peer with the child). Both scales, and the
rationale for combining them into a more
molar scale ("Boundary Dissolution Com-
posite"), are described in a prior
publication
(Sroufe et al., 1985). Both the 24-month and
the 42-month scales were broadly consid-
ered as assessments of parent-child genera-
tional boundaries.
Results
RELIABILITY AND CONVERGENT VALIDITY
Cross-Gender
Behavior Codes
Percent agreement of the two coders for
the cross-gender behavior codes ranged
from .41 to .71. Despite extensive training,
certain categories proved difficult to differ-
Sroufe et al. 461
entially code; for example, hovering (.41)
versus joining (.54). Reliabilities for fre-
quently occurring categories were more ade-
quate; for example, border work (.71) and
relaxed interaction (.68). When boundary
vi-
olating and nonviolating categories were
collapsed, reliability improved to .83. Agree-
ment on violating behaviors was .80, on non-
violating behaviors .84. This collapsed
score, the proportion of violating behaviors
observed, was used in the analyses below.
(All disagreements were conferenced to con-
sensus.)
Ratings
Across the three camps, reliability for
the Gender Boundary Violation Scale was
.79 and for the Boundary
Maintenance Scale
.67 (Pearson r's). The composite of the two
coders served as data,
and the estimated reli-
ability of these composites (Spearman-
Brown formula) is .88 and .80, respectively,
for the two scales.
Correlations of Ratings and Behaviorally
Based Indices
The Boundary Maintenance Scale
scores and Boundary
Violation Scale scores,
which were as predicted modestly nega-
tively correlated (r = -.20, p < .10), were
combined by subtracting
the latter from the
former. The correlation of this index with
the violation behavioral code index in the
first
camp was .68 (p < .01). The correlations
of the behavioral index with each of the two
scale scores taken individually were .69
(p < .01) for boundary violation and .36
(p < .10) for
maintenance; that is, the bound-
ary violation behavioral code index corre-
lated better with the Boundary Violation
Scale. Overall, this convergence attests to
the meaningfulness and behavioral base of
the ratings used as primary
data below.
Frequency of Cross-Gender
Association
In accord with the previous literature,
preadolescents in our camp rarely
associated
exclusively with members of the other gen-
der. Based on extensive child sampling, with
an average of 317 observations per child,
there was a mean of 17.2 instances of cross-
gender association or 6%
of the observations.
However, there was ample variation
to allow
this index to be used in other analyses.
Contemporary Correlates of Boundary
Violation and Maintenance
The major index of social competence
available was the composited social skill rat-
ing of the four counselors. Both the Gender
Boundary
Violation Scale and the Boundary
Maintenance Scale correlated significantly
and in the expected direction with this mea-
sure (see Table 2). The composite of the
scales (Maintenance Scale score minus Vio-
lation Scale score) correlated .53 with the
counselor judgments of competence (p <
.001). Those who frequently violated gender
boundaries were independently judged by
counselors to be lower on social competence
(r = -.33); those who actively maintained
boundaries were judged to be higher (r =
.53). Likewise, for Camp 1, those higher on
the behavioral index of boundary violation
were judged as less competent (r = -.36),
although this does not reach the .05 level of
confidence with 16 subjects. Even the crude
index of sheer frequency of association with
the other gender was negatively correlated
with counselor judgments of social compe-
tence (-.32 and .30, respectively, for the
two competence indices; p < .02).
The counselor data are supported by
corroborating observational data, namely,
the friendship score. Correlations with the
violation and maintenance scales were -.18
and .55, respectively, and for the composite
of these ratings .44 (p < .001). The violation
behavior index from Camp 1 (N = 16) also
correlated significantly with the camp
friendship measure (r = -.43, p < .05), and
the more specific "joining" score ap-
proached significance (r = -.41, p < .06).'
Finally, children themselves indicated
that those who violate gender boundaries
and show little evidence of maintaining
them (our composited index) are less popu-
lar, based on our camp interview (r = .37, p
< .005). Again, this index of competence also
correlated negatively with sheer frequency
of contact (r = -.30, p < .05). For Camp 1,
the correlation
with the violation behavioral
index was -.60 (p < .01) and with "joining"
was -.38 (p < .10).
Sex Differences
While it was found that boys received
both higher boundary violation and bound-
ary maintenance scale scores than girls (pre-
1
While the reliability of the "joining" index was only .54, the conferenced scores used can
be assumed to be more reliable. Still, the low reliability constrains the relationship with other
variables.
.0 .
00 00-4(7 -*?
0 0
0
-* *
CdCCk
z z 0
u -0
Q <
m
z4 9
*.*** **
S 0 (1) 0*
0 4
1-4 "0 ? . II
ZQ
EV4
* * 0
* **
00
C
~O
t-
V V
CO
-
V? CO
V
01
z Cd *
C0CC
wR
?
, In lIn
o 0 r--4
Uz
Z0
CC
Z00 C
OE O
oo
0 0
r.4 -2
00
0
q -4
TdCd
boIC~C~~
5
p
4
0 0 .,*.*4
z
01-4. . 0
cr0z0Q
0
o
cd
) ov
L~ cl
r4
en
n
ek
i
Sroufe et al. 463
sented below in the "antecedents" section),
there was no difference in the patterns of
correlations for boys and girls. For example,
the correlation between the counselor social
skills rating
and the composite Boundary
Vi-
olation-Boundary Maintenance Scale was
exactly .53 for boys and for girls. Moreover,
regression analyses in which gender was en-
tered first still consistently showed signifi-
cant effects for the competence variables
when predicting boundary violation and
boundary
maintenance. Moreover,
the inter-
action effect between gender and the social
competence indices never approached sig-
nificance in these regressions.
Controlling for Social Activity
We doubted that our competence mea-
sures were merely reflections of social activ-
ity, even though proportion of time spent
with other children did correlate signifi-
cantly and positively with the social skill rat-
ing (.63, p < .001) and modestly with peer-
based popularity (.21, p < .10). Nonetheless,
it was possible to control for amount of child
contact using partial
correlations.
Proportion
of time spent with other children was used
in these analyses for the social skill rating
and the peer popularity variable. A partial
correlation was not necessary for the friend-
ship score, since this is already a proportion
score and thus accounts for individual differ-
ences in activity. This approach
is very con-
servative, because meaningful variance is
being partialled out. Social competence
does involve being with other children.
Nonetheless, for the social skills variable,
the resulting partial correlation was signifi-
cant for the Boundary
Violation Scale (-.35,
p < .01), the Boundary Maintenance Scale
(.26, p < .05) and the combined scales (.40,
p < .01). For popularity, the violation and
combined scales were significant (-.33 and
.32, respectively, p < .02). The result was
not significant with the maintenance scale.
It also should be mentioned that proportion
of time spent with the opposite gender cor-
related negatively (-.26, p < .05) with
boundary maintenance, while proportion of
time in isolation correlated positively with
boundary violation (.22, p < .10). Neither of
these results would follow from a social ac-
tivity interpretation.
ANTECEDENTS OF BOUNDARY VIOLATION
AND MAINTENANCE
Attachment History
Since attachment history has proven to
be a robust predictor of later social compe-
tence (e.g., Elicker et al., 1992) and because
it has been viewed as integral in develop-
ment of the self (e.g., Sroufe, 1990), it was
used as another external correlate for assess-
ing the validity of the gender boundary con-
struct. All camp children had previously
been grouped into secure and anxious at-
tachment categories. For this analysis A and
C anxious patterns were combined in car-
rying out posttests.
Analyses of variance were carried
out on
the violation and maintenance scale scores
and the composite index (maintenance - vi-
olation), in each case yielding significant
main effects for attachment history, F =
5.287, p < .002; F = 12.507, p < .001; F =
14.389, p < .001; respectively. Posttesting
revealed no significant differences between
the two anxious attachment groups, which
were, taken together, significantly different
from the secure group for each scale (p <
.005 for the Boundary Violation Scale, p <
.001 for the Boundary Maintenance and
Composite Scales). Those with secure histo-
ries more likely maintained, and less likely
violated, gender boundaries. In addition, for
each of the two scales there was a main ef-
fect for sex: boys showed significantly more
boundary violation and boundary mainte-
nance.
Generational Boundary Violation
Each measure of generational boundary
violation available from our longitudinal as-
sessments in early childhood was related to
later gender boundary violation and mainte-
nance, always in the expected direction.
For the categorical "seductive behav-
ior" index at 24 months, a chi-square analy-
sis was carried out. In the camps seven chil-
dren were present whose mothers had
behaved seductively in the clean-up task at
24 months;- five of these children were
among the 10 children (of 47) who had
scores of 5 or higher on the boundary viola-
tion scale. The resulting chi square was
12.35, p < .001.
The two scales from
the 42-month
teach-
ing task (Generational
Boundary
Dissolution
and Nonresponsive Physical Intimacy), and
the combination of these, all are continuous
variables, so correlational
analyses were car-
ried out. The Generational Boundary Disso-
lution Scale at 42 months correlated signifi-
cantly, and in the expected direction, with
both gender boundary scales at age 10-11
years and their composite (range .29-.44).
The Nonresponsive Physical Intimacy Scale
correlated significantly with the later Gen-
464 Child Development
der Boundary
Violation Scale (r = .38). This
patterning may be seen in Table 2.
Discussion
Substantial validation for the gender
boundary construct was obtained in this
study, in accord with the previous literature.
Support for the idea of gender boundaries
as a normative developmental issue derives
from the finding of predominantly within-
gender associations during free choice time.
It seems likely that all of the children ob-
served have some understanding of this
norm and the rules governing it, including
contexts in which cross-gender association
would be appropriate (Maccoby, 1990). No
child associated predominantly with the
other gender. From a normative point of
view it was also noteworthy that boys
showed both more boundary maintenance
and boundary
violation than girls. Past liter-
ature
has suggested greater
sanctions against
boys for boundary violation, as well as
greater
stigma for doing so (Thorne & Luria,
1986). We do not view our data as inconsis-
tent with this. Rather, we take the set of
findings as implying that gender boundary
issues are of great salience for boys and that
boys may do more work at the boundaries.
It is important
to note that past studies have
not documented less violation of boundaries
by boys, only more sanctions.
Beyond these normative issues, we
view these data as strongly validating gen-
der boundary violation/maintenance as an
individual differences construct as well.
Both gender boundary violation and gender
boundary maintenance were reliably rated.
It was difficult to achieve reliability for some
of the specific behavioral codes, but these
too were reliable when collapsed across cat-
egories. Moreover,
the ratings and collapsed
behavioral codes were in good agreement,
attesting to the behavioral basis of the rat-
ings. More important, external validity was
demonstrated for all assessments of gender
boundary violation or maintenance. Both a
broad range of contemporary competence
correlates
and meaningful antecedents were
demonstrated.
With the most robust external
correlates-the composited counselor rating
of social skill and the composited genera-
tional boundary dissolution index-impres-
sive correlations were obtained with the
combined Gender Boundary Maintenance
Scale (.53 and -.41, respectively).
Our strong support for boundary viola-
tion/maintenance as an individual differ-
ences construct may at first seem inconsis-
tent with an earlier claim by Maccoby and
Jacklin (1987) that differences in cross-
gender interaction are not stable. Without
such stability one would never have ex-
pected external correlates of such a measure.
However, their subjects were 41/2-61/2
years
old, an age when boundary maintenance
may not yet be a salient issue. Moreover,
they did not directly assess boundary main-
tenance and violation but gender segrega-
tion per se.
The correlations with social compe-
tence, popularity, and friendship suggest
that those who cross gender boundaries are
lacking in social competence, rather than
being precociously competent, and support
the hypothesis that maintenance of gender
boundaries is an important sign of positive
adaptation
in late middle childhood. Our in-
terpretation of these findings follows from
the notion that acquiring peer group norms
is an important
task in middle childhood and
preadolescence. One such norm may con-
cern the maintenance of gender boundaries.
Children who violate such boundaries are
generally unpopular
with peers. In addition,
such violation is one marker of social incom-
petence. Even more strongly, maintenance
of gender boundaries is a marker of social
competence; as such, it correlates with other
indices of social competence.
The link between attachment history
and the gender boundary assessments can
be explained in several ways. It may again
be interpreted
in terms of all measures being
from the competence domain. Since attach-
ment security predicts broad indices of com-
petence, so too it predicts these particular
indicators in preadolescence. Along these
lines, one reason children may cross gender
boundaries is failure within the same-
gender peer group. Being unsuccessful with
same-gender peers, they may spend more
time with members of the other gender. This
would not seem to be the total picture, how-
ever, since behaviors in the boundary viola-
tion category imply more than de facto inter-
action with members of the opposite gender.
Another interpretation of the attach-
ment-gender boundary linkage is more
deeply theoretical and centers on the notion
of self-development. Identity and, as part of
that, gender identity may have important
roots in the dyadic child-caregiver relation-
ship. Boundaries are an important
aspect of
self-development, as the young child learns
to be both autonomous and connected. A se-
Sroufe et al. 465
cure attachment relationship, in which the
child achieves a smooth balance between
exploring away from the caregiver and seek-
ing contact when threatened or distressed,
may mark the early origins of a clear sense
of self or identity.
Such theoretical concerns led us to ex-
amine aspects of our data set targeted at the
boundaries between parent (mothers in our
risk sample) and child. These measures
showed clear associations with the gender
boundary outcome measures. The fact that
the parent-child boundary dissolution mea-
sures also correlated with other aspects of
later social competence, as well as the corre-
lational nature of this study, suggests that
conclusions must be tempered at this point.
However, the patterning of the findings,
with seductiveness at 24 months and non-
responsive physical intimacy at 42 months
correlating with later gender boundary vio-
lating and the broader generational bound-
ary dissolution scale correlating (negatively)
with boundary maintenance, are suggestive.
We believe that
two things may be involved:
first, through clear parent-child boundaries
clear self or ego boundaries may be evolved,
and, second, such parent-child boundary
dissolution may in many cases be gender
based. Unfortunately, with our predomi-
nantly single mother sample we are not able
adequately to investigate differential bound-
ary dissolution across parents. At the least,
these findings again attest to the meaning-
fulness of gender boundary maintenance as
a developmental construct.
An important task for future research
will be to relate individual differences on
gender boundary violation and maintenance
in preadolescence to peer relationship vari-
ables in adolescence. If we are correct that
adhering to the gender boundary norm pro-
motes acquisition of the experiences and
skills that will promote later successful func-
tioning in the adolescent peer group, numer-
ous predictions follow. In an earlier study
Kagan and Moss (1962) reported that boys
who were centrally involved in the male
peer group in preadolescence were less
avoidant of sexual activity in early adoles-
cence. More recently, Feldman et al. (in
press) reported different pathways to adoles-
cent heterosexual functioning for boys who
had been popular or rejected in middle
childhood. Both of these studies defined
preadolescent competence in broad terms.
We would propose more specifically that in-
dividuals following the culturally normative
pathway of cross-gender engagement with
boundaries maintained in late middle child-
hood should as adolescents be able to deal
well with the social complexities of that pe-
riod;
namely, coordinate
same-gender
friend-
ships and group functioning with cross-
gender relationships and functioning in the
mixed-gender crowd. In particular,
mainte-
nance of cross-gender boundaries in preado-
lescence may predict successful cross-
gender relationships in adolescence.
Appendix
Rating Scale for Boundary-violating
Behaviors
1. This child never violates gender bound-
aries (0%). When interactions occur, the child is
always with gender mates, and cross-gender be-
haviors are always in the service of boundary
maintenance
(show some disavowal of interest). It
may be that this child is never seen interacting
with the other gender.
2. Child seldom violates boundaries
and typi-
cally does so by either hovering or by showing
ambiguous
behavior.
3. Occasionally violates boundaries-there
may be a mixture of hovering and joining.
4. Child has a moderate amount of low-
intensity violations (hovering, joining, partici-
pating).
5. Boundary violations may be rare but are
strong when they occur, involving sexualized
comments or inappropriate physical contact, or
these may occur in addition to what happens in 4.
Or,
the child spends a great deal of time hovering,
joining, or participating.
6. The child occasionally makes intimate
physical contact of a sensual or aggressive nature,
or the child makes frequent sexualized comments/
gestures toward
the opposite gender, but they are
not so intense or found in the diversity of settings
as 7. Or, the child spends an inordinate
amount
of
time hovering,
joining, or participating
and seems
to associate more readily with the other gender
than their own.
7. Child regularly
violates boundaries under
a variety of circumstances. The majority
of these
occur in the absence of gender mates. The child
often makes intimate physical contact of a sensual
or aggressive nature.
Rating Scale for Boundary
Maintenance Behaviors
1. Child never maintains boundaries during
cross-gender encounters. They seem to associate
more readily with the opposite gender than their
own. This is true whether the child is alone with
the other gender or part of a mixed group. Or,
this child is never seen interacting
with the other
gender.
466 Child Development
2. Child uses only passive maintenance, on
an occasional basis.
3. Passive maintenance only, in a moderate
amount.
4. Child engages in passive maintenance
nearly all the time. Or, does a small amount of
active maintenance.
5. Moderately active at maintaining bound-
aries. May do some passive maintenance.
6. Boundary is maintained nearly all of the
time, through
a mixture
of active (mainly)
and pas-
sive maintenance.
7. Boundary
is always maintained. Child uses
most or all opportunities for active maintenance.
Child is quick to leave situations where alone
with the other gender.
Definitions
Active maintenance: border work and escape
strategies, such as calling for support
from
gender
mates or leaving the scene when none are
available.
Passive maintenance:
relaxed interaction,
on-
looking, contexts where boundary is provided by
"cover,"
for example, being in a group or with an
adult.
References
Ainsworth, M., Blehar, M., Waters, E., & Wall, S.
(1978). Patterns of attachment. Hillsdale, NJ:
Erlbaum.
Egeland, B., & Brunnquell, D. (1979). An at-risk
approach to the study of child abuse: Some
preliminary
findings.
Journal of the American
Academy of Child Psychiatry, 18, 219-225.
Elicker, J., Englund, M., & Sroufe, L. A. (1992).
Predicting peer competence and peer rela-
tionships in childhood from early parent-
child relationships. In R. Parke & G. Ladd
(Eds.), Family-peer relationships: Modes of
linkage. Hillsdale, NJ: Erlbaum.
Feldman, S., Rosenthal, D., Brown, N., & Can-
ning, R. (in press). Predicting sexual experi-
ence in adolescent boys from peer rejection
and acceptance during childhood. Child De-
velopment.
Gove, F. (1983).
Patterns and organizations of be-
havior and affective expression during the
second year of life. Unpublished doctoral
dis-
sertation, University of Minnesota.
Hartup,
W. (in press). Peer relations in early and
middle childhood. In V. VanHasselt & M.
Hersen (Eds.), Handbook of social develop-
ment: A lifespan perspective. New York:
Plenum.
Kagan,
J., & Moss, H. (1962). Birth to maturity.
New York:
Wiley.
Maccoby, E. (1990). Gender and relationships.
American Psychologist, 45, 513-520.
Maccoby, E., & Jacklin, C. (1987). Gender segre-
gation in childhood. In H. W. Reese (Ed.),
Advances in child development and behavior
(Vol. 20, pp. 239-288). New York:
Academic
Press.
Schofield, J. (1981). Complementary
and conflict-
ing identities: Images and interaction in an
interracial school. In S. Asher & J. Gottman
(Eds.), The development of children's
friend-
ships (pp. 53-90). Cambridge: Cambridge
University Press.
Sroufe,
L. A. (1990).
An organizational
perspective
on the self. In D. Cicchetti & M. Beeghly
(Eds.), The self in transition: Infancy to
childhood (pp. 281-307). Chicago: University
of Chicago Press.
Sroufe, L. A. (1991). Considering normal and ab-
normal together: The essence of develop-
mental psychopathology. Development and
Psychopathology, 2, 335-347.
Sroufe, L. A., Carlson, E., & Shulman, S. (in
press). The development of individuals in re-
lationships: From infancy through adoles-
cence. In D. C. Funder, R. D. Parke,
C. Tom-
linson-Keasey, & K. Widaman (Eds.),
Studying lives through time: Approaches to
personality and development. Washington,
DC: American
Psychological Association.
Sroufe, L. A., & Fleeson, J. (1988). The coherence
of family relationships. In R. A. Hinde &
J. Stevenson-Hinde (Eds.), Relationships
within families: Mutual influences (pp. 27-
47). Oxford: Oxford University.Press.
Sroufe, L. A., Jacobvitz, J., Mangelsdorf, S.,
DeAngelo, E., & Ward,
M. J. (1985). Genera-
tional boundary
dissolution between mothers
and their preschool children: A relationships
systems approach. Child Development, 56,
317-325.
Sroufe, L. A., & Ward,
M. J. (1980). Seductive be-
havior of mothers of toddlers: Occurrence,
correlates,
and family origins. Child Develop-
ment, 51, 1222-1229.
Sullivan, H. S. (1953). The interpersonal theory of
psychiatry. New York: Norton.
Thorne, B. (1986). Girls and boys together ... but
mostly apart: Gender arrangements in ele-
mentary schools. In W. Hartup & Z. Rubin
(Eds.), Relationships and development (pp.
167-184). Hillsdale, NJ: Erlbaum.
Thorne, B., & Luria,
Z. (1986). Sexuality and gen-
der in children's daily worlds. Social Prob-
lems, 33, 176-190.
Waters, E., & Sroufe, L. A. (1983). Social compe-
tence as a developmental construct.
Develop-
mental Review, 3, 79-97.
Whiting, B. B., & Edwards,
C. P. (1988). Children
of different worlds. Cambridge,
MA:
Harvard
University Press.