Clinical Psychology and Psychotherapy

Clin. Psychol. Psychother. 12, 67–79 (2005)

Assessment The Inventory of

Parent and Peer
Attachment—Revised
(IPPA-R) for Children:
A Psychometric
Investigation
Eleonora Gullone* and Kym Robinson
Department of Psychology, Monash University, Australia

Despite the importance of attachment theory in developmental

research, there is an absence of valid and reliable tools with which to
assess attachment beyond infancy and prior to late adolescence. To
address this issue the present investigation reports on the revision
and psychometric evaluation of the Inventory for Parent and Peer
Attachment (IPPA, Armsden & Greenberg, 1987), a measure of attach-
ment developed for use with older adolescents and young adults. The
simplified revised measure (IPPA-R) was administered along with the
Parental Bonding Inventory (PBI; Parker, Tupling, & Brown, 1979)
and the Coopersmith Self-Esteem Inventory (SEI) to two samples of
youth: 118 youth aged 9–11 years; 163 youth aged 14–15 years. Our
findings provided support for the reliability and validity of the
revised measure. It was concluded that the IPPA-R constitutes a
useful tool for the assessment of both parent and peer attachment in
youth aged between 9 and 15 years. Copyright © 2005 John Wiley &
Sons, Ltd.

According to Bowlby’s (1969) evolutionary– infant and caregiver that may be characterized in
ethological attachment theory, the infant is terms of regulation of infant emotion.
endowed with an ‘attachment behavioural An impressive amount of research has been
system’, which ensures sufficient proximity to carried out, providing strong empirical support
primary caregivers to promote the infant’s sur- for most of the key components of attachment
vival. Essentially, attachment theory describes a theory (Carlson & Sroufe, 1995; Rutter, 1995).
fundamental normative process in early develop- Given Bowlby’s proposals that the attachment
ment defined in terms of behavioural and affective bond develops from birth and that it promotes sur-
regulation. The attachment relationship represents vival in the young infant, it is not surprising that
a ‘special type of social relationship’ (Bowlby, 1969, the majority of this research has focussed on
p. 376) and involves an affective bond between infancy or early childhood. It is equally not sur-
prising that methods and measures for assessing
attachment have been developed primarily for
*Correspondence to: Associate Professor, E. Gullone, PhD,
FAPS, Department of Psychology, Monash University, these early years of life. Perhaps the most fre-
Monash, Victoria, Australia, 3800. quently cited method of assessing attachment is
E-mail: E.Gullone@med.monash.edu.au the observational ‘strange situation’ method devel-

Copyright © 2005 John Wiley & Sons, Ltd.

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/cpp.433
68 E. Gullone and K. Robinson

oped by Mary Ainsworth and colleagues over-control to encouragement of independence

(Ainsworth, Blehar, Waters, & Wall, 1978). Using and autonomy. Studies using this measure with
this method, researchers demonstrated that indi- adolescent samples have consistently found a pos-
vidual differences in patterns of behaviour during itive association between self-esteem and the Care
infancy can reliably be classified as secure or inse- dimension and a negative association between self-
cure. Consistent with Bowlby’s theoretical propos- esteem and the Overprotection dimension (e.g.
als, these patterns of behaviour have been shown Herz & Gullone, 1999; Rice & Cummins, 1996).
to be moderately stable over long periods of time Another self-report measure of attachment is the
under stable family and caretaking conditions (see, Inventory of Parent and Peer Attachment (IPPA)
e.g. Fraley, 2002; Hamilton, 2000; Lewis, Feiring, & developed by Armsden and Greenberg (1987). In
Rosenthal, 2000). Moreover, an affective bond char- line with Bowlby’s attachment theory, the IPPA
acterized by warmth, availability, trust, and measures psychological security derived from rela-
responsiveness with at least one individual tionships with significant others. In particular,
throughout the lifespan has been proposed to be this measure assesses the quality of attachment to
important for psychological adjustment (see, e.g. parents and peers. For each of the parent and peer
MacDonald, 1992). Indeed, much research has scales of the IPPA, items are included to assess
shown such a bond to be an important factor in three aspects of attachment including trust, com-
predicting resilience in individuals faced with sub- munication and alienation. Specifically, the Trust
stantial adversity (e.g. Cicchetti & Garmezy, 1993; scale measures the degree of mutual understand-
O’Connell-Higgins, 1994). ing and respect in the attachment relationship, the
Reflecting the increased recognition of the impor- Communication scale assesses the extent and
tance of attachment across the lifespan (see, e.g. quality of spoken communication and the Alien-
Griffin & Bartholomew, 1994), research into attach- ation scale assesses feelings of anger and inter-
ment relationships beyond the preschool years is personal alienation.
increasingly being conducted (e.g. Armsden & In their psychometric investigation of the IPPA,
Greenberg, 1987; Gullone, King, & Ollendick, involving a sample of adolescents aged between
2002). Self-report measures of attachment have 16 and 20 years, Armsden and Greenberg (1987)
been utilized to assess attachment in adolescence, found significant inter-correlations between all
in comparison with the observational measures subscales. Specifically, Trust and Communication
utilized in infancy (Bretherton, 1985). Such were found to be positively correlated (r = 0.76 for
measures allow researchers to gain insight into each of the parent and peer scales). In contrast, the
the cognitively based representations of individu- Alienation subscale was inversely correlated with
als’ internal working models (Berlin & Cassidy, each of the Communication (r = -0.70 for parents
1999). and r = -0.40 for peers) and Trust (r = -0.76 for
However, only a very limited number of psy- parents and r = -0.46 for peers) subscales.
chometrically validated self-report measures that Past research has demonstrated that relation-
assess cognitive representations of attachment ships characterized by high levels of acceptance,
have been developed (e.g. West, Rose, Spreng, warmth and trust between children and adoles-
Sheldon-Keller, & Adam, 1998). These have pri- cents and their parents as well as their peers are
marily been developed for use with older adoles- positively associated with self-esteem (e.g. Gecas,
cents and/or adults. For example, the initial 1971; Greenberg, Siegal, & Leitch, 1983; Walker
version of the Parental Bonding Inventory (PBI) & Green, 1986). Consistent with such research,
developed by Parker et al. (1979) requires that Armsden and Greenberg (1987) found that higher
adults retrospectively respond in relation to their quality attachments with significant others (both
experiences of parenting during their adolescent parents and peers), as assessed with the IPPA, were
years. Versions enabling current reporting of par- related to more positive perceptions of oneself as
enting perceptions by adolescents have since been a family member and to higher scores on self-
developed (e.g. Cubis, Lewis, & Dawes, 1989; Herz concept as measured using the Tennessee Self-
& Gullone, 1999; Klimidis, Minas, & Ata, 1992; Concept Scale (Fitts, 1965). Each of peer and parent
Klimidas, Minas, Ata, & Stuart, 1992). attachment scores were found to be significant pre-
The PBI assesses attachment on two dimensions, dictors of adolescent self-esteem.
these being parental nurturance, ranging from It is noteworthy, however, that the IPPA has been
parental care and involvement to indifference and demonstrated to be a valid measure of attachment
neglect, and parental overprotection, ranging from for the developmental periods of mid- to late ado-

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
IPPA-R for Children 69

lescence and early adulthood. Therefore, there Measures

remains, in large part, a paucity of assessment tools
The Inventory of Parent and Peer Attachment
available to measure attachment representations
(IPPA; Armsden & Greenberg, 1987)
post-infancy and pre-adolescence (Green &
The original version of the IPPA was developed
Goldwyn, 2002). Given the identified need for a
to measure attachment in older adolescents. It
measure of attachment appropriate for use with
assesses the positive and negative affective and
children and young adolescents, in the present
cognitive dimensions of adolescents’ relationships
study we report the revision and psychometric
with their parents and close friends. For each of the
evaluation of the IPPA for children and young ado-
28 items assessing parent attachment and the 25
lescents. On the basis of the research reviewed
items assessing peer attachment, respondents are
above, we propose that subscales of the revised
required to rate the degree to which each item is
IPPA will correlate significantly both with each
true for them on a five-point scale ranging from
other and with a valid measure of self-esteem (i.e.
‘Almost always or always true’ to ‘Almost never or
Coopersmith Self-Esteem Inventory), in line with
never true’. The items in each of the scales (i.e.
correlations reported by Armsden and Greenberg
parent, peer) were demonstrated through principal
(1987). We also predict that scores on each of the
components analysis to cluster into three factors
two dimensions of the PBI will correlate signifi-
(trust; communication; anger and alienation). In
cantly with the IPPA-R subscales. Specifically, we
the present study, the IPPA-R was scored in accor-
predict that the Alienation subscale scores (parent
dance with directions provided by Armsden and
and peer) will correlate positively with scores on
Greenberg (1987) (see Armsden & Greenberg, 1987,
the PBI Overprotection dimension but negatively
for details).
with the PBI Care dimension. In addition, we
Using two samples of undergraduate students
predict that each of the Communication and Trust
who ranged in age from 16 to 20 years, Armsden
subscales will correlate positively with the Care
and Greenberg (1987) reported good internal
dimension but negatively with the Overprotection
consistency for the IPPA with Cronbach’s alpha
dimension.
coefficients ranging between 0.72 and 0.91 for the
sub-scales across both the parent and peer scales.
Good test–retest reliability for a sample of 18–20-
METHOD year-olds over a three-week period was also
Participants reported with correlation coefficients ranging
between 0.86 for peer attachment and 0.93 for
Twenty-five schools were contacted to participate parent attachment (Armsden & Greenberg, 1987).
in the study, out of which six primary and three Convergent validity has been reported on the
secondary schools agreed to participate. This rep- basis of moderate correlations between the IPPA
resents a 36% response rate for schools. The total and other measures, including the Family Self-
number of students recruited from the nine partic- Concept subscale of the Tennessee Self-Concept
ipating schools was 281. The sample was divided Scale (r = 0.78 with parent attachment; r = 0.28 with
into two age-groups for data analysis. The first, peer attachment) and the Social Self-Concept sub-
constituting the child group, was aged between 9 scale (r = 0.46 with Parent attachment; r = 0.57 with
and 11 years (age: M = 9.97, SD = 0.72) and com- Peer attachment). Also, significant positive corre-
prised a total of 118 (91 males, 27 females) grade lations between parent attachment and the Cohe-
four and five students. The second group was sion (r = 0.56), Expressiveness (r = 0.52) and
aged between 14 and 15 years (age M = 14.16, SD Organization (r = 0.38) subscales of the Family
= 0.37) and comprised a total of 163 (33 males, Environment Scale (FES) have been reported. In
130 females) participants. This group will subse- addition, significant negative correlations with the
quently be referred to as the young adolescent Conflict (r = -0.36) and Control (r = -0.20) subscales
sample. Only participants who were given writ- of the FES were reported (Armsden & Greenberg,
ten parental consent to participate were involved 1987).
in the study. The overall parent consent rate
was 38.76%. This is consistent with other studies
requiring active consent (Hollman & McNamara, The Self-Esteem Inventory—School Form
1999). (SEI; Coopersmith, 1981)
The SEI was used to measure self-esteem. This
measure was designed for use with respondents

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
70 E. Gullone and K. Robinson

aged between 8 and 15 years. It consists of 58 items dimension (see, e.g. Parker, 1979a, 1979b, 1983;
of which 50 items constitute a Global self-esteem Silove, 1986).
scale and eight items constitute a lie or defensive- Of particular relevance in the current study, the
ness scale scale). Respondents are required to rate version of the PBI used was the revision reported
each item on a two-point scale as either ‘Like me’ by Herz and Gullone (1999) wherein the wording
or ‘Unlike me’. Higher scores indicated higher self- of the PBI was changed from retrospective to
esteem. Although the SEI also yields four sub-scale current. Demonstrating validity for this revised
scores, for the present purposes only the total version, in two adolescent samples (an Anglo-
Global self-esteem score was used. Australian and a Vietnamess sample), Herz and
The SEI has been shown to have good psycho- Gullone reported significant positive correlations
metric properties. Kudar–Richardson reliability between the SEI and the Care dimensions of the
estimates and split-half reliability coefficients have PBI and significant negative correlations with the
been determined across large samples to range Overprotection dimension. Good internal consis-
between 0.81 and 0.92. The test–retest reliability tency for each of the dimensions was also reported
of the measure over a 12-month period has also in each of the adolescent samples.
been shown to be good, with a test–retest coeffi-
cient of 0.64 for 104 children in grades 5 and 6
Procedure
(Coopersmith, 1981). Adequate validity has also
been reported on the basis of significant positive The IPPA items were reviewed and revised with a
correlations between the SEI and an achievement view to simplifying the wording so as to promote
(r = 0.33) as well as an intelligence measure (r = their comprehension by children and younger ado-
0.30) (Coopersmith, 1981). lescents. In addition, the five-point response scale
was simplified to a three-point scale with ‘always
true’, ‘sometimes true’ and ‘never true’ as the
The Parental Bonding Instrument response options. As shown in Table 1, 16 of the 28
(PBI; Parker et al. 1979) parent attachment items and 14 of the 25 peer
As previously noted, the PBI was originally attachment were revised. The revised items, as
developed to assess adults’ perceptions of their detailed in Table 1, were independently evaluated
parents’ behaviours and attitudes in their first 16 and endorsed by two primary school teachers prior
years of life. The instrument comprises 25 state- to being administered to the respondents.
ments assessing each of the Care (12 items) and Prior to data collection, approval was obtained
Overprotection (13 items) dimensions. The Care from the University ethics committee and the
item ‘Is emotionally cold to me’ was excluded governing body of Catholic schools in Victoria,
upon request from the university ethics committee, Australia. Following this, the principals of all
who considered the item to be inappropriate for schools approached to participate were sent a letter
the young sample involved in the present study. informing them of the study. The final nine princi-
Respondents were required to rate each of the pals who gave permission for their schools to par-
remaining 24 items as to how closely the statement ticipate also nominated the classes of children in
represented their parenting perceptions using a their school which would be approached to be
four-point Likert scale ranging from very unlike to involved in the study.
very like. All parents of children in nominated classes
The original version of the PBI has been shown were provided with an explanatory statement and
to have sound psychometric properties. Parker and consent form, via their children. Parents who pro-
colleagues (1979) reported a split-half reliability of vided consent for their child to participate in the
0.88 for the Care scale and 0.74 for the Overpro- study were required to return a signed consent
tection scale. Test–retest reliability coefficients over form. Children were also asked to complete a
a three-week period were reported to be 0.76 for consent form as per ethics committee requirement.
the Care scale and 0.63 for the Overprotection The questionnaires were completed on a small
scale. Predictive validity has also been reported group basis and in a quiet room at the child’s
through a number of studies examining the asso- school during school hours. The measures were
ciation with psychosocial morbidity. For example, counterbalanced across different groups to control
people with anxiety, phobic and depressive disor- for possible order effects. The voluntary nature of
ders have been found to score higher on the Over- the child’s participation was clearly stated prior to
protection dimension but lower on the Care distributing the questionnaires. It was also empha-

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
IPPA-R for Children 71

Table 1. Original and revised items of the IPPA

Original item Revised item (where changed)

Parent Attachment Items
I feel my parents are successful as parents. My parents are good parents.
I have to rely on myself when I have a problem to solve. I can depend on my parents to help me solve a
problem (R)
I like to get my parents’ point of view on things I’m I like to get my parents’ view on things I’m
concerned about. worried about.
I feel it’s no use letting my feelings show. It helps to show my feelings when I’m upset (R).
My parents sense when I’m upset about something. My parents can tell when I’m upset about something.
Talking over my problems with my parents makes me I feel silly or ashamed when I talk about my
feel ashamed or foolish. problems with my parents.
I get upset easily at home. I easily get upset at home.
When we discuss things, my parents consider my point When I talk about things with my parents they listen
of view. to what I think.
My parents trust my judgement. My parents listen to my opinions.
My parents encourage me to talk about my difficulties. My parents support me to talk about my worries.
I don’t know whom I can depend on these days. I don’t know who I can depend on.
When I am angry about something, my parents try to When I am angry about something, my parents try
be understanding. to understand.
My parents don’t understand what I’m going through My parents understand my problems (R).
these days.
I can count on my parents when I need to get something I can count on my parents when I need to talk about
off my chest. a problem.
I feel that no one understands me. No one understands me.
If my parents know something is bothering me, they ask If my parents know that I am upset about
me about it. something, they ask me about it.
Peer Attachment Items
I like to get my friends’ point of view on things I’m I like to get my friends’ opinions on things I’m
concerned about. worried about.
My friends sense when I’m upset about something. My friends can tell when I’m upset about something.
When we discuss things, my friends consider my point When we talk, my friends listen to my opinion.
of view.
Talking over my problems with my friends makes me feel I feel silly or ashamed when I talk about my
ashamed and foolish. problems with my friends.
My friends encourage me to talk about my difficulties. My friends support me to talk about my problems.
I feel the need to be in touch with my friends more often. I feel the need to be around my friends.
My friends don’t understand what I’m going through My friends don’t understand my problems.
these days.
I feel alone or apart when I am with my friends. I do not feel like I belong when I am with my friends.
I feel my friends are good friends. My friends are good friends
When I am angry about something, my friends try to be When I am angry about something, my friends try
understanding. to understand.
My friends are concerned about my well-being. My friends care about the way I feel.
I can count on my friends when I need to get something I can count on my friends to listen when something
off my chest. is bothering me.
It seems as if my friends are irritated with me for no My friends get annoyed with me for no reason.
reason.
I tell my friends about my problems and troubles.
If my friends know something is bothering me, they ask If my friends know that I am upset about something,
me about it. they ask me about it.
(R) refers to revised items that are scored in reverse to their equivalent item on the original IPPA.

sized to the children that there were no correct for questionnaire completion and the individual
or incorrect answers but rather that it was the questionnaire items were read aloud to the stu-
answers that were most true for them that were of dents by the administrator of the questionnaires
interest to us. For the child sample, the directions (i.e. second author).

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
72 E. Gullone and K. Robinson

All participants completed the SEI and the mod- For example, it was found that 10 cases had
ified version of the IPPA but only a sub-sample of missing data for item 45 of the SEI. There were
participants completed the PBI. For the Parent other instances of missing data, although this was
Attachment section of the IPPA-R, participants the item missing the largest number of values.
were asked to answer with regard to their rela- Given that the missing data appeared to be ran-
tionship with both parents. The sub-sample who domly scattered across cases and questionnaires, it
completed the PBI comprised 43 participants (21 was considered acceptable to replace the missing
males, 22 females); 15 in the child group and 28 item value with the mean for that particular item.
in the young adolescent group. Children were To detect possible outliers, standardized scores
instructed to nominate either mother or father and were computed for the total scores of each of the
to respond to the PBI items in relation to the nom- variables. Scores in excess of ±3.29 were identified
inated parent. The average time required by the as outliers. This resulted in the deletion of four
children to complete the questionnaires varied cases, reducing the overall sample size from 281 to
between 20 and 30 minutes depending on the age 277.
of participants and the number of questionnaires
administered.
Descriptive Statistics
A total score for each of the IPPA-R Parent and Peer
RESULTS
Attachment scales was calculated by obtaining a
Below we report descriptive statistics for the mea- sum of the Trust and Communication subscales
sures used. These are followed by the internal con- and then subtracting the Alienation subscale score.
sistency coefficients for the IPPA-R total scores and The IPPA-R overall and sub-scale score means and
subscales by age group and sex. We then report the standard deviations are presented in Table 2. These
correlations between the IPPA-R scores and the are provided for the overall sample as well as for
other measures (i.e. PBI, SEI). We also report cor- each age group and sex. Similarly, total SEI score
relations between the two scales of the IPPA-R and PBI nurturance and overprotection means and
(i.e. Parent Attachment and Peer Attachment) as standard deviations for the overall sample as well
well as inter-correlations between the IPPA-R sub- as for each age and sex group are provided in Table
scales. These are reported separately for each of the 2. No significant group differences were found for
age and sex groups. self-esteem. In contrast, for the IPPA-R scores, there
Prior to conducting statistical analyses, integrity were significant age-group and sex differences for
issues related to the data were examined. Missing all of the IPPA-R sub-scales, with one exception
data and outliers were identified and dealt with. being the Alienation sub-scale of the Parent Attach-

Table 2. Means and standard deviations for the SEI, and the IPPA-R by age group and sex

Measure Children Adolescents Males Females Overall sample

M SD M SD M SD M SD M SD
Total SEI score 79.95 16.49 79.44 14.16 ns 77.66 16.01 78.39 14.60 ns 79.09 15.17
IPPA-R—Parents
Trust 14.83 2.00 13.83 2.67*** 14.79 1.99 13.81 2.70** 14.22 2.48
Communication 14.70 2.74 12.79 3.42*** 14.26 2.99 13.02 3.43** 13.53 3.31
Alienation 4.39 3.14 4.92 3.56 ns 3.94 2.80 5.26 3.69** 4.72 3.41
Total score 25.14 6.78 21.70 8.71*** 25.11 6.37 21.57 8.97*** 23.03 8.18
IPPA-R—Peers
Trust 14.26 3.06 15.76 2.97*** 13.94 3.07 16.06 2.79*** 15.19 3.08
Communication 11.51 3.73 14.12 3.55*** 10.84 3.39 14.70 3.29*** 13.11 3.83
Alienation 4.22 2.48 3.44 2.32** 4.40 2.39 3.28 2.32*** 3.74 2.41
Total score 21.55 8.16 26.45 7.94*** 20.37 7.73 27.48 7.52*** 24.56 8.36
PBI Care 27.87 3.70 26.71 4.81 ns 26.75 3.04 27.43 5.43 ns 27.12 4.44
PBI Overprotection 15.73 7.38 12.14 6.89 ns 13.67 6.56 13.14 7.89 ns 13.40 7.19
t-test outcomes noted as ns = not significant, ** p < 0.01, ***p < 0.001.

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
IPPA-R for Children 73

ment scale. No significant age-group differences determined for the two dimensions of the PBI, for
were found for this sub-scale. The significant age- overall SEI, and for the overall IPPA-R Peer and
group differences were due to the child sample Parent Attachment scales as well as for each IPPA-
scoring significantly higher on the Communication R sub-scale. These were determined for the overall
and Trust sub-scales as well as on the overall sample and by age and sex groups. The results
Parent Attachment scale compared with the ado- are shown in Table 3. As is evident upon examina-
lescent sample. In contrast, the differences relating tion of Table 3, the internal consistency coefficients
to Peer Attachment were due to the adolescent did not differ markedly across the sub-groups,
sample scoring significantly higher than the child with the exception of the alpha coefficients for PBI
sample on Trust, Communication, and overall Peer Care. For this variable, there was a marked differ-
Attachment but lower on the peer Alienation ence, with a more acceptable alpha being yielded
sub-scale. for the older age group of participants and for
With regard to the significant sex differences, females. On the whole, the coefficients demon-
males scored higher than females on parent Trust strated good internal consistency for all of the
and Communication as well as on overall IPPA-R variables, with the exception of those for the
parent attachment but lower on parent Alienation. Alienation sub-scale of the Peer Attachment scale,
In contrast, females scored higher on two (i.e. Trust which were somewhat weaker, but nevertheless
and Communication) of the IPPA-R peer attach- acceptable.
ment sub-scales but lower on the Alienation sub-
scale. Females also scored higher than males on
overall Peer Attachment. It is important to
note, however, that the female sample was over- Pearson’s Correlation Coefficients Between
represented by older participants while the male the IPPA-R and the Other Measures
sample was over-represented by younger partici-
Convergent validity coefficients between the IPPA-
pants. This may explain why the trends for males
R and self-esteem scores, as well as with the PBI
and females are generally the same as those for the
dimensions of Care and Overprotection, are shown
two age-groups.
in Table 4. What is clearly evident is that the over-
Descriptive statistics and t-test comparisons for
all Parent Attachment score of the IPPA-R was
the PBI revealed no significant age-group or sex
strongly positively correlated with the Care dimen-
differences in ratings for parental nurturance or
sion of the PBI and moderately negatively corre-
overprotection. The means and standard devia-
lated with the Overprotection dimension of the
tions for the two dimensions are shown in Table 2.
PBI. With the exception of the Parent Communica-
tion subscale of the IPPA-R, the sub-scales of the
Parent Attachment scale were moderately corre-
Cronbach’s Alpha Internal Consistency
lated with the PBI dimensions. However, the cor-
Internal consistency analyses were carried out for relations were generally smaller than those with
each of the measures. Alpha coefficients were the total Parent Attachment score. Peer Attachment

Table 3. Cronbach’s alpha coefficients for the overall SEI, PBI Care and Overprotection dimensions, and the IPPA-
R scales as well as its sub-scales, by age group and sex

Measure Children Adolescents Males Females Overall

Self-Esteem 0.90 0.87 0.89 0.87 0.88
Parent Attachment
Trust 0.83 0.85 0.77 0.77 0.78
Communication 0.79 0.79 0.79 0.83 0.82
Alienation 0.76 0.81 0.60 0.69 0.79
Peer Attachment
Trust 0.80 0.86 0.82 0.88 0.86
Communication 0.84 0.86 0.80 0.86 0.87
Alienation 0.66 0.68 0.63 0.65 0.69
Parental Care 0.65 0.85 0.52 0.88 0.80
Parental Overprotection 0.80 0.86 0.74 0.86 0.84

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
74 E. Gullone and K. Robinson

Table 4. Pearson’s product-moment correlation coefficients between the IPPA-R scores with self-esteem and the PBI
by age group and sex

Measure Self-esteem score PBI (overall sample)

Children Adolescents Males Females Care Overprotection
Parent Attachment 0.51*** 0.65*** 0.49*** 0.62*** 0.73*** -0.51**
Trust 0.41*** 0.60*** 0.41*** 0.57*** 0.45** -0.40*
Communication 0.28** 0.51*** 0.20* 0.49*** 0.45** -0.05
Alienation -0.60*** -0.65*** -0.63*** -0.64*** -0.58*** 0.57***
Peer Attachment 0.50*** 0.33*** 0.45*** 0.47*** 0.36* -0.24
Trust 0.49*** 0.33*** 0.42*** 0.44*** 0.44** -0.25
Communication 0.31** 0.27*** 0.22* 0.43*** 0.42** -0.29
Alienation -0.59*** -0.35*** -0.59*** -0.38*** -0.33* 0.15
Significance levels are for one-tailed correlations.
*p < 0.05, **p < 0.01, ***p < 0.001.

Table 5. Pearson’s product-moment inter-correlation coefficients between the IPPA-R scale and subscale scores by
age group

Measure Parent Attachment Peer Attachment

Trust Communication Alienation Trust Communication Alienation
Parent Attachment (overall) 0.87*** 0.85*** -0.87***
0.90*** 0.90*** -0.90***
Trust 0.67*** -0.64*** 0.30** 0.28** -0.15
0.74*** -0.74*** 0.24** 0.21** -0.18*
Communication -0.53*** 0.26** 0.30** -0.16
-0.68*** 0.20** 0.24** -0.20**
Alienation -0.31** -0.20* -0.33***
-0.13* -0.06 0.29***
Peer Attachment (overall) 0.92*** 0.91*** -0.78***
0.93*** 0.93*** -0.81***
Trust 0.79*** -0.62***
0.82*** -0.66***
Communication -0.53***
-0.61***
Significance levels are for one-tailed correlations.
*p < 0.05, **p < 0.01, ***p < 0.001.
Coefficients for the adolescent sample are given in italics and bold font.

and its sub-scales were also found to be moderately coefficients were positive for all IPPA-R scales with
correlated with the PBI Care dimension. In con- the exception of Alienation.
trast, the Overprotection dimension was not found
to correlate significantly with the IPPA-R peer
attachment measure.
Inter-Correlations Between the IPPA-R Scales
With regard to self-esteem, on the whole, the cor-
and Subscales by Age Group
relation coefficients were found to be significant
and moderately sized. Whilst for the child group Table 5 shows the inter-correlation coefficients
there was no apparent difference in the strength of between the Parent and Peer Attachment total and
the correlations across Parent and Peer Attach- sub-scale scores, by age group. Not surprisingly,
ment, for the adolescent group the coefficients the within scale (i.e. Parent Attachment, Peer
were somewhat stronger between self-esteem and Attachment) correlations were found to be consis-
Parent Attachment compared with Peer Attach- tently higher than those across scales (i.e. Parent
ment. No marked sex differences were apparent in Attachment scores with Peer Attachment scores).
the strength of the correlations. As expected, the This was true for each of the child and adolescent

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
IPPA-R for Children 75

Table 6. Pearson’s product-moment inter-correlation coefficients between the IPPA-R scale and subscale scores by
sex

Measure Parent Attachment Peer Attachment

Trust Communication Alienation Trust Communication Alienation
Parent Attachment (overall) 0.86*** 0.82*** -0.79***
0.90*** 0.92*** -0.92***
Trust 0.61*** -0.59*** 0.27** 0.25** -0.17 ns
0.78*** -0.74*** 0.30*** 0.31*** -0.19*
Communication -0.35*** 0.23* 0.30** -0.09 ns
-0.74*** 0.21** 0.26** -0.24**
Alienation -0.29** -0.16 ns 0.39***
-0.24** -0.23** 0.33***
Peer Attachment (overall) 0.92*** 0.89*** -0.78***
0.93*** 0.93*** -0.81***
Trust 0.75*** -0.64***
0.83*** -0.63***
Communication -0.51***
-0.60***
Significance levels are for one-tailed correlations.
*p < 0.05, **p < 0.01, ***p < 0.001.
Coefficients for the female sample are given in italics and bold font.

age groups. Also, each of the Parent and Peer of age appropriate measures (Green & Goldwyn,
overall attachment scores correlated highly with 2002).
their respective sub-scale scores. Correlations On the basis of several criteria, including its
between sub-scales, within each of the Parent and assessment of key aspects of the attachment rela-
Peer attachment scales, were also consistently tionship (i.e. trust, communication and alienation)
moderately high. Finally, there were no marked and its demonstrated psychometric soundness, we
differences in the patterns of association between considered the IPPA an appropriate measure to
the two age groups, although the correlations adapt for use with children. To that end, we sim-
between scales tended to be stronger for the plified the wording of the IPPA items. Specifically,
younger group and those within scales tended to 16 of the 28 parent attachment items were revised,
be somewhat higher for the older group. as were 14 of the 25 peer attachment items.
Table 6 shows the correlations between the same Investigation of the revised measure’s psycho-
variable pairs as Table 5, this time by sex. As with metric properties included an initial set of analyses
the coefficients in Table 5, the within scale correla- aimed at examining whether age and sex differ-
tions were found to be consistently higher than ences on the IPPA-R scale and subscale scores
those across scales. This was true for each of the would be found. These analyses yielded significant
male and female sub-groups. Again, the overall differences, indicating that the IPPA-R is sensitive
attachment scores for each of Parent and Peer to age and sex differences. In relation to age dif-
Attachment correlated strongly with their respec- ferences, the early adolescent group scored signi-
tive sub-scale scores. As with the age-groups, there ficantly higher than the child group on overall Peer
were no marked differences in the patterns of asso- Attachment as well as Trust and Communication.
ciation between males and females. In contrast, adolescents scored lower on the peer
Alienation subscale compared to the child group.
When examining Parent Attachment, on the whole,
DISCUSSION the child group scored significantly higher than the
Despite its central place in developmental theories adolescent group on all scores, with the exception
(Rutter, 1995), the incorporation of the attachment of the Alienation subscale, for which there were
construct into research involving participants no age-group differences. It is important to note,
beyond infancy and prior to late adolescence however, that this should not be interpreted as an
has been difficult given the limited availability indication that attachments to parents are less

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
76 E. Gullone and K. Robinson

important for adolescents’ psychological wellbeing that post-adolescence, youths’ perceptions of their
compared to that of children (cf. Gecas, 1972; relationships with their parents remain more
O’Donnell, 1976) as is indicated by the correlations important than those with peers (see, e.g., Gecas,
found between IPPA-R scale and subscale scores 1972; O’Donnell, 1976).
and self-esteem (as will be discussed in more detail As was predicted, the overall Parent Attachment
below). score of the IPPA-R was strongly positively
Sex differences on the IPPA-R scale and subscale correlated with the Care dimension of the PBI
scores were the result of males generally scoring and moderately negatively correlated with the
higher than females on Parent Attachment but Overprotection dimension of the PBI. For the most
females scoring higher than males on peer attach- part, the sub-scales of the Parent Attachment scale
ment. Given that the sex differences on the were also moderately correlated with the PBI
Alienation subscale for both parent and peer were dimensions. Peer Attachment and its sub-scales
in the opposite direction, the findings indicated were also found to be significantly correlated with
that males reported more positive attachments the PBI Care dimension. In contrast, the Overpro-
with their parents than did females. In contrast, tection dimension was not found to correlate
females reported more positive attachments with significantly with the IPPA-R peer attachment
their peers compared with males. These differences measure. The different pattern of association
are intriguing and warrant further investigation to between the PBI with IPPA-R Parent versus
determine whether they are a reflection of real Peer Attachment is not surprising since parental
differences in attachment relationships or whether overprotection would not be expected to
they are related to response style on the self-report significantly spill over into peer relationship
measure. quality. In contrast, care, otherwise referred to as
Analyses relating to the reliability of the IPPA-R warmth and nurturance, has been identified as a
demonstrated that the internal consistency coeffi- central factor in the development of intimate
cients did not differ markedly by age or sex group. relationships across the lifespan (cf. Griffin &
The coefficients ranged between 0.60 (on Parent Bartholomew, 1994; MacDonald, 1992). It is also of
Alienation for males) and 0.88 (on Peer Trust for interest that, in contrast to the findings relating to
females) and thus indicated adequate to good self-esteem, the correlation analyses between the
internal consistency for each of the IPPA-R sub- PBI and the IPPA-R did not reveal the Communi-
scales across the sub-samples investigated. The cation subscale to be weakly correlated with the
coefficients yielded were highly comparable to Care dimension.
those reported by Armsden and Greenberg (1987) Finally, we examined the interrelationships
for the IPPA, which ranged between 0.72 and 0.91 between the IPPA-R scales and subscales. Our find-
for the sub-scales across both the parent and peer ings were highly consistent with those reported by
scales. Armsden and Greenberg (1987). Specifically, we
Consistent with the work by Armsden and found that each of the total IPPA-R Parent and Peer
Greenberg with the IPPA, we examined convergent Attachment scores correlated strongly with their
validity by correlating reports on the IPPA-R with respective sub-scale scores. Correlations between
reports on self-esteem. In general, moderate corre- sub-scales, within scales, were also consistently
lations in the predicted direction were found for all moderately strong. Comparison of the correlation
scores on the IPPA-R and therefore provided coefficients across age and sex sub-groups did not
support for the validity of the revised measure. The reveal any major differences, suggesting that the
reason is not clear but correlations were weakest IPPA-R assessed the constructs in a comparable
on the Communication subscale scores across both way across sub-groups. These findings therefore
Parent and Peer Attachment and particularly for lend further support to the validity of the revised
the male and child sub-samples. It is also impor- measure.
tant to note that, whilst the differences were not Notwithstanding these promising findings, the
large, there was some tendency for the correlations limitations of the study need to be acknowledged.
between self-esteem and Parent Attachment to be First, our sample was limited by an unequal rep-
stronger for the adolescent sub-sample compared resentation of males and females in the two age
with the child sub-sample. This is consistent with groups of participants, with the child group being
outcomes reported in studies comparing peer and over-represented by males and the adolescent
parent relationships in relation to psychological group being over-represented by females. Thus,
adjustment. Such studies have primarily shown the analyses involving the age and sex sub-samples

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
IPPA-R for Children 77

warrant replication in order to determine their 15. My parents have their own problems, so I don’t
validity. A second limitation relates to the small bother them with mine.
sample involved in the convergent validity analy- 16. My parents help me to understand myself
ses using the PBI. Although the outcomes were as better.
predicted, these findings can only be considered 17. I tell my parents about my problems and
tentative and need to be replicated. troubles.
In conclusion, our findings provide strong 18. I feel angry with my parents.
support for the reliability and validity of the 19. I don’t get much attention at home.
revised IPPA. We found adequate to good internal 20. My parents support me to talk about my
consistency for the IPPA-R Parent and Peer Attach- worries.
ment scales as well as for each of the subscales. The 21. My parents understand me.
adequately sized correlations between the IPPA-R 22. I don’t know who I can depend on.
total and subscale scores and another measure of 23. When I am angry about something, my parents
parent bonding (i.e. PBI: Parker et al., 1979) indi- try to understand.
cate that the IPPA-R is a valid measure of attach- 24. I trust my parents.
ment in children and young adolescents. The 25. My parents don’t understand my problems.
equally robust correlations found with self-esteem 26. I can count on my parents when I need to talk
are consistent with past research using the IPPA about a problem.
with adolescents and young adults (Armsden & 27. No one understands me.
Greenberg, 1987). This consistency provides good 28. If my parents know that I am upset about
indication that the revised IPPA is a sound tool for something, they ask me about it.
the assessment of attachment in children and
adolescents aged between 9 and 15 years. Given
the limited attachment measures for children and
Peer Scale Items
younger adolescents, the present study makes an
important contribution to the developmental 1. I like to get my friends’ opinions on things I’m
literature. worried about.
2. My friends can tell when I’m upset about
something.
3. When we talk, my friends listen to my opinion.
APPENDIX—THE IPPA-R 4. I feel silly or ashamed when I talk about my
problems with my friends.
Parent Scale Items
5. I wish I had different friends.
1. My parents respect my feelings. 6. My friends understand me.
2. My parents are good parents. 7. My friends support me to talk about my
3. I wish I had different parents. worries.
4. My parents accept me as I am. 8. My friends accept me as I am.
5. I can’t depend on my parents to help me solve 9. I feel the need to be around my friends more
a problem. often.
6. I like to get my parents’ view on things I’m 10. My friends don’t understand my problems.
worried about. 11. I do not feel like I belong when I am with my
7. It does not help to show my feelings when I am friends.
upset. 12. My friends listen to what I have to say.
8. My parents can tell when I’m upset about 13. My friends are good friends.
something. 14. My friends are fairly easy to talk to.
9. I feel silly or ashamed when I talk about my 15. When I am angry about something, my friends
problems with my parents. try to understand.
10. My parents expect too much from me. 16. My friends help me to understand myself
11. I easily get upset at home. better.
12. I get upset a lot more than my parents know 17. My friends care about the way I feel.
about. 18. I feel angry with my friends.
13. When I talk about things with my parents they 19. I can count on my friends to listen when some-
listen to what I think. thing is bothering me.
14. My parents listen to my opinions. 20. I trust my friends.

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
78 E. Gullone and K. Robinson

21. My friends respect my feelings. and peers during adolescence. Journal of Youth and
22. I get upset a lot more than my friends know Adolescence, 12, 373–386.
about. Griffin, D., & Bartholomew, K. (1994). Models of the self
and other: Fundamental dimensions underlying
23. My friends get annoyed with me for no reason. measures of adult attachment. Journal of Personality and
24. I tell my friends about my problems and Social Psychology, 67, 430–445.
troubles. Gullone, E., King, N.J., & Ollendick, T.H. (June, 2002).
25. If my friends know that I am upset about some- The role of attachment style in the development of
thing, they ask me about it. childhood depression. Paper presented at the Third
International Conference on Child and Adolescent Mental
Health, Brisbane.
Hamilton, C.E. (2002). Continuity and discontinuity of
REFERENCES attachment from infancy through adolescence. Child
Development, 71, 690–694.
Ainsworth, M.D.S., Blehar, M.C., Waters, E., & Wall, S. Herz, L., & Gullone, E. (1999). The relationship between
(1978). Patterns of attachment: A psychological study of the self-esteem and parenting style: A cross-cultural
strange situation. Hillsdale, NJ: Erlbaum. comparison of Australian and Vietnamese-Australian
Armsden, G.C., & Greenberg, M.T. (1987). The inventory adolescents. Journal of Cross-Cultural Psychology, 30,
of parent and peer attachment: Individual differences 742–761.
and their relationship to psychological well-being Hollman, C.M., & McNamara, J.R. (1999). Considerations
in adolescence. Journal of Youth and Adolescence, 16, in the use of active and passive parental consent
427–454. procedures. The Journal of Psychology, 133, 141–156.
Berlin, L., & Cassidy, J. (1999). Relations among relation- Klimidis, S., Minas, I.H., & Ata, A.W. (1992). The PBI-BC:
ships: Contributions from attachment theory and A brief current form of the Parental Bonding Instru-
research. In J. Cassidy, & P.R. Shaver (Eds), Handbook ment for adolescent research. Comprehensive Psychiatry,
of attachment: Theory, research, and clinical applications 33, 374–377.
(pp. 688–712). New York: Guilford. Klimidis, S., Minas, I.H., Ata, A.W., & Stuart, G.W. (1992).
Bowlby, J. (1969). Attachment and loss: Vol. 1. Attachment. Construct validation in adolescents of the brief current
New York: Basic. form of the Parental Bonding Instrument. Comprehen-
Bretherton, I. (1985). Attachment theory: Retrospect and sive Psychiatry, 33, 378–383.
prospect. Monographs of the Society for Research in Child Lewis, M., Feiring, C., & Rosenthal, S. (2000). Attachment
Development, 50, 3–35. over time. Child Development, 71, 707–720.
Carlson, E.A., & Sroufe, L.A. (1995). Contribution of MacDonald, K. (1992). Warmth as a developmental con-
attachment theory to developmental psychopathology. struct: An evolutionary analysis. Child Development, 63,
In D. Cicchetti, & D.J. Cohen (Eds), Developmental 753–773.
psychopathology Volume 1: Theory and methods (pp. O’Connell-Higgins, G. (1994). Resilient adults: Overcoming
581–616). New York: Wiley. a cruel past. San Francisco: Jossey-Bass.
Cicchetti, D., & Garmezy, N. (1993). Prospects and O’Donnell, W.J. (1976). Adolescent self-esteem related
promises in the study of resilience. Development and feelings toward parents and friends. Journal of Youth
Psychopathology, 5, 497–502. and Adolescence, 5, 179–185.
Coopersmith, S. (1981). Self-esteem inventories. Palo Alto Parker, G. (1979a). Reported parental characteristics in
CA: Consulting Psychologists Press. relation to depressive disorders. British Journal of
Cubis, J., Lewis, T., & Dawes, F. (1989). Australian ado- Psychiatry, 134, 138–147.
lescents’ perceptions of their parents. Australian and Parker, G. (1979b). Reported parental characteristics
New Zealand Journal of Psychiatry, 23, 35–47. of agoraphobics and social phobics. British Journal of
Fitts, W.H. (1965). Tennessee Self-Concept Scale Manual. Psychiatry, 135, 555–560.
Nashville: Counselor Recordings and Tests. Parker, G. (1983). Parental ‘affectionless control’ as
Fraley, R.C. (2002). Attachment stability from infancy to an antecedent to adult depression: A risk factor
adulthood: Meta-analysis and dynamic modeling of delineated. Archives of General Psychiatry, 40, 956–
developmental mechanisms. Personality and Social 960.
Psychology Review, 6, 123–151. Parker, G., Tupling, H., & Brown, L.B. (1979). A parental
Gecas, V. (1971). Parental behavior and dimensions of bonding instrument. British Journal of Medical Psychol-
adolescent self-evaluation. Sociometry, 34, 466–482. ogy, 52, 1–10.
Gecas, V. (1972). Parental behavior and contextual varia- Rice, K.G., & Cummins, P.N. (1996). Late adolescent and
tions in adolescent self-esteem. Sociometry, 35, 332–345. parent perceptions of attachment: An exploratory
Green, J., & Goldwyn, R. (2002). Annotation: Attachment study of personal and social well-being. Journal of
disorganisation and psychopathology: New findings Counseling and Development, 75, 50–57.
in attachment research and their potential implications Rutter, M. (1995). Clinical implications of attachment
for developmental psychopathology in childhood. concepts: Retrospect and prospect. Journal of Child Psy-
Journal of Child Psychology and Psychiatry, 43, 835–846. chology and Psychiatry, 4, 549–571.
Greenberg, M., Siegal, J., & Leitch, C. (1983). The nature Silove, D. (1986). Perceived parental characteristics and
and importance of attachment relationships to parents reports of early parental deprivation in agoraphobic

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)
IPPA-R for Children 79

patients. Australian and New Zealand Journal of Psychia- West, M., Rose, S., Spreng, S., Sheldon-Keller, A., &
try, 20, 365–369. Adam, K. (1998). Adolescent attachment question-
Walker, L.S., & Green, J.W. (1986). The social context of naire: A brief assessment of attachment. Journal of Youth
adolescent self-esteem. Journal of Youth and Adolescence, and Adolescence, 27, 661–673.
15, 315–322.

Copyright © 2005 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 12, 67–79 (2005)