Original article
Validation of a brief measure of adolescents’ parent attachment based on Armsden and Greenberg’s three-dimension model

https://doi.org/10.1016/j.erap.2004.04.003Get rights and content

Abstract

The three studies that are reported examined the validity of a three-dimension model of adolescents’ attachment to parents based on Armsden and Greenberg’s IPPA [J. Youth Adolesc. 5 (1987) 427]. In Study 1, 289 14–15-year-old participants assessed their parent attachments using a French version of the original scale. Using factor analyses, we selected those items that tap one of the three expected dimensions (communication, trust, and alienation) especially well for each parent relationship. This shortened scale was tested in Study 2 with 206 14–15-year-old participants. Confirmatory factor analyses revealed that the three-dimension model provides acceptable fit, in contrast to one- and two-dimension models. In Study 3, 283 18–19-year-old participants responded to this short form and to a parenting styles questionnaire. Based on confirmatory factor analyses and convergent validity indices, we conclude that this questionnaire is a reliable basis on which to develop a more complex multidimensional inventory.

Résumé

Dans les trois études dont il est rendu compte, on a examiné la validité d’un modèle d’attachement des adolescents à leurs parents en trois dimensions fondé sur l’IPPA d’Armsden et Greenberg [J. Youth Adolesc. 5 (1987) 427]. Dans la première étude, 289 participants âgés de 14–15 ans ont évalué les attachements à leurs parents en répondant à une version française de l’échelle originale. A l’aide d’analyses factorielles nous avons sélectionné pour chaque dimension (communication, confiance et aliénation) les items qui portent plus spécialement sur elle, à la fois pour l’attachement à la mère et l’attachement au père. Cette échelle courte a été testée dans la deuxième étude auprès de 206 participants de 14–15 ans. Une analyse factorielle confirmatoire indique que le modèle en trois dimensions s’ajuste de manière acceptable aux données, à la différence des modèles uni- et bidimensionnels. Dans la troisième expérience, 283 participants de 18–19 ans ont répondu à cette même échelle ainsi qu’à un questionnaire évaluant les styles éducatifs parentaux. Compte tenu des analyses factorielles confirmatoires et des indices de validité convergente, nous concluons que cet instrument constitue une base solide pour la construction d’un questionnaire multidimensionnel plus complexe.

Introduction

In accordance with Bowlby’s (1969) view of attachment as a lifespan need, numerous empirical studies have supported the positive role of secure attachment to parents for adolescent development and adjustment (for an overview, see Allen and Land, 1999). During a period characterized by various broad and normative changes, secure attachment to parents provides a source of support for exploration and for mastering developmental tasks. The role of parent–adolescent attachment relationships is especially important during developmental transitions, such as school transitions (e.g., Papini and Roggman, 1992). Longitudinal research has demonstrated that the quality of parent attachment predicts the evolution of socioemotional adjustment during such transition (e.g., Larose and Boivin, 1998). The aim of the present study was to generalize the results of Armsden and Greenberg’s (1987) three-dimension model of adolescents’ attachment to parents to French middle-to-late adolescents, by means of exploratory factor analyses. We also attempted to validate this model by means of confirmatory analyses, and we compared its fit with those of one- and two-dimension models.

One of Bowlby’s (1969) central assertions is that attachment behavioural systems arise from characteristic patterns of interactions with significant others. As stated by Sroufe and Waters (1977), a consequence of this assertion is that “establishing and describing individual differences in the way attachment behaviours are organized is crucial, both for a complete understanding of normative patterns and in order to examine the origins and consequences of individual differences” (p. 1190). Several instruments have been devised to assess adolescents’ parent attachment (for a review, see Lopez and Gover, 1993 or Crowell et al., 1999). One of the first and most frequently used is Armsden and Greenberg’s (1987) questionnaire. It was designed for the general population of adolescents rather than specifically for those with disturbed relations to their parent.

Greenberg et al. (1984) created the first version of this self-report inventory, to assess adolescents’ attachment to parents and peers. The authors empirically supported their hypothesis that, for attachment to parents as well as for attachment to peers, “there were two separate dimensions of attachment relationships: (1) felt security, operationalized as the quality of affect toward such figures, and (2) proximity seeking, the degree to which the youth seek out such figures in times of stress and need” (p. 376). Regarding attachment to parents, the subscale assessing the quality of affect was only moderately (r = 0.24, P = 0.001) correlated to the subscale assessing the degree to which the adolescent would talk to parents or siblings in five hypothetical situations eliciting strong emotional states. It must be noted that the first dimension was assessed with items tapping attachment to both parents together, and the second dimension was assessed with items tapping attachment to the whole family, a fact that increased the likelihood of obtaining two highly independent dimensions.

In order to improve the instrument, and according to the multidimensional character of attachment, Armsden and Greenberg (1987) decided to isolate the various qualities within the affective dimension. Taking two of the four items that assessed the felt security of the relation in the first questionnaire, they then added 26 more items. These were designed to assess “the adolescent’s trust (felt security) that attachment figures understand and respect her/his needs and desires, and perceptions that they are sensitive and responsive to her/his emotional states and helpful with concerns” (p. 432). Other items were designed to tap the negative experiences of anger, hopelessness or detachment related to perceived unresponsive or inconsistently responsive partners. An exploratory factor analysis revealed a meaningful three-dimension model: communication, trust, and alienation. Communication may be viewed as the adolescent analogy of the infant proximity seeking dimension, trust is related to felt security, and alienation signifies an emotional distance from an unempathic partner.

This inventory of parent and peer attachment (IPPA) was frequently used to assess parent attachment alone (rather than peer attachment alone or both relations). Peer relationships are generally studied with instruments designed to assess not only attachment features but also companionship or other relevant features more or less specific to adolescent peer relationships (e.g., the Friendship Qualities Scale of Bukowski et al., 1994). Armsden and Greenberg (1987) did not present the factor analysis results of their scale separately for attachment to mother and attachment to father. Participants in their study were requested to assess the relation to both parents or the relation with the parent who had most influenced them, if they felt they had very different relationships with mother and father. Given the frequently reported within-person variation in attachment security across relationships (e.g., LaGuardia et al., 2000), and following the recommendation provided subsequently by Armsden and Greenberg (1989), the parent attachment scale was generally used in subsequent studies to assess separately attachment to mother and attachment to father separately.

As far as we know, all the authors who used the parent attachment scale of the IPPA analysed only the total scale score (obtained by summing trust and communication scores and subtracting from this sum the alienation score). This is in accordance with Armsden and Greenberg’s (1989) suggestion. Although in their 1987 article, they reported that the three dimensions were highly correlated (range –0.76 to 0.76; N = 179), we reasoned that this three-dimension model provides one of the few multidimensional models of adolescent’s parent attachment. The attachment behavioural system, especially for adolescents, cannot be accurately assessed by a single dimension. Therefore, it was of interest to generalize the three-dimension model, which has never been replicated, to adolescents of another country. This was the aim of the first study, which consisted in an exploratory analysis of middle adolescents’ perception of attachment to mother and father, separately, using the original 28 IPPA items.

The second study was intended to confirm the three-dimension model, which was empirically supported by Armsden and Greenberg (1987), and which we obtained in the first study by selecting the most relevant items. It was also intended to compare the fit of this expected model to those of one-dimension and two-dimension models. Most of the studies implicitly relied on the one-dimension model, since they used only the total scale score. The two-dimension model, which was hypothesized by Armsden and Greenberg (1987), could be expected because the two most salient categories that seem to characterize the items are the positive (trust and communication) and negative (alienation) attachment experiences. The third study was carried out to: (a) replicate Study 2 confirmatory factor analysis with a sample of late adolescents; and (b) provide a first step of convergent analysis of the proposed brief measure of adolescents’ attachments to parents.

Section snippets

Participants

The respondents were 289 middle-to-high socio-economic status middle adolescents (47% boys) living in the eastern part of France. They attended two similar high schools and were recruited through a form letter which was sent out to the 303 potential participants. Among these students, 292 returned a parental consent form, though three of them were absent when the data were collected. Their average age was 14.64 years (S.D. = 0.76 years).

Instrument

The 28 items devised by Armsden and Greenberg (1987) to

Participants

Participants were 206 middle adolescents (45% boys) of the same age as those in the first study (mean age = 14.71 years, S.D. = 0.66 years). They were from middle-to-high socio-economic backgrounds and were recruited in a high school in the south of France, following the same process as in the first study. All participants provided a parental consent form. Four adolescents did not complete the scale for father, and one did not for mother.

Procedure and measures

Participants responded to the 14-item version of the

Study 3

In addition to the analysis of internal consistency and latent structure, we assessed adolescents’ perception of parenting styles, as a first approach to the convergent validity of the parent attachment short form. Four parenting styles, originally defined by Baumrind (1971), were considered: authoritative, authoritarian, permissive, and neglectful (for a review on parenting styles, see Collins et al., 2000).

Authoritarian parents impose many rules, expect obedience and rely on punitive,

General discussion

Overall the three studies support the validity of Armsden and Greenberg’s (1987) construct of adolescents’ attachment to parent, which they proposed in order to study the “working models” of adolescents in the framework of Bowlby’s (1969) attachment theory. In three independent samples of French middle-to-late adolescents, we observed consistent individual differences in the way they perceive the emotional support provided by each of their parents. The questionnaire elaborated by these authors

Acknowledgements

We thank the adolescents who participated in this study, the school staffs and parents who made it possible, and the reviewers for their helpful comments on a first version of the manuscript. The study was supported by the Research Center for Cognition, Language, and Emotion of the University of Provence and the Research Center “Milieu and Development” of the University of Paris, 10 Nanterre.

References (27)

  • J.P. Allen et al.

    Attachment in adolescence

  • G.C. Armsden et al.

    The inventory of parent and peer attachment: individual differences and their relationship to psychological well-being in adolescence

    J. Youth Adolesc.

    (1987)
  • G.C. Armsden et al.

    The inventory of parent and peer attachment (IPPA) (unpublished technical report)

    (1989)
  • F. Bacher

    L’utilisation des modèles dans l’analyse des structures de covariance

    Annee Psychol.

    (1999)
  • D. Baumrind

    Current patterns of parental authority

    Developmental Psychology Monographs 4(1, part 2)

    (1971)
  • C. Bourcet

    Evaluation de soi, climat familial et adaptation scolaire à l’adolescence (Self evaluation, family climate, and school adaptation in adolescence)

    (1994)
  • J. Bowlby

    Attachment and Loss. Attachment, vol. 1.

    (1969)
  • M.W. Browne et al.

    Alternative ways of assessing model fit

  • W.M. Bukowski et al.

    Measuring friendship quality during pre- and early adolescence: the development and psychometric properties of the Friendship Quality Scale

    J. Soc. Pers. Relat.

    (1994)
  • W.A. Collins et al.

    Contemporary research on parenting. The case of nature and nurture

    Am. Psychol.

    (2000)
  • J.A. Crowell et al.

    Measurement of individual differences in adolescent and adult attachment

  • M.T. Greenberg et al.

    The nature and importance of attachment relationships to parents and peers during adolescence

    J. Youth Adolesc.

    (1984)
  • E.V.E. Hodges et al.

    Skewed autonomy-related in preadolescent's conceptions of their relationships with mother, father, and best friend

    Developmental Psychology

    (1999)
  • Cited by (44)

    • Parents vs peers’ influence on teenagers’ Internet addiction and risky online activities

      2018, Telematics and Informatics
      Citation Excerpt :

      It was derived from factor analytic studies that explicitly distinguish addictions from (non-pathological) high engagement. To reduce the length of the survey, the scale used to measure parent-attachment construct was adapted from Vignoli and Mallet’s (2004), which is a shortened version of Armsden and Greenberg’s (1987) scale, consisting of 14 items. Peer-attachment was measured using Armsden and Greenberg’s (1987) scale of 17 questions.

    View all citing articles on Scopus
    View full text