IDENTIFICATION OF SECONDARY ATTACHMENT FIGURES IN

IDENTIFICATION OF SECONDARY ATTACHMENT FIGURES IN MIDDLE
CHILDHOOD
A thesis submitted
to Kent State University in partial
fulfillment of the requirements for the
degree of Master of Arts
by
Ashley C. Seibert
May, 2007
Thesis written by
Ashley C. Seibert
B.S., University of Pittsburgh, 2004
M.A., Kent State University, 2007
Approved by
Kathryn A. Kerns, PhD
Advisor
Janis Crowther, PhD
Chair, Department of Psychology
Jerry Feezel, PhD
Dean, College of Arts and Science
ii
TABLE OF CONTENTS
LIST OF TABLES…………………………………………………….......…...……..…..iv
INTRODUCTION…………………………………………………………...…...…….…1
STUDY 1 - INTRODUCTION……………..…...…………………………...…...…….…8
METHOD - STUDY 1……….……………..…...………………………...…...…….…11
Participants……………………………………………………………...…………11
Procedures..………………………………………………………………………..11
Measures...…………………………………………………………………………12
RESULTS - STUDY 1……….…………………………………..……..…..…….......…16
DISCUSSION - STUDY 1………………………………………………………............29
STUDY 2 - INTRODUCTION…….…………………………………………………….33
METHOD - STUDY 2………………………………………………………………...…36
Participants……………………………………………………………...…………36
Procedures..………………………………………………………………………..36
Measures….......……………………………………………………………………37
RESULTS – STUDY 2..…………………………………………………………………40
DISCUSSION - STUDY 2………………………………………………………………48
GENERAL DISCUSSION………………………………………………………………51
REFERENCES……..……………………………………………………………………56
APPENDIX A……………………………………………………………………………60
iii
LIST OF TABLES
Table
Page
1. Who Children Seek First for General Attachment and Companionship,
Context-Specific Attachment and Companionship, and
School Attachment Situations…….………………………….………………………..15
2. Mean Scores for Who Children Nominated
as First Choice or Ever…….………………………….…………………………...…..19
3. All People Children Seek for General Attachment and Companionship,
Context-Specific Attachment and Companionship, and
School Attachment Situations…….………………………….………………………..22
4. Age Effects for First Choice and Ever……….……….…………………………...…..25
5. Sibling Effects for Ever……………….…………………………..........................…..27
6. Gender Effects for Ever…..………………………….….………………………...…..28
7. Who Children Say They Would Go To and Who Children Say
They Would Go To Plus Provided an Example…….…………...………………...…..41
8. Sibling Effect for Who Children Say They Would
Go To when Scared……………………………………………………………......…..43
9. Gender Effects for Who Children Say They Would
Go To when Sad or Scared……………….……………………….…………..…...…..45
10. Mean Security Scores for Who Children Would
Go To if They Were Sad or Scared……………….……………………………....…..46
11. Children’s Security with Mom Correlated with
Comfort Ratings……………….…………………………………………….…...…..47
iv
INTRODUCTION
The attachment relationship has been described as an enduring, emotional bond a
child forms with a particular attachment figure who provides the child with security and
comfort (Ainsworth, 1989). The child uses the attachment figure as a secure base from
which to explore the environment and as a safe haven in times of distress. The formation
of attachment to caregivers is a normative event. That is, all children form attachments to
their caregivers even if they do not receive adequate care (Bowlby, 1982). Attachment
relationships are thought to be long-enduring, and the attachment figure is not
interchangeable with another person (Ainsworth, 1989). There is a desire to maintain
closeness to the attachment figure and reestablish proximity if the bond is threatened. In
an attachment relationship, a child may experience distress if separated from his or her
attachment figure and will experience grief if there is permanent loss of the attachment
figure.
Until fairly recently attachment theory has been mainly applied to parent-child
relationships, and most often to the mother-child relationship. More recently, researchers
have begun to acknowledge and focus on multiple attachments. Even Bowlby (1982)
argued that children form multiple attachments, which are organized in a hierarchy, with
some attachment relationships more important than others. While children’s primary
attachments are to their mother and father, subsidiary attachment figures, as Bowlby
1
2
called them, are people such as siblings and grandparents. Webb (1984) noted that
virtually all researchers from Bowlby on confirm the existence of multiple attachments in
most children by twelve to eighteen months of age. She also mentioned that the “normal
thrust” of a child’s growth and development is toward expanding relationships.
Therefore, forming relationships with people besides the primary caregiver, such as nonparental caregivers and siblings, is a normal part of a child’s life. There has been
research on the role of siblings and teachers as secondary attachment figures, but only in
early childhood. These studies have shown that children in full-time daycare use teachers
as attachment figures (Howes & Oldman, 2001; Howes, 1999; Howes & Hamilton,
1992). In addition, toddlers will use their older siblings as attachment figures in stressful
situations where their parents are not available (Stewart & Marvin, 1984).
During middle childhood, parents are still the principal attachment figures for
most children (Kerns, Tomich, & Kim 2006). However, children may begin to direct
attachment behaviors to partners in other close relationships. In addition, children may
show a tendency to rely on different attachment figures in different contexts rather than
one attachment figure for all situations (Mayseless, 2005). Children may also need to
rely on other people to fulfill their attachment needs when access to their primary
attachment figure is blocked, such as when children are attending school and are
physically separated from their parents. Therefore, secondary attachment relationships
may become more prevalent during middle childhood. We define secondary attachment
relationships as relationships that are similar to primary attachment relationships in that
children use secondary attachment figures as a secure base from which to explore the
3
environment as well as a haven of safety in times of distress. Unlike primary attachment
relationships, secondary attachment relationships do not have to be long-enduring, and it
is possible that secondary attachment figures may be interchangeable with one another.
Thus, by our definition, secondary attachment relationships do not have to encompass all
of the criteria that need to be present in primary attachment relationships. Instead,
secondary attachments might be more time limited and situation dependent.
Several researchers have focused on hypothesized changes in the attachment
relationship between early childhood and middle childhood (Bowlby, 1982; Mayseless,
2005). Due to these changes, secondary or subsidiary attachments may become more
common in middle childhood. First, the intensity and frequency of attachment behavior
may decrease. This change may allow children to experience more freedom and explore
other relationships (Mayseless, 2005). Second, there may be a shift in responsibility
between parent and child for monitoring and maintaining availability of the caregiver
(Mayseless, 2005; Kerns, Aspelmeier, Gentzler, & Grabill, 2001; Marvin & Britner,
1999). The child may need to call upon the attachment figure less often and may more
often cope with situations on his/her own (Marvin & Britner, 1999). Third, the
attachment system may become more differentiated and diversified. That is, children
may form bonds with other people besides their primary caregivers, such as teachers,
grandparents, siblings, other adults, etc. (Mayseless, 2005). Thus, attachment-related
behaviors may begin to occur in non-parental relationships of children in middle
childhood.
4
There may also be evolutionary reasons to believe that secondary attachments
may occur in middle childhood (Mayseless, 2005). It has been suggested that humans
have an evolutionary-based developmental tendency to partially withdraw their
investments with their primary caregivers and to invest instead in getting along with their
peer group and securing their cooperation and protection as children get older
(Mayseless, 2005). This process is hypothesized to start during middle childhood. This
partial withdrawal of investment and its refocusing may be reflected in the tendency to
turn to others besides primary caregivers in times of distress, especially peers.
Consequently, there may be a shift in middle childhood in preparation for adolescence
and the development of peer attachments.
In summary, we hypothesize that children form secondary or subsidiary
attachments in middle childhood, and that the incidence of secondary attachments may
increase at the end of the middle childhood period. Children may still rely on their
primary attachment figures first, but go to a secondary figure if their primary figure is not
available. As Cassidy (1999) suggests, the bond with primary figures does not become
weaker as the child matures. Instead, the relationship with the primary figure may not
infiltrate as many aspects of the child’s life as the child begins to spend more time away
from parents and to develop new relationships. Thus, in middle childhood secondary
attachments supplement rather than replace attachments to primary caregivers (usually
parents).
There has been research on who are children’s attachment figures for children in
middle childhood, with attention to whether peers might function as attachment figures.
5
Hazan and Zeifman (1994) have proposed that there are four components of attachment:
proximity maintenance, safe haven, separation distress, and secure base. In addition, they
propose that the first two attachment components are transferred from parents to peer
during middle childhood. To test their hypothesis, they conducted an interview study of
6- to 17-year-olds to examine whether and when children transfer the four attachment
components from parents to peers. They found that most children, at all age groups,
preferred to spend their time with peers than with parents (i.e. the proximity maintenance
component). Hazan and Zeifman (1994) found an age trend for the safe haven
component with a shift from parents to peers occurring between the 8-10 year-old group
and the 11-14 year-old group. Children in all age groups reported directing separation
distress and secure base behavior to parents over peers. Hazan and Zeifman (1994)
interpreted these findings as showing that children in middle childhood have already
transferred the proximity maintenance component from parents to peers and are in the
process of transferring the safe haven component.
Nickerson and Nagle (2005) also tested Hazan’s hypothesis of transferring
attachment components from parents to peers by interviewing fourth, sixth, and eighth
graders. They found that grade was a significant predictor of whether children selected
parents or peers to fulfill the proximity seeking component (eighth graders were more
likely than fourth graders to turn to peers for proximity seeking). Grade was also a
significant predictor for the safe haven component, again with eighth graders more likely
than fourth graders to pick peers over parents to fulfill the safe haven function. Grade
was not a significant predictor, however, for the secure base component. The majority of
6
children in all three grades selected parents to fulfill the secure base component.
Nickerson and Nagle’s (2005) findings were similar to Hazan and Zeifman’s (1994) in
showing the progression from parents to peers for the proximity seeking and safe haven
components of attachment, although their study suggested that the transition occurs in
early adolescence rather than middle childhood.
Kobak, Rosenthal, and Serwik (2005) adapted the Hazan and Zeifman (1994)
interview by asking children to name the four most important people in their life. The
interviewer then asks the child which of the people he or she would like to go to or be
around in different attachment situations. The child continues to nominate people until
all four people are ranked for each situation. In their sample of 9-13 year-olds, Kobak et
al. (2005) found that parents were ranked highest for all four attachment components.
These findings were consistent with Nickerson and Nagle (2005) in suggesting that
parents are still the primary attachment figures for children in middle childhood.
Kerns et al. (2006) questioned Hazan and Zeifman’s (1994) interpretations,
particularly that of proximity maintenance. Kerns et al. (2006) suggested that some
contacts might involve proximity, but do not necessarily involve attachment needs. For
example, a child might approach a peer to play, which could be interpreted as proximity
maintenance, although in this case the proximity is not in the service of the child’s
attachment needs. Because of the way the questions were phrased in the Hazan and
Zeifman (1994) interview, it was not clear whether children’s answers to the proximity
maintenance questions reflected seeking others for companionship needs or for
attachment needs. In their study, they examined who children seek out for attachment
7
needs and companionship needs in middle childhood by asking children who they would
seek out in the two types of situations (e.g. when they were scared versus when they
wanted to play). Their attachment questions thus captured what Hazan and Zeifman
(1994) termed the safe haven component of attachment. In their study of third and sixth
graders, they found that children wanted to be around parents when the attachment
system would be activated (i.e., the safe haven questions) and want to be around peers for
companionship.
While it has now been established that parents function as primary attachment
figures in middle childhood, very little is known about the identity and use of secondary
attachment figures. The main goal of the two studies in this paper is to identify
secondary attachment figures in middle childhood. In Study 1, we used an open ended
interview where children were allowed to generate their own list of important people in
their lives. We then used the list the child generated to ask whom they would go to in
both attachment and non-attachment (i.e. companionship) situations. In Study 2, we used
a more structured interview where we explicitly asked children whether they go to
particular individuals in attachment situations. We included children 7-12 years of age to
explore whether there are age differences in reliance on secondary attachment figures.
STUDY 1
Study 1 extends the literature in several ways. First, we will expand upon the
Kerns et al. (2006) study by differentiating between attachment needs and companionship
needs in a variety of contexts. We included questions that would evoke the attachment
system, companionship questions, and questions that would evoke the attachment system
while the child was at school. Second, in order to identify secondary attachment figures,
we will expand upon the earlier studies by examining not only who children nominate as
their first choice, but also whether or not they ever nominate certain individuals (e.g.,
siblings, peers) in response to attachment questions. If certain individuals do serve as
secondary attachment figures for children, children should nominate them at some point
for the attachment questions, but if they are never nominated for attachment questions we
have evidence that these individuals are not being used as secondary attachment figures.
In addition, our Study 1 interview allowed children to nominate as many
important people they can think of in their lives (instead of limiting them to four people),
and to select as many or as few people as they would like for each question. Giving
children more options allowed us to investigate potential secondary attachment figures
that may be overlooked when children are restricted in the number of people they are
allowed to nominate. Also, by not specifying a minimum number of nominations, the
procedure does not force children to nominate or rank a particular figure.
8
9
Our first hypothesis was that children would nominate parents first for the general
attachment questions and peers first for general companionship questions, replicating
earlier research (Kerns et al., 2006). We also extended earlier research by including
context-specific attachment and companionship questions with more concrete examples.
For these questions we expected to see the same pattern as the general questions with
children choosing parents for attachment situations and peers for the companionship
situations. However, if children are beginning to show differentiation in their use of
attachment figures, it is possible that a preference for parents may be less pronounced in
specific attachment situations. We also included questions about events that would evoke
the attachment system while the child was at school because we were interested in what
children would do in the physical absence of the parent, that is, whether this may be a
context in which children may use secondary attachment figures. One possibility is that
children might still pick their parents as the first person they would go to, meaning they
would not seek out alternative attachment figures at school, but would wait until they
returned home. Another possibility is that they might use an alternative adult such as a
teacher or a peer to fulfill attachment needs when a parent is not available.
Our second hypothesis was that children would show more diversification of
attachment networks with age. We therefore expected that older children would be more
likely than younger children to choose someone other than a parent as a secondary
attachment figure. As another test of the hypothesis of an increased diversification of
attachment networks with age, we expected that older children would nominate more
people or more unique individuals per question than would younger children.
10
We also explored whether the above hypotheses might vary with certain child
characteristics. Earlier literature shows that children do use siblings as attachment
figures. An attachment figure is defined as someone who is older and wiser (Bowlby,
1982). Therefore, our third hypothesis was that children with at least one older sibling
would be more likely than children with only younger siblings to nominate a sibling as an
attachment figure. Finally, we examined gender differences in attachment figure
preference, although we did not have any a priori hypotheses.
METHOD
Participants
Participants were recruited through elementary schools and a summer camp
program. Recruitment letters were sent home to the parents of second through sixth
grade children explaining the study and inviting the child to participate. The final sample
(N = 114) consisted of 55 boys and 59 girls. Thirteen of the participants were from the
summer camp program, 27 were from a Catholic school, and 74 were from a public
school. The average age of the participating children was 9 years, with an age range of 7
to 12 years. Ninety percent of the participants were Caucasian, 1 percent were AfricanAmerican, 3 percent were Asian, and 6 percent reported other ethnic origins. Family
status categories included two-parent intact families (90 percent), two-parent blended
families (1 percent), and single-parent families (9 percent). Sixty percent of the
participants had at least one older sibling, 31 percent had only younger siblings, and 9
percent were only children.
Procedure
All data were collected at the participating elementary schools or the summer
camp program. A research assistant worked with a teacher from each classroom and with
the director of the summer camp program to set up an interview schedule. Participating
11
12
children were taken from their classrooms or from their summer camp activities to a
separate interview room for approximately 15 minutes to complete the interview.
Measures
Social Network Interview. Antonucci’s (1986) network mapping procedure was
used to identify the important people in a child’s life. Participants were shown a drawing
depicting three concentric circles. In the middle of the smallest circle was the word
“you.” Participants were told to think of “people who are important in your life right
now” and then place these people in the three circles based on how close they felt to that
person. Instructions for the inner circle were: “Beginning with the people you feel
closest to, is there one person or persons that you feel so close to that it’s hard to imagine
life without them?” The participants nominated these people for the innermost circle and
the interviewer wrote these names down on separate note cards. The same procedure was
followed for the next circle by asking participants to name “people to whom you may not
feel quite that close but who are still very important to you.” Once again, the same
procedure was followed for the outer circle. Participants were asked to name “people
whom you haven’t already mentioned but who are close enough and important enough in
your life that they should be placed in your personal network.”
Demographic information, such as age, gender, and relationship to child, was
collected about each person the child nominated. After the child nominated people the
interviewer put their names on note cards and laid them out in front of the child to remind
him or her whom he or she had nominated. The interviewer also put a “nobody” card in
13
front of each child so that the child would feel free to select “nobody” if he or she
wanted.
Attachment Figure Interview. To assess who children would seek out in different
situations they completed an author-constructed attachment figure interview. The
interviewer said to each child, “now that you have told me about all of the important
people in your life, I am going to ask you a few questions.” For each question, the child
was asked whom he or she would go to first. After this choice was made, the interviewer
kept asking who the child wanted to go to until everyone on the child’s list had been
nominated or until the child said that he or she would no longer go to anyone for that
particular situation. This allowed children to make an unrestricted number of
nominations. Children were also allowed to nominate “nobody” if they wanted to.
The attachment situations were situations that would evoke the need for an
attachment figure. For example, the attachment situations assessed the safe haven
component for attachment as well as affective sharing of positive events. The attachment
situations were distinguished from companionship situations which focused on fun,
shared activities. First, children were asked four general questions, two of which
assessed attachment needs and two of which assessed companionship needs. These
questions were taken from Kerns et al. (2006). They were then asked several contextspecific questions. Six assessed who the child would go to in situations likely to evoke
seeking an attachment figure, and four of the specific questions assessed companionship
needs. Finally, to examine who children might seek out in situations where they would
want an attachment figure, but parents are not available, they were presented with five
14
school situations. See Appendix A for a list of all general attachment and companionship
questions, specific attachment and companionship questions, and all school attachment
questions.
To examine who children nominated as primary attachment figures, we coded
their first response. Responses were coded into 10 categories: mom, dad, both mom and
dad, grandparent, sibling, peer, teacher, adult relative, nobody, and other. Many children
said “both mom and dad” and did not want to choose a single primary figure. Therefore,
the three parent categories were subsequently combined to form a single parent category.
The adult relative and other categories were rarely used, (0%-5%) and (0%-6%)
respectively, so they were not included in further analyses. We included a “nobody”
column in Table 1 to examine whether or not children say they would go to someone for
the different situations. Table 1 shows that children usually do nominate someone for
their first choice.
To examine who children nominated as secondary attachment figures we looked
at all nominations across a particular question and scored (yes/no) whether a particular
figure (i.e., sibling) was ever nominated by the child for that question. We tabled
nominations for mom, dad, grandparent, sibling, peer, and teacher for each question since
they were the most nominated categories.
To calculate the total number of people nominated, we summed the total number
of people nominated for each question. To calculate the number of unique people
nominated, we summed the total number of unique relationship types nominated for each
question.
15
Table 1. Who Children Seek First for General Attachment and Companionship, ContextSpecific Attachment and Companionship, and School Attachment Situations
General
Attachment
Sad
Scared
Companionship
Secret
Play
Parent
Peer
Sibling
Grandparent
Nobody
67%
65%
18%
10%
5%
9%
0%
2%
0%
3%
8%
0%
72%
73%
11%
18%
1%
0%
3%
0%
Context-Specific
Attachment
New School
Best Friend Fight
Summer Camp
Best Friend Move
Bully
Won Trophy
Companionship
Movie
Funny Story
Bike Ride
Talk About Books
47%
41%
76%
58%
41%
55%
23%
27%
17%
26%
27%
28%
9%
11%
3%
6%
10%
11%
4%
2%
1%
2%
2%
2%
4%
1%
0%
2%
7%
0%
28%
15%
16%
17%
52%
54%
61%
57%
10%
18%
14%
10%
4%
3%
0%
4%
2%
2%
2%
0%
School Attachment
Hurt During Recess
Won Contest
Mean Friend
Got Into Trouble
Drop Lunch Tray
Parent
18%
29%
28%
41%
22%
Peer
54%
59%
31%
39%
48%
Sibling
2%
7%
7%
5%
2%
Teacher
13%
2%
28%
4%
14%
Nobody
8%
2%
3%
7%
10%
Note: Percentages do not add to 100 because low frequency categories were not included
in the table. Items in bold indicate the highest number in each row.
RESULTS
Preliminary Analyses
Analyses were conducted separately for intact families (n=103) and single parent families
(n=10). The results for both types of families were nearly identical so the entire sample
was used for the following analyses. Chi-square analyses indicated that age, gender, and
sibling status were not significantly related.
Choice of Primary Attachment Figures
Our first analyses describe who children chose as their first choice or ever for
general attachment and companionship questions, context-specific attachment and
companionship questions, and school attachment questions. In addition, in order to
determine whether children were significantly more likely to choose some people more
than others for the attachment questions, we conducted dependent T-tests. For these
analyses, we summed the number of times children endorsed, respectively, parents, peers,
siblings, and grandparents or teachers for each set of attachment questions (i.e. general
attachment, context-specific attachment, and school attachment).
General Attachment and Companionship Questions
We first hypothesized that children would seek out parents first for general
attachment needs (i.e., sad, scared) and peers first for companionship needs (i.e., play,
16
17
sharing secrets), replicating earlier research. As shown in Table 1, when children were
asked whom they would seek first to fulfill attachment needs, they reported going to
parents at least 65% of the time. In contrast, when asked about companionship needs,
they reported going to peers at least 72% of the time. Although parents are chosen first
most often for attachment questions and peers are chosen first most often for
companionship questions, we do see a small percentage of children endorsing siblings,
and to a lesser extent grandparents, for both types of situations. The hypothesis of
preference for parents for attachment questions was further confirmed with T-tests
demonstrating that children were significantly more likely to choose parents first for the
general attachment questions (see Table 2).
Context-Specific Attachment and Companionship Questions
Next, we examined whom children would seek for context-specific attachment
and companionship situations. We expected the same general pattern of seeking out
parents first for context-specific attachment and peers first for context-specific
companionship, although we expected more variability given that children’s choices may
vary somewhat in response to situational factors. As shown in Table 1, children reported
going to parents first most often for context-specific attachment situations and peers first
most often for context-specific companionship situations. There was also more
situational variability for the context-specific questions. For example, peers were
endorsed more for context-specific attachment questions than they were for general
attachment questions, and parents were endorsed more for these types of companionship
questions than they were for general companionship questions. Even though there was
18
more situational variability for these questions, children were still significantly more
likely to choose parents first for the context-specific attachment questions (see Table 2).
19
Table 2. Mean Scores for Who Children Nominated as First Choice or Ever
First Choice
Parent
Peer
Sibling
Grandparent Teacher
General Attachment
1.32a
.28b
.14b
.02c
Context-Specific Attachment
3.16a
1.48b
.49c
.11d
School Attachment
1.38a
2.30b
.23c
Ever
Mom
Dad
Peer
Sibling
Grandparent Teacher
General Attachment
1.46a
1.27a
.56b
.51b
.24c
Context-Specific Attachment
4.03a
3.34b
2.59b
1.51c
.59d
School Attachment
2.15a
1.68b
2.82a
.88c
.61d
1.03c
Note: Means with different subscripts within a row are significantly different from one
another. Using the Bonferroni correction procedure, p < .008 for First Choice or p < .005
for Ever.
Scale range: 0 to 2 (General), 0 to 6 (Context-Specific), 0 to 5 (School).
School Attachment Questions
Next, we examined attachment situations while the child was at school because
this is a situation when secondary attachment figures might be used in the physical
absence of the parent. As shown in Table 1, it appears that for most of the school
attachment situations, children are choosing to go to someone at school first to fulfill
their attachment needs. In particular, children are turning to their peers in these
situations. In general, they are not waiting until they return home to have their parents
fulfill their attachment needs. As shown in Table 2, children were significantly more
likely to choose peers first for the school attachment questions. For the school questions
in Table 1, we have excluded the grandparent category and instead included the teacher
20
category because a sizeable number of children did endorse going to teachers in some of
these situations.
Choice of Secondary Attachment Figures
Our first analysis focused only on children’s first choice in the situations. To
examine who might function as secondary attachment figures for children, we examined
whether children ever nominated certain individuals for a specific question. This was
calculated separately for each of the general, context-specific, and school questions. For
this question we were able to distinguish between mother and father so we separated
them for descriptive purposes. When interpreting the answers to the “father” question it
is important to keep in mind that most children in the sample were living with their
biological father.
General Attachment and Companionship Questions
On average, children nominated 2.60 people for the attachment questions and 2.75
people for the companionship questions. What is most striking about examining whether
children ever nominate certain individuals is that some people who are rarely chosen as a
first choice, such as siblings and grandparents, do get nominated more by children later
on for these questions (see Table 3). For example, for “sad” grandparents are never
chosen first. However, if one looks to see if children ever report going to grandparents,
the percentage increases to 15%. Also, for “sad” siblings are only nominated first 5% of
the time. However, if one looks to see if children ever nominate siblings, the percentage
increases to 24% (see Table 3). As shown in Table 2, children are significantly more
21
likely to ever choose mom or dad than any other figure for the general attachment
questions. In addition, children are significantly more likely to ever choose a peer or
sibling than a grandparent.
22
Table 3. All People Children Seek for General Attachment and Companionship, ContextSpecific Attachment and Companionship, and School Attachment Situations
General
Attachment
Sad
Scared
Companionship
Secret
Play
Context-Specific
Attachment
New School
Best Friend Fight
Summer Camp
Best Friend Move
Bully
Won Trophy
Companionship
Movie
Funny Story
Bike Ride
Talk About Books
School Attachment
Hurt During Recess
Won Contest
Mean Friend
Got Into Trouble
Drop Lunch Tray
Mom
Dad
Peer
Sibling
Grandparent
82%
64%
62%
35%
36%
20%
24%
27%
15%
9%
14%
5%
9%
8%
82%
90%
25%
40%
2%
3%
61%
60%
83%
70%
59%
71%
44%
43%
71%
54%
54%
68%
40%
33%
44%
34%
48%
60%
18%
19%
28%
21%
21%
44%
7%
4%
14%
11%
6%
18%
29%
33%
27%
32%
33%
26%
30%
27%
76%
75%
75%
65%
34%
47%
36%
26%
5%
5%
0%
8%
Mom
29%
56%
48%
49%
33%
Dad
21%
50%
37%
38%
23%
Peer
63%
72%
40%
48%
58%
Sibling
11%
30%
15%
15%
17%
Teacher
22%
6%
38%
12%
25%
Note: Percentages over 50% are in bold.
23
Context-Specific Attachment and Companionship Questions
Next, we examined all people children would seek for context-specific attachment
and companionship questions. On average children nominated 2.65 people for the
attachment questions and 2.93 people for the companionship questions. Again, we see
the same general pattern that emerged when considering all of a child’s nominations for a
question rather than looking at first choices. Parents are still the figures who are chosen
most in specific attachment situations and peers are the figures who are chosen most in
specific companionship situations. Peers, siblings, and grandparents, who are not chosen
first very often for attachment situations, do get endorsed by children later on for the
context-specific questions (see Table 3). As shown in Table 2, children are significantly
more likely to ever choose mom than any other figure for the context-specific attachment
questions. In addition, children are significantly more likely to ever choose dad or a peer
than a sibling or grandparent, and children are also significantly more likely to ever
choose a sibling than a grandparent.
School Attachment Questions
We then examined all people children would seek for the school questions. On
average children nominated 2.43 people for the school attachment questions. Again, we
see the same general pattern emerge when we look to see whether children would ever
nominate certain individuals rather than looking at their first choice. Parents, siblings,
and teachers, who are not chosen first very often, do get endorsed by children later on for
the school attachment questions. As shown in Table 2, children are significantly more
likely to ever choose mom or peer than any other figure for the school attachment
24
questions. In addition, children are significantly more likely to ever choose dad than a
sibling or teacher.
Total/Unique Individuals Nominated
We then examined whether older children would nominate more people or more
unique individuals (i.e., number of relationship types) per question to test the hypothesis
of increased diversification of attachment networks with age. We did not find any
significant age differences in how many people total children nominated per question or
how many unique people children nominated per question.
Age, Sibling, and Gender Effects for First Choice
We then performed subsequent analyses to examine age, sibling, and gender
effects in children’s choices for the attachment questions. Because of the sample size, we
looked at age and gender effects in one MANOVA and sibling effects in a separate
MANOVA. In addition, only children were not included in the sibling MANOVA, so
examining all three variables simultaneously would have excluded only children from the
age and gender MANOVA. Also, we only looked at parents, peers, and siblings in the
MANOVA analyses because these were the categories children nominated most often.
A series of two-way MANOVAs were conducted to examine age, sibling, and
gender differences in whom children seek first for general, context-specific, and school
attachment questions. Only one effect emerged. A 2 (older, younger) by 2 (male,
female) MANOVA on children’s choices for the general attachment questions revealed
25
an age effect for whom children picked as their first choice for general attachment
questions, F(3, 108) = 3.32, p < .05 (partial η2 = .08). As shown in the first column of
Table 4, results further revealed that the age effect occurred because older children
(M=.42) were more likely than younger children (M=.15) to choose peers as their first
choice, F(1, 110) = 7.42, p < .01 (partial η2 = .06). No age effects emerged for the
context-specific attachment questions or the school attachment questions. Also, no
gender effects or sibling effects emerged for any of the attachment questions.
Table 4. Age Effects for First Choice and Ever
First Choice
M (SD)
General Attachment a
Parents
Older
1.31 (.74)
Younger
1.34 (.71)
Mom
Older
Younger
Dad
Older
Younger
Peers
Older
.42 (.63)**
Younger
.15 (.41)**
Siblings
Older
.09 (.35)
Younger
.19 (.47)
Note: a Scale range 0 to 2.
** p < .01, + p = .08
Ever
M (SD)
1.48 (.66)
1.42 (.70)
1.30 (.76)
1.26 (.73)
.74 (.78)+
.39 (.64)+
.48 (.74)
.54 (.68)
26
Age, Sibling, and Gender Effects for Choice of Secondary Attachment Figures
Age Differences
A 2 (older, younger) by 2 (male, female) MANOVA on children’s choices for the
general attachment questions revealed an age trend for who children ever nominated for
general attachment questions, F(4, 107) = 2.17, p = .08 (partial η2 = .08). The results
showed a similar pattern for who children picked as their first choice and who children
ever nominated. As shown in the second column of Table 4, results further revealed that
the age trend occurred because older children (M=.74) were more likely than younger
children (M=.39) to ever choose peers for the general attachment questions, F(1, 110) =
7.03, p < .01 (partial η2 = .06).
Similar analyses were also performed on mean scores for endorsement of parents,
peers, and siblings in context-specific attachment situations and at school. No effects
emerged in these analyses.
Sibling Differences
Results indicated that there were sibling effects only for who children ever
nominated for the general attachment questions, F(4, 99) = 3.88, p < .01 (partial η2 = .14)
and context-specific attachment questions, F(4, 99) = 3.24, p < .05 (partial η2 = .12). As
shown in Table 5, results further revealed that children with at least one older sibling
(M=.70) were more likely than children with only younger siblings (M=.29) to ever
choose a sibling for the general attachment questions, F(1, 102) = 8.00, p < .01 (partial η2
= .07). In addition, children with at least one older sibling (M=1.91) were more likely
than children with only younger siblings (M=1.14) to ever choose a sibling for context
27
specific attachment questions, F(1, 102) = 4.82, p < .05 (partial η2 = .05). Also, children
with only younger siblings (M=3.14) were more likely than children with at least one
older sibling (M=2.20) to ever choose a peer for context-specific attachment questions,
F(1, 102) = 5.60, p < .05 (partial η2 = .05).
Similar analyses were also performed on mean scores for endorsement of parents,
peers, and siblings in school attachment situations. No effects emerged in these analyses.
Table 5. Sibling Effects for Ever
General a
M (SD)
Mom
Only Younger
1.57 (.56)
At Least One Older
1.38 (.75)
Dad
Only Younger
1.40 (.69)
At Least One Older
1.16 (.76)
Peers
Only Younger
.69 (.83)
At Least One Older
.49 (.68)
Siblings
Only Younger
.29 (.52)**
At Least One Older
.70 (.77)**
a
b
Note: Scale range 0 to 2. Scale range 0 to 6.
** p < .01, * p < .05
Context-Specific b
M (SD)
4.20 (1.61)
3.99 (1.89)
3.40 (1.85)
3.36 (1.90)
3.14 (1.93)*
2.20 (1.91)*
1.14 (1.48)*
1.91 (1.79)*
Gender Differences
The previous analyses also indicated that there were gender effects for who
children ever nominated for all three types of questions (i.e. general, F(4, 107) = 6.39, p
< .001 [partial η2 = .19], context-specific, F(4, 107) = 6.67, p < .001 [partial η2 = .20],
and school attachment, F(4, 107) = 3.89, p < .01 [partial η2 = .13]). As shown in Table 6,
results further revealed a similar pattern for all three types of questions. Girls (M=1.61)
28
were more likely than boys (M=1.29) to ever choose “Mom” for general attachment
questions, F(1, 110) = 6.37, p < .05 (partial η2 = .06). Girls (M=4.52) were also more
likely than boys (M=3.49) to ever choose “Mom” for context-specific attachment
questions, F(1, 110) = 10.51, p < .01 (partial η2 = .09). Girls (M=2.53) were also more
likely than boys (M=1.75) to ever choose “Mom” for school attachment questions, F(1,
110) = 6.35, p < .05 (partial η2 = .06). Thus, girls more often than boys included their
mom in their attachment networks.
Table 6. Gender Effects for Ever
General a
M (SD)
Context-Specific b
M (SD)
School c
M (SD)
Mom
Girls
1.61 (.56)*
4.52 (1.44)**
2.52 (1.61)*
Boys
1.29 (.76)*
3.49 (1.91)**
1.75 (1.65)*
Dad
Girls
1.15 (.76)
3.30 (1.77)
1.73 (1.64)
Boys
1.41 (.71)
3.39 (1.98)
1.63 (1.53)
Peers
Girls
.66 (.76)
2.78 (2.00)
2.90 (1.55)
Boys
.47 (.69)
2.39 (1.82)
2.73 (1.72)
Siblings
Girls
.39 (.62)
1.45 (1.66)
.81 (1.15)
Boys
.64 (.78)
1.57 (1.77)
.95 (1.25)
a
b
c
Note: Scale range 0 to 2. Scale range 0 to 6. Scale range 0 to 5.
** p < .01, * p < .05
DISCUSSION
Although not the focus of the study, we did ask children about primary attachment
figures, defined as whom children picked as their first choice. The findings confirmed
our first hypothesis that children would nominate parents first for general attachment
questions and peers first for general companionship questions, which is consistent with
Kerns et al. (2006). Our results also showed the same pattern of nominating parents first
for attachment questions and peers first for companionship questions when we asked
children context-specific questions. We did find some situational variability in the
answers to the context-specific questions as parents were endorsed slightly more for the
context-specific companionship questions than they were for the general companionship
questions while peers were endorsed somewhat more for the context-specific attachment
questions than they were for the general attachment questions. All of these results
support the idea put forth by Kerns et al. (2006) that when determining whether children
are turning to parents or peers to fulfill attachment needs in middle childhood, it is
important to include information about context. In addition, our findings showed that
children are turning to peers to fulfill their attachment needs when they are at school.
This finding shows that children may use peers as secondary attachment figures when
access to their primary attachment figure is blocked (e.g., when they are at school).
The main goal of the study was to identify secondary attachment figures, defined
as individuals who children ever nominate in response to attachment questions. The logic
29
30
behind this definition is that if certain individuals do serve as secondary attachment
figures for children, children should nominated them at some point for the attachment
questions, but if they are never nominated for attachment questions we have evidence that
these individuals are not being used as secondary attachment figures. The results showed
that when one looks at whether children would ever nominate certain individuals for the
general, context-specific, and school questions people who do not get chosen first very
often, such as peers and siblings, do get endorsed more often later on even though the
rates of endorsement for some of the individuals are relatively low. This finding shows
that while peers and siblings may not be the primary attachment figures for children
during middle childhood, children may use them as secondary attachment figures.
Our findings also partially confirmed our second hypothesis that children would
show more diversification of attachment networks with age. We found mixed evidence
as the results did not confirm our hypothesis in terms of total number of people
nominated per question or number of unique people nominated per question, however the
results did confirm our hypothesis in that there was a greater report of peers in children’s
networks with age. It is important to keep in mind that the results confirmed our
hypothesis only in terms of one alternative attachment figure (peers) and only for general
attachment questions. We found that older children were more likely than younger
children to choose peers as their first choice and to ever choose peers for the general
attachment questions only. Our findings show that as children get older they are relying
on peers more as secondary attachment figures. In addition, older children are more
likely to nominate a peer as their first choice. This finding is consistent with the
31
hypothesis that older children are beginning to direct some attachment behaviors to peers
in preparation for the development of peer attachments during adolescence (Mayseless,
2005).
The findings confirmed our third hypothesis that children with at least one older
sibling were more likely than children with only younger siblings to ever nominate a
sibling as an attachment figure for the general and context-specific questions. This
finding extends previous research which found that children in early childhood will use
their older siblings as attachment figures when their parents are not available (Stewart &
Marvin, 1984). In addition, we found that children with only younger siblings were more
likely than children with at least one older sibling to ever nominate a peer as an
attachment figure for the context-specific questions only. It is interesting to note that
children with only younger siblings used peers rather than their siblings as secondary
attachment figures.
Although we did not have any a priori hypotheses regarding gender, the results
showed gender effects for who children ever nominated for general, context-specific, and
school questions. The pattern was the same for all three questions in that girls were more
likely than boys to ever choose “Mom”. Attachment theory does not clearly state what
impact gender may have on attachment. Our results are consistent with Kerns et al.
(2006), which found that girls reported higher reliance on mothers than did boys and girls
also perceived their mothers to be more available than did boys. Further research is
needed to determine whether the gender differences are specific to the attachment
32
component of mother-child relationships or if they reflect a more general trend of
increased mother-child closeness in middle childhood for girls relative to boys.
STUDY 2
One of the strengths of Study 1 was that the open-ended format allowed children
to nominate whomever they wished and as many or as few important people in their lives
as they wanted. One potential downside to this methodology is that children may not
have thought to nominate people who are used only infrequently to fulfill attachment
needs. To decrease the influence of salience, memory, and frequency of usage in Study 2
we explicitly asked children whether or not they would ever go to specific people in their
lives (e.g., mom, dad, brother, sister, teacher, grandparent, and friend) in two general
attachment situations. If children say that they would go to these individuals to fulfill
their attachment needs we have evidence that children are using these individuals as
secondary attachment figures. As a stricter criterion for whether or not we can call these
individuals secondary attachment figures, we also asked children who said they would go
to these individuals to provide us with an example of a time the person fulfilled their
attachment needs.
For Study 2, we examined age, sibling status, and gender differences just as we
did in Study 1 so that we could compare the results using the two different
methodologies. The hypotheses were similar. Our first hypothesis was that children
would show more diversification of attachment networks with age. Therefore, we
expected that older children would be more likely than younger children to say they
33
34
would go to non-parental attachment figures (e.g., siblings, teachers, grandparents, and
friends). Our second hypothesis was that children with at least one older sibling would
be more likely than children with only younger siblings to nominate a sibling as an
attachment figure. In addition, we examined gender differences in who children said they
would go to.
Finally, we explored some new questions regarding whether the use of secondary
attachment figures is related to the security of children’s attachments to their mother.
The mother-child attachment relationship may serve as a model for other close
relationships (Kerns, 1996). One way mother-child attachment may influence other close
relationships is through working models, which are rules or expectations children have
about relationships (Bowlby, 1973; Main, Kaplan, & Cassidy, 1985). Children who have
experienced responsive and caring relationships with their mother may come to expect
other close partners to be responsive and caring as well (Bowlby, 1973; Kerns, 1994).
Thus, more securely attached children may utilize secondary attachment figures more,
and they may also perceive secondary attachment figures to be more comforting, due to
their self-fulfilling expectations. An alternative hypothesis is that children who are less
securely attached to their mothers and who may therefore have experienced a less
responsive caregiver may attempt to fulfill attachment needs within other close
relationships. Sullivan (1953) suggested that the chumships children form with peers
during middle childhood may compensate for impoverished parent-child relationships.
Thus, less securely attached children may utilize secondary attachment figures more,
particularly peers, to compensate for inadequacies in parent-child relationships. Study 2
35
was designed to test these two hypotheses regarding the use of secondary attachment
figures and the security of children’s attachments to their mother.
METHOD
Participants
Participants were recruited through elementary schools. Recruitment letters were
sent home to the parents of second through sixth grade children explaining the study,
inviting the child to participate. The final sample (N = 94) consisted of 51 boys and 43
girls. Twenty-five of the participants were from a Catholic school, and 69 were from a
public school. The average age of the participating children was 9 years, with an age
range of 7 to 12 years. Ninety-five percent of the participants were Caucasian, 2 percent
were Asian, and 3 percent reported other ethnic origins. Family status categories
included two-parent intact families (88 percent), two-parent blended families (1 percent),
and single-parent families (11 percent). Fifty-three percent of the participants had at least
one older sibling, 37 percent had only younger siblings, and 10 percent were only
children.
Procedure
All data were collected at the participating elementary schools. A research assistant
worked with a teacher from each classroom to set up an interview schedule. Participating
children were taken from their classrooms to a separate interview room for approximately
15 minutes to complete the interview. Participating children in grades 3-6 completed the
interview and the Security Scale. Children in second grade only completed the interview
36
37
given concerns that they may not be able to understand the instructions for the Security
Scale.
Measures
Identification of Attachment Figures. An interview was developed to explicitly
ask children about certain people who might serve as secondary attachment figures. The
methodology is different from Study 1 because instead of asking children to
spontaneously nominate important people, they were explicitly asked about particular
people in Study 2. It is possible that children may be using certain individuals, such as
grandparents, as secondary attachment figures but they may not spontaneously nominate
these less salient secondary attachment figures who are used rarely. Therefore, we asked
children whether they would ever go to their parents, siblings, teachers, grandparents, and
peers for two situations that might elicit the need for an attachment figure. If children
still said that they would not go to these individuals, we have evidence that these
individuals probably are not used at all as secondary attachment figures.
We began by showing children a sheet of paper that had mom, dad, brother, sister,
non-custodial parent, teacher, grandparent, and friend listed on it. We then asked the
children whether or not they had each of the individuals available to them. This way, we
could be sure that we only asked children about the individuals who were relevant for
them. For this interview, children were only asked about two general attachment
situations. These questions assessed the safe haven function of the attachment
relationship. The safe haven questions were the same as the general attachment questions
38
for Study 1 (i.e., who gives you comfort when sad or scared). Children were asked if a
particular individual ever comforted the child when he/she was sad or scared. If children
nominated someone they were asked, for verification, to give an example of a time when
the person did this for them. Once they gave the example, children were asked to rate
how comforting or helpful that person was to them on a scale of 1 to 10.
Children’s responses for each person were coded as yes or no. If a particular
person was not in a child’s network, it was treated as missing data. This way, the
percentages are based only on the children who had the particular individual available in
their network.
Security of Attachment. To assess children’s perceptions of attachment security to
their mothers, children in third grade and up (N = 75) completed the Security Scale
(Kerns, Klepac, & Cole, 1996). This is a 15-item, self-report questionnaire that assesses
children’s perceptions of attachment security to a particular attachment figure. The scale
assesses perceptions of the availability and responsiveness of the parent, ease and interest
in communicating with the parent, and tendency to rely on the parent in times of stress.
Using Harter’s (1982) “Some kids … other kids …” format, the children were asked to
indicate which statement was most like them, and then to indicate whether it was “really
true” or “sort of true” for them. An example item includes: “Some kids find it easy to
trust their mom. Other kids are not sure if they can trust their mom.” Items on this scale
are scored from “insecure” (1) to “secure” (4). The Security Scale has been found to
have adequate internal consistency and test-retest reliability and there is also evidence of
validity as it has shown associations with other measures of attachment, parenting, and
39
other theoretically related constructs (Kerns, Schlegelmilch, Morgan, & Abraham, 2005).
This scale had adequate internal consistency for the current sample, Cronbach’s α = .76.
RESULTS
Choice of Attachment Figures
We first looked at whether children said they would ever go to particular
individuals if they were sad or scared (see Table 7). We tabled the data in two ways: 1)
percentage of children who said they would go to a particular figure and 2) percentage of
children who said they would go to a particular figure and were able to provide an
example. Most children who said they would go to an individual were able to generate
an example (percentages ranged from 71% - 95%). Age was dichotomized into 7-9 year
olds (younger) and 10-12 year olds (older). Chi-square analyses indicated that the ability
to generate an example was not dependent on age. We focused on the latter because it
provides a more conservative test of whether children are using secondary attachment
figures.
40
41
Table 7. Who Children Say They Would Go To and Who Children Say They Would Go
To Plus Provided an Example
Sad
Scared
Percent
Nominated
Percent Nominated
+ Example
Percent
Nominated
Percent Nominated
+ Example
Mom
98%
91%
97%
90%
Dad
91%
79%
85%
74%
Non-custodial Dad
83%
67%
50%
50%
Friend
85%
76%
59%
50%
Grandparent
82%
63%
70%
50%
Brother
62%
59%
45%
41%
Sister
57%
52%
41%
35%
Teacher
60%
48%
31%
22%
Note: Percentages are based on participants who had these individuals available in their
network.
It is interesting to note that endorsement rates in general are higher for sad than
they are for scared, suggesting that children are more selective about their choice of
figure when they are scared. As in Study 1, this shows that context is important when
asking children if they ever use a particular individual as an attachment figure.
Most children (over 74%) say they would go to their mom or their dad if they were sad or
scared, confirming their status as primary attachment figures. Other individuals (i.e., not
mom or dad) are nominated by about 22% - 76% (depending on the figure) of the
children, suggesting that these individuals are used as secondary attachment figures. For
sad, after mom and dad, most children say they would go to a friend followed by a
42
grandparent, sibling, and teacher. For scared, after mom and dad, most children say they
would go to a grandparent followed by a friend, sibling, and teacher.
Age, Sibling, and Gender Effects
We conducted a series of Chi-square analyses to look for age, sibling status, and
gender differences in who children would go to if they were sad or scared. The Chisquare analyses indicated that gender was significantly related to sibling status. Girls
were more likely than boys to have an older sibling χ2 (1) = 13.99, p < .001. Age was
dichotomized into 7-9 year olds (younger) and 10-12 year olds (older). Since we were
interested in secondary attachment figures, we did not look at mom or dad in the
following analyses. We did look at sibling, teacher, grandparent, and friend for a total of
24 tests performed. The test for sibling status differences for ‘go to friend when sad’ was
not interpreted because it had expected cell frequencies of less than 5.
Age Differences
We hypothesized that children would show more diversification of attachment
networks with age. This hypothesis was not confirmed in that there were no age
differences in whether children said they would go to specific secondary attachment
figures if they were sad or scared.
Sibling Differences
We expected that children with at least one older sibling would be more likely
than children with only younger siblings to say they would go to a sibling. Based on
Study 1, we expected that children with only younger siblings would be more likely than
43
children with at least one older sibling to say they would go to a peer. Although there
were no sibling differences for sad, we did find sibling differences for scared. Children
with at least one older sibling were more likely than children with only younger siblings
to say they would go to their sibling when they were scared, χ2 (1) = 4.64, p < .05 (see
Table 8). The results of the Mantel-Haenszel Common Odds Ratio Estimate also show
that the results are practically significant. Children with at least one older sibling were
2.65 times more likely than children with only younger siblings to go to a sibling when
scared.
Table 8. Sibling Effect for Who Children Say They Would Go To when Scared
Go To Sibling
No
23
Yes
12
Total
35
At Least One
Older Sibling
21
29
50
Total
44
41
85
Only Younger
Siblings
Gender Differences
There was one gender effect for who children said they would go to when they
were sad (see Table 9). Girls were more likely than boys to say they would go to a friend
when they were sad, χ2 (1) = 3.92, p < .05. The results of the Mantel-Haenszel Common
Odds Ratio Estimate also show that the results are practically significant. Girls were 3.67
times more likely than boys to go to a friend when sad. There was one gender effect and
one trend for who children said they would go to when they were scared. Girls were
44
more likely than boys to say they would go to their sibling when they were scared, χ2 (1)
= 8.51, p < .01 (see Table 9). The results of the Mantel-Haenszel Common Odds Ratio
Estimate also show that the results are practically significant. Girls were 3.72 times more
likely than boys to go to a sibling when scared. Boys were more likely than girls to say
they would go to their teacher when they were scared, χ2 (1) = 3.66, p = .056 (see Table
9). The results of the Mantel-Haenszel Common Odds Ratio Estimate also show that the
results are practically significant. Boys were 2.44 times more likely than girls to go to a
teacher when scared.
45
Table 9. Gender Effects for Who Children Say They Would Go To when Sad or Scared
Sad:
Go To Friend
Boy
No
11
Yes
40
Total
51
Girl
3
40
43
Total
14
80
94
Scared:
Go To Sibling
Boy
No
30
Yes
15
Total
45
Girl
14
26
40
Total
44
41
85
Scared:
Go To Teacher
Boy
No
31
Yes
20
Total
51
Girl
34
9
43
Total
65
29
94
46
Security of Attachment to Mother and Experiences with Secondary Attachment Figures
Our last question was whether the use of secondary attachment figures is related
to the security of children’s attachments to their mothers. For these analyses we used Ttests to examine whether the children who did or did not report going to a particular
attachment figure differed on security. The analyses only included secondary attachment
figures (friend, grandparent, sibling, teacher) because almost all children go to mom and
dad and because few children had non-custodial parents so there were insufficient cases
for these analyses. As shown in Table 10, the mean security scores for children who said
they would go to a non-parental figure were not very different from the means of children
who said they would not go to non-parental figures. Not surprisingly, none of the T-tests
were significant.
Table 10. Mean Security Scores for Who Children Would Go To if They Were Sad or
Scared
Sad
Scared
Yes
M
No
M
Yes
M
No
M
Friend
3.30
3.39
3.35
3.25
Grandparent
3.34
3.27
3.31
3.34
Sibling
3.35
3.21
3.27
3.35
Teacher
3.37
3.21
3.36
3.29
Next, we looked to see if children’s security with their mother was related to how
comforting they said certain individuals were when they fulfilled their attachment needs.
Children’s scores on the security scale were significantly correlated with children’s
47
reports of how comforting their teacher was when they were sad, and how comforting
they perceived their mom, dad, teacher, and grandparent to be when they were scared
(see Table 11). In general, it appears that secure children are more likely to say that
alternative adults are comforting when they use these adults to fulfill their attachment
needs. Interestingly, security with mother did not predict how comforting children rated
the support from siblings or friends.
Table 11. Children’s Security with Mom Correlated with Comfort Ratings
Sad
n
Scared n
Mom Comfort
.11
74
.24*
73
Dad Comfort
.25
59
.31*
59
Brother Comfort
-.01
28
-.32
22
Sister Comfort
.18
26
.25
20
Teacher Comfort
.37*
46
.52*
24
Grandparent Comfort
.18
60
.29*
50
Friend Comfort
.06
64
.20
48
*p < .05
DISCUSSION
Although not the main goal of the study, we asked children whether or not they
would ever go to their mom and dad (i.e., primary attachment figures) if they were sad or
scared. Our findings confirmed that parents are still the primary attachment figures for
children in middle childhood, which is consistent with previous research (Study 1; Kerns
et al., 2006; Nickerson & Nagle, 2005; Kobak et al., 2005). The results also showed that
children were more selective about their choice of figure when they were scared, which is
consistent with Study 1 and Kerns et al. (2006) in showing that context is important when
determining who children use to fulfill attachment needs in middle childhood.
The main goal of our study was to identify secondary attachment figures. The
results showed that when we explicitly ask children if they would ever go to certain
figures (siblings, peers, etc.) when they were sad or scared 22% - 76% (depending on the
figure) of the children said yes. This finding shows that while these individuals may not
be the primary attachment figures for children during middle childhood, children may use
them as secondary attachment figures. Also, the finding that the majority of children who
say they would go to these figures if they were sad or scared are able to come up with an
example gives us more confidence that children really are using these individuals as
secondary attachment figures (i.e., responses do not simply reflect an acquiescence bias).
Our findings did not confirm our first hypothesis that children would show more
diversification of attachment networks with age in that older children and younger
48
49
children did not show any differences in who they would go to if they were sad or scared.
The findings partially confirmed our second hypothesis that children with at least one
older sibling were more likely than children with only younger siblings to say they would
go to a sibling if they were scared. Finally, the results showed gender effects for who
children said they would go to if they were sad or scared. Girls were more likely than
boys to say they would go to a friend when they were sad and to a sibling when they were
scared. It is important to keep in mind that girls were significantly more likely than boys
to have an older sibling. Therefore, the fact that girls were more likely to go to a sibling
when scared may be driven by the fact that they are more likely than boys to have an
older sibling available to them. Boys were more likely than girls to say they would go to
a teacher when they were scared. It may be that girls are more likely to turn to alternative
figures who are emotionally close to them (siblings, peers) while boys are more likely to
turn to an alternative figure where there is more distance between them (teacher). In
addition, the finding that girls go to more alternative figures than boys is consistent with
research that has found that females tend to rely on broader social networks (Levitt,
Guacci-Franco, & Levitt, 1993). Also, it is interesting to note that boys say they would
go to their teacher when they were scared. Perhaps boys were thinking about a time
when they were threatened by a bully (which usually happens at school) and this is why
boys were more likely than girls to say they would go to a teacher when they were scared.
Our findings also showed that there was no difference between more securely
attached and less securely attached children in terms of whether they go to alternative
figures when they were sad or scared. However, secure children were more likely to rate
50
alternative adults as more comforting when they used these adults to fulfill their
attachment needs. It is interesting to note that security with mother did not predict how
comforting children rated the support from siblings or friends. Children more securely
attached to their mother may generalize expectations of comfort only to adults.
GENERAL DISCUSSION
The two studies in this paper have given us information about the identity and use
of attachment figures in middle childhood. First, both studies showed that parents are
still the primary attachment figures for children in middle childhood, even in a variety of
contexts, which is consistent with previous research (Kerns et al., 2006; Nickerson &
Nagle, 2005; Kobak et al., 2005). Second, and most importantly, the two studies provide
us information on the identity and use of secondary attachment figures in middle
childhood, which was previously a virtually unexplored topic. Crowell and Waters
(1994) suggest that, “in a storm, any port will do” when children are in extremely
stressful situations. We have shown that children may turn to other people even in less
critical situations. For example, when primary attachment figures are not physically
available, such as when children are at school, children do turn to others, particularly
peers, to fulfill their attachment needs.
Neither study may provide us with exact estimates of rates of utilizing secondary
attachment figures. Rather, the two methodologies provide us with two different ways of
identifying secondary attachment figures. In Study 1 we allow children to generate their
own list of important people and we allow them to nominate anyone they want in
response to the attachment questions. Therefore, in Study 1, we may be identifying
secondary attachment figures children use frequently. Children nominated approximately
51
52
2-3 people per question, revealing that children are fairly selective in who they will turn
to. In Study 2, we explicitly ask children about particular figures. Therefore, in Study 2,
we may be identifying secondary attachment figures who may be less salient to children
and are used less frequently. As for the identity of secondary attachment figures, Study 1
showed us that children in middle childhood may use peers and siblings, and to a lesser
extent, grandparents and teachers. Study 2 showed us that 22% - 76% of the children use
peers, siblings, grandparents, or teachers as secondary attachment figures.
When comparing the two studies, our findings showed that when children are
asked explicitly about certain individuals instead of nominating people on their own, the
percentage of children who said they would use these individuals increases. Also, the
results showed that the rank ordering of who children would go to changed slightly from
Study 1 to Study 2. Specifically, more children said they would go to a grandparent in
Study 2. Both of these findings show that children may use certain individuals as
secondary attachment figures but may not nominate them very often in Study 1 because
they are less salient to children (e.g., infrequently used). Therefore, the methodology of
Study 2 (asking children explicitly about certain individuals) may ensure that we do not
overlook individuals who do in fact serve as secondary attachment figures for children.
Additionally, these two studies provide us with information about the use of
secondary attachment figures and how it might vary with certain child characteristics.
For example, we found that older children are more likely than younger children to
nominate a peer for general attachment questions in Study 1, providing support for the
proposed hypothesis that children in middle childhood are preparing for peer attachments
53
during adolescence (Mayseless, 2005). Also, we found that children with at least one
older sibling are more likely than children with only younger siblings to choose a sibling
for general and context-specific attachment questions in Study 1 and to say they would go
to a sibling if they were scared in Study 2. Both of these findings show that older
siblings are more likely than younger siblings to be used as secondary attachment figures
in middle childhood, which is consistent with previous research which has shown that
older siblings are used as secondary attachment figures in early childhood (Stewart &
Marvin, 1984). In addition, in Study 2 we found that children more securely attached to
their mothers are more likely to report that alternative adults are comforting when they
use them to fulfill their attachment needs. Thus, Study 1 and Study 2 have shown us that
the identity and use of secondary attachment figures may differ depending on a child’s
age, sibling status, and security with mother.
As mentioned above, previous research has focused on who are children’s
attachment figures for children in middle childhood, with attention to whether peers
might function as attachment figures (Hazan & Zeifman, 1994; Kerns et al., 2006;
Nickerson & Nagle, 2005; Kobak et al., 2005). The two studies in the current paper
extend this research as they have shown that while peers may not be primary attachment
figures for children in middle childhood, they often do serve as secondary attachment
figures, especially when parents may not be readily accessible. It would be interesting to
extend this research into the adolescent years to see if and when adolescents use peers as
primary attachment figures. While these two studies do identify who children utilize as
secondary attachment figures in middle childhood, the rates are fairly low for siblings,
54
grandparents, and teachers, especially in Study 1. This does not mean that these figures
are unimportant. Instead, these figures may more often play a different role in children’s
lives. For example, siblings may provide one another with protection and they may be
social models for one another (Lewis, 2005). In addition, grandparents may rarely be
utilized as attachment figures but may provide children with validation and self-esteem
enhancement.
The majority of the participants from the current sample were Caucasian and
came from two-parent, intact families. Therefore, it is important to keep in mind that our
findings and conclusions may not generalize to other groups. It would be interesting to
extend this research to other ethnic groups and more diverse family structures. It is
possible that we may identify different or additional secondary attachment figures for
children from different ethnic groups or family structures. For example, in some ethnic
groups where it is customary for children to be surrounded by a large, extended family
we might find that children more often nominate people such as adult relatives and
cousins as secondary attachment figures. Therefore, we might get a fuller understanding
of the identity of secondary attachment figures if we include children from more diverse
ethnic groups and family structures.
In conclusion, these two studies have provided novel information on the identity
and use of secondary attachment figures in middle childhood. The two studies have
shown that peers, and to a lesser extent siblings, grandparents, and teachers, may serve as
secondary attachment figures for children in middle childhood. These studies have also
shown that the identity and use of secondary attachment figures may differ depending on
55
context and on children’s age, sibling status, and security with mother. Adolescence has
been identified as the age period when attachments to peers emerge (Allen & Land,
1999). It would be interesting to extend this research into adolescence to see when
attachments to peers emerge and what factors are related to the timing and usage of peers
as attachment figures.
REFERENCES
Ainsworth, M. D. S. (1989). Attachments beyond infancy. American Psychologist, 44,
709-716.
Allen, J. P., & Land, D. (1999). Attachment in adolescence. In J. Cassidy & P. R. Shaver
(Eds.), Handbook of attachment: Theory, research, and clinical applications (pp.
319-335). New York: Guilford Press.
Antonucci, T. C. (1986). Hierarchical mapping technique. Generations, 10, 10-12.
Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation. New York: Basic Books.
Bowlby, J. (1982). Attachment and loss: Vol 1. Attachment (2nd ed.). New York: Basic
Books.
Cassidy, J. (1999). The nature of the child’s ties. In J. Cassidy & P. R. Shaver (Eds.),
Handbook of attachment: Theory, research, and clinical applications (pp. 3-20).
New York: Guilford Press.
Crowell, J. A., & Waters, E. (1994). Bowlby’s theory grown up: The role of attachment
in adult love relationships. Psychological Inquiry, 5, 31-34.
Hazan, C., & Zeifman, D. (1994). Sex and the psychological tether. In K. Bartholomew
& D. Perlman (Eds.), Attachment Process in Adulthood (pp. 151-178). Bristol,
PA: Jessica Kingsley Publishers.
56
57
Howes, C. (1999). Attachment relationships in the context of multiple caregivers. In J.
Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and
clinical applications (pp. 671-687). New York: Guilford Press.
Howes, C., & Hamilton, C. E. (1992). Children’s relationships with caregivers: Mothers
and childcare teachers. Child Development, 63, 859-866.
Howes, C., & Oldham, E. (2001). Processes in the formation of attachment relationships
with alternative caregivers. In A. Goncu & E. L. Klein (Eds.), Children in play,
story, and school (pp. 267-287). New York: The Guilford Press.
Kerns, K. A. (1994). A developmental model of the relations between mother-child
attachment and friendship. In R. Erber & R. Gilmour (Eds.), Theoretical
frameworks for personal relationships (pp. 129-156). Hillsdale, NJ: Lawrence
Erlbaum Associates, Inc.
Kerns, K. A. (1996). Individual differences in friendship quality: Links to child-mother
attachment. In W. M. Bukowski, A. F. Newcomb, & W. W. Hartup (Eds.), The
company they keep: Friendship in childhood and adolescence (pp. 137-157). New
York: Cambridge University Press.
Kerns, K. A., Aspelmeier, J. E., Gentzler, A. L., & Grabill, C. M. (2001). Parent-child
attachment and monitoring in middle childhood. Journal of Family Psychology,
15, 69-81.
Kerns, K. A., Klepac, L., & Cole, A. (1996). Peer relationships and preadolescents’
perceptions of security in the child-mother relationship. Developmental
Psychology, 32, 457-466.
58
Kerns, K. A., Schlegelmilch, A., Morgan, T. A., & Abraham, M. M. (2005). Assessing
attachment in middle childhood. In K. A. Kerns & R. A. Richardson (Eds.),
Attachment in Middle Childhood (pp. 46-70). New York: Guilford Press.
Kerns, K. A., Tomich, P. L., & Kim P. (2006). Normative trends in children’s perceptions
of availability and utilization of attachment figures in middle childhood. Social
Development, 15, 1-22.
Kobak, R., Rosenthal, N., & Serwik, A. (2005). The attachment hierarchy in middle
childhood: Conceptual and methodological issues. In K. A. Kerns & R. A.
Richardson (Eds.), Attachment in Middle Childhood (pp. 71-88). New York:
Guilford Press.
Lewis, M. (2005). The child and its family: The social network model. Human
Development, 48, 8-27.
Levitt, M. J., Guacci-Franco, N., & Levitt, J. L. (1993). Convoys of social support in
childhood and early adolescence: Structure and function. Developmental
Psychology, 29, 811-818.
Main, M., Kaplan, N., & Cassidy, J. (1985). Security of infancy, childhood, and
adulthood: A move to the level of representation. In I. Bretherton & E. Waters
(Eds.), Monographs of the Society for Research in Child Development, 50, (Serial
No. 209).
Marvin, R. S., & Britner, P. A. (1999). Normative development: The ontogeny of
attachment. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment:
59
Theory, research, and clinical applications (pp. 44-67). New York: Guilford
Press.
Mayseless, O. (2005). Ontogeny of attachment in middle childhood: Conceptualization of
normative changes. In K. A. Kerns & R. A. Richardson (Eds.), Attachment in
Middle Childhood (pp. 1-23). New York: Guilford Press.
Nickerson, A. B., & Nagle, R. J. (2005). Parent and peer attachment in late childhood and
early adolescence. Journal of Early Adolescence, 25, 223-249.
Stewart, R. B., & Marvin, R. S. (1984). Sibling relations: The role of conceptual
perspective-taking in the ontogeny of sibling caregiving. Child Development, 55,
1322-1332.
Sullivan, H. S. (1953). The interpersonal theory of psychiatry. New York: Norton.
Webb, N. (1984). Advantages of multiple caretaking experiences. In N. Webb (Ed.),
Preschool children with working parents (pp. 131-156). New York: University
Press of America.
APPENDIX A
60
61
General Attachment and Companionship Questions, Context-Specific Attachment and
Companionship Questions, and School Attachment Questions for Study 1
General Attachment
1. If you felt really sad, who would you go to first?
2. If you felt scared, who would you go to first?
General Companionship
1. If you had a special secret, who would you want to tell it to first?
2. If you wanted someone to play with, who would you go to first?
Context-Specific Attachment
1. Imagine that you are getting ready to go to a new school and you are a little bit worried. Who
would you most want to talk to about how you feel about going to this new school?
2. Imagine that you get into a fight with one of your best friends and you feel really lonely and
sad. Who would you most want to talk to about this?
3. Imagine that you were away at summer camp for two weeks. At the end of the first week you
get to call someone. Who would you most want to call?
4. Imagine that your best friend is moving so far away that you won’t see him/her anymore and
you are really sad. Who would you most want to talk to about this?
5. Imagine that on the way home from school a bully stops you and picks on you. You feel really
upset and afraid. Who would you most want to talk to about this?
6. Imagine that you won a trophy and you are very excited. Who would you most want to show
your trophy?
Context-Specific Companionship
1. Imagine that you want to go see a new movie. Who would you most want to go with you?
2. Imagine that you have a really funny joke or story you want to tell someone. Who would you
most want to tell your joke or story to?
3. Imagine that you want to go for a bike ride. Who would you most want to go with you?
4. Imagine that you wanted to talk to someone about books you’ve read and movies you’ve seen.
Who would you most want to talk to?
School Attachment
1. Imagine that you are playing outside during recess and you fall down and get hurt. Who
would you want to comfort you first?
2. Imagine that when you are at school, your teacher told you that you won a contest. Who
would you want to tell first?
3. Imagine that you are at school and one of your friends says something really mean to you.
Who would you want to talk to about this first?
4. Imagine that you are at school and you are upset because you just got in trouble. Who would
you want to talk to about this first?
5. Imagine that you are at school and you drop your lunch tray. You feel really embarrassed.
Who would want to talk to about this first?