You are on page 1of 8

Revue européenne de psychologie appliquée 66 (2016) 109–116

Disponible en ligne sur

ScienceDirect
www.sciencedirect.com

Original article

Personal assessment of intimacy in relationships: Validity and


measurement invariance across gender
Évaluation personnelle de l’intimité au sein des relations : validité et mesure de
l’invariance selon le genre
E. Constant ∗ , F. Vallet , J.-L. Nandrino , V. Christophe ∗
Lille University, UMR CNRS 9193 SCALab “Cognitive and Affective Sciences”, rue du Barreau, BP 60149, 59653 Villeneuve-d’Ascq cedex, France

a r t i c l e i n f o a b s t r a c t

Article history: Introduction. – Intimacy is central in close relationships and a variety of definitions and instruments exist
Received 26 May 2015 in scientific literature, making this concept complex to investigate. Furthermore, gender differences were
Received in revised form 5 February 2016 identified in the definition and experience of intimacy.
Accepted 28 April 2016
Objectives. – This study aims to confirm the three-factor structure of the personal assessment of inti-
macy in relationships (PAIR, Schaefer & Olson, 1981), commonly used in research and marital therapy,
Keywords: developed by Moore et al. (1998) and to examine the measurement equivalence of the scale across gender.
Intimacy
Method. – A confirmatory factor analysis (CFA) was performed on the responses of 313 women and 251
Couples
Confirmatory factor analysis
men in committed relationships in a French population. Measurement equivalence of the scale across
Measurement invariance gender was then conducted with the best-fitting model.
Scale validation Results. – The results of CFA and post-hoc analyses revealed that the three-factor model revised with 18
items was the best fit to the data: (1) engagement (10 items), (2) communication (5 items), (3) shared
friends (3 items). PAIR-18 displayed a lack of metric and scalar invariance across gender but estimated
means of latent variables is not very different between the full invariant model and the unconstraint
models.
Conclusion. – Nevertheless, consideration of gender differences in the construct of intimacy is very impor-
tant for theoretical research on intimacy in romantic relationships and for the interventions of marital
therapists based on intimacy.
© 2016 Published by Elsevier Masson SAS.

r é s u m é

Mots clés : Introduction. – L’intimité est centrale dans les relations proches et un grand nombre de définitions et
Intimité d’outils existent, rendant ce concept complexe à étudier. De plus, des différences de genre ont été
Couples identifiées concernant la définition et l’expérience de l’intimité.
Analyse factorielle confirmatoire Objectifs. – Cette étude vise à confirmer la structure en 3 facteurs développée par Moore et al. (1998) de
Mesure d’invariance
l’échelle originale « évaluation personnelle de l’intimité au sein des relations » (PAIR, Schaefer & Olson,
Genre
1981), couramment utilisée en recherche et thérapie conjugale et d’examiner l’équivalence de l’échelle
Validation d’échelle
selon le genre.
Méthode. – Une analyse factorielle confirmatoire (AFC) a été réalisée auprès de 313 femmes et
251 hommes engagés dans une relation sérieuse au sein de la population française. La mesure d’invariance
selon le genre a ensuite été réalisée sur le modèle le mieux ajusté.
Résultats. – Les résultats de l’AFC et des analyses post-hoc ont révélé qu’un modèle en 3 facteurs révisé
en 18 items est le mieux ajusté aux données : (1) engagement (10 items), (2) communication (5 items),
(3) amis communs (3 items). La PAIR-18 présente un manque d’invariance métrique et scalaire selon le
genre mais les moyennes estimées des variables latentes ne sont pas très différentes entre le modèle
totalement contraint à invariance et le modèle non contraint.

∗ Corresponding authors.
E-mail addresses: emilie.constant@univ-lille3.fr (E. Constant), veronique.christophe@univ-lille3.fr (V. Christophe).

http://dx.doi.org/10.1016/j.erap.2016.04.008
1162-9088/© 2016 Published by Elsevier Masson SAS.
110 E. Constant et al. / Revue européenne de psychologie appliquée 66 (2016) 109–116

Conclusion. – Néanmoins, prendre en considération les différences de genre dans le construit de l’intimité
est très important pour les recherches théoriques explorant l’intimité au sein des relations amoureuses
ainsi que pour les interventions des thérapeutes conjugaux basées sur l’intimité.
© 2016 Publié par Elsevier Masson SAS.

1. Introduction Among the different available tools, there are three popu-
lar scales used by researchers and couple therapists in this field
Intimacy is at the core of loving relationships (Sternberg & (Hook et al., 2003): the Miller Social Intimacy Scale (MSIS; Miller
Grajek, 1984) and many researchers have attempted to conceptual- & Lefcourt, 1982), the Fear of Intimacy Scale (FIS; Descutner &
ize it. According to Sternberg’s triangular theory of love, intimacy is Thelen, 1991) and the Personal Assessment of Intimacy in Rela-
the emotional component of love and refers particularly to feelings tionships (PAIR, Schaefer & Olson, 1981). The MSIS is a 17-item
of closeness, bondedness, and connectedness (Sternberg, 1986). questionnaire, which assesses the frequency of intimacy and the
Intimate relationships are important for human functioning and intensity of the intimacy experienced in a given relationship.
well-being (Baumeister & Leary, 1995; McAdams & Vaillant, 1982; The FIS (Descutner & Thelen, 1991) concerns items about anxiety
Reis, Collins, & Berscheid, 2000) and many developmental theo- to exchange thoughts and feelings of personal significance with
ries agree that intimacy is essential for individual adjustment and another individual on three dimensions. Particularly, the PAIR is
needs (Erikson, 1950; Maslow, 1954; Sullivan, 1953). The concept commonly used to assess relationship intimacy in research and
of intimacy has often been investigated in the scientific litera- marital therapy (Elliott, Bingham, Nielsen, & Warner, 1986; Harper
ture where a variety of definitions and operationalizations can & Elliott, 1988; Heller & Wood, 1998; Nezhad & Goodarzi, 2011;
be found (see Heller & Wood, 1998; Schaefer & Olson, 1981). Reiter, Richmond, Stirlen, & Kompel, 2009; Worthington, Buston,
Moss and Schwebel (1993) proposed that “intimacy in enduring & Hammonds, 1989) and to examine differences between partners
romantic relationships is determined by the level of commit- (Schaefer & Olson, 1981). The original instrument was developed
ment and positive affective, cognitive, and physical closeness one in 36 items to investigate the experience of intimacy and contained
experiences with a partner in a reciprocal (although not neces- six subscales: emotional, social, sexual, intellectual, recreational
sarily symmetrical) relationship”. Besides, Reis and Shaver (1988) and conventionality. This latter dimension is only used to “indicate
defined intimacy as an interpersonal and transactional process with how much individuals attempt to ‘fake good’.” (Schaefer & Olson,
two principal components: self-disclosure and partner responsive- 1981). However, serious concerns were pointed involving furthers
ness. Indeed, intimacy is initiated when a person share relevant research in order to demonstrate the reliability of the scale (Hook
information, thoughts, and feelings to another person (nonver- et al., 2003). Researchers attempted to investigate the factor struc-
bal or verbal disclosures). To maintain intimacy, the listener must ture of the PAIR in general (e.g., Moore, McCabe, & Stockdale, 1998)
communicate understanding and validation in response to a part- and clinical (e.g., Walker, Hampton, & Robinson, 2014) populations,
ner’s self-disclosure (Laurenceau, Rivera, Schaffer, & Pietromonaco, but none of these studies confirmed these six dimensions. Moore,
2004; Laurenceau, Barrett, & Rovine, 2005). Due to its interpersonal McCabe, and Stockdale (1998) proposed a 27-item scale with three
nature and multidimensional definition, intimacy is complex to subscales using Exploratory Factorial Analysis (EFA):
measure (Kouneski & Olson, 2004). Nevertheless, its assessment
is important for researchers as well as in couple counseling in • communication: experiencing an open and fluent exchange of
order to help each partner express their intimacy needs, under- ideas;
stand their partner’s needs (Hook, Gerstein, Detterich, & Gridley, • engagement: feeling connected with one’s partner;
2003) and improve feelings of intimacy in their romantic relation- • shared friendships: sharing common activities with friends.
ship (Schaefer & Olson, 1981).
In the same way, Walker et al. (2014) proposed a 23-item scale
1.1. Assessment of intimacy and validation studies of the personal with three subscales using also EFA:
assessment of intimacy in relationships (PAIR, Schaefer & Olson,
1981) • understanding: ability to understand one’s partner and to express
one’s emotions and beliefs freely to them. This subscale expands
Empirical studies confirmed a multidimensional conceptualiza-
beyond the conceptualized “communication” subscale of Moore
tion of intimacy (for a review see Van den Broucke, Vandereycken,
et al.;
& Vertommen, 1995) and enabled the examination of individual
• engagement: degree to which a person feels connected to their
differences through the degree and needs of intimacy.
partner;
Van den Broucke et al. (1995) examined the psychological litera-
• shared friends: degree to which a couple participates in valued
ture about marital intimacy and they identified several self-report
social activities with mutual friends.
questionnaires that could be used in order to assess intimacy in
couple relationships. The Waring Intimacy Questionnaire (WIQ;
Waring & Reddon, 1983) describes close, dyadic relations as being We focused our attention on the factorial validation developed
composed of three interpersonal dimensions: boundary, power in general population by Moore et al. (1998) and to our knowledge,
and intimacy (Gaia, 2002); the Marital Intimacy Questionnaire there are no empirical studies supporting the PAIR three-dimension
(MIQ; Van den Broucke et al., 1995) assess intimacy problems, structure by a confirmatory factorial validation.
consensus between partners, openness, affection and commit-
ment; the Intimate Relationship Scale (IRS; Hetherington & Soeken, 1.2. Gender differences in intimacy
1990) measures the perception of change in the physical, cog-
nitive, and emotional dimensions of intimacy and sexuality and Identifying similarities or differences between each partner in
finally, the Relationship Intimacy Styles Questionnaire (RISQ; Moss a romantic relationship seems to be helpful for couples in order
& Schwebel, 1993) measures the cognitive, affective, physical and to understand and accept their partner’s feelings (Heller & Wood,
commitment intimacy (Prager, 1995). 1998). Some authors support the hypothesis that the perception of
E. Constant et al. / Revue européenne de psychologie appliquée 66 (2016) 109–116 111

intimacy occurs when both partners have similar intimacy needs Olson, 1981). The scale was translated from English into French by a
(Schaefer & Olson, 1981). However, studies have reported gen- researcher and then independently back-translated by a bilingual
der differences in the definition and expression of intimacy. Thus, researcher. Discrepancies were analyzed and discussed by three
women may define intimacy as love, affection, and the expres- French researchers of the university. Participants answered items
sion of warm feelings, while men define it as sexual behavior and using a 5-point Likert scale ranging from “strongly disagree” to
physical closeness (Ridley, 1993). In addition, gender differences “strongly agree” with 18 of the items being reversed. A high score
are revealed in the experience and expression of intimacy (Greeff indicated that participants had a high level of intimacy.
& Malherbe, 2001; Patrick & Beckenbach, 2009). Women report
greater feelings of intimacy than men on several dimensions of 2.2. Data analysis
the PAIR (emotional, intellectual and recreational; Heller & Wood,
1998). Women also express more affection, love and personal vali- 2.2.1. Preliminary analysis
dation and have a better understanding of their partner’s needs in Data on 574 participants were initially conducted in order to
their relationships (Hook et al., 2003). check statistical assumption of normality. Skewness and Kurto-
Investigating how women and men define and experience inti- sis indices have been use to assess respectively the degree of
macy is an important issue in order to understand this concept asymmetry and “peakedness” in the distribution of our different
more precisely according to gender and the partner’s adjustment variables (Hopkins & Weeks, 1990). Moreover, extreme multivari-
within a stable romantic relationship. These gender differences in ate outliers were removed from the analyses in our sample using
the expression of intimacy may lead to marital conflict or problems the Squared Mahalanobis Distance test (Byrne, 2009; Kline, 2005),
during the romantic relationship, in particular to misunderstand- which resulted in a sample of 564 participants. Based on our final
ing by the partners. Furthermore, these differences regarding the sample, the analyses were performed in three steps based on the
meaning of intimacy may have repercussions on its assessment, 27-item version of the scale proposed by Moore et al. (1998).
for example in the examination of mean differences across gen-
der. Comparisons can lead to misleading interpretations in cases
2.2.2. Step 1: testing for the factorial structure of the three-factor
where measures are not invariant across gender (Borsboom, 2006).
model
A scale is considered invariant when its measurement properties
The factor structure of the PAIR proposed by Moore et al. (1998)
function in the same way across gender. Because the meaning of
was first examined using confirmatory factor analysis (CFA) and
intimacy seems to differ according to gender, the examination of
maximum likelihood (ML) estimation in Amos 21.0. We evaluated
measurement equivalence is particularly important when explor-
how well the model fit the data using multiple goodness-of-fit
ing intimacy in close relationships.
indices including the 2 /df ratio, the comparative fit index (CFI), the
Thus, the present paper aims:
root mean square error of approximation (RMSEA), and the stan-
dardized root mean square residual (SRMR). A well-fitting model
• to confirm the previous three-factor structure based on the origi-
is indicated by a cutoff value lower than or equal to 3 for the 2 /df
nal scale developed by Moore et al. (1998) in general population; ratio, equal to or higher than .95 for the CFI, close to .06 or less for
• to examine the measurement equivalence of the scale across gen-
the RMSEA, and close or less to .08 for the SRMR (Hu & Bentler,
der. 1999).

To our knowledge, there are no empirical studies supporting 2.2.3. Step 2: measurement invariance across gender
the PAIR three-dimension structure by a confirmatory factorial vali- The best-fitting model was then examined for its invariance
dation and gender similarities/differences in factorial structure and across gender by performing multigroup CFAs to test a similar
measurement invariance have not yet been tested for this inven- structure for women and men (e.g. Gregorich, 2006; Schmitt &
tory. Kuljanin, 2008; Vandenberg & Lance, 2000). For these models, the
mean was fixed at 0 and the variance at 1 in order to examine all
2. Method factor loadings and intercepts invariance for each latent variable.
An increasing level of constraints was used across different mod-
The sample consisted of 313 French women (aged 18 years to els as follows: model 1 represented configural invariance, which
73 years, M = 33.9, SD = 12.1) and 251 men (aged 20 years to 71 imposed the same structure for each group; model 2 represented
years, M = 37.0, SD = 12.3) in a committed romantic relationship for metric invariance, which constrained the factor loadings to be
at least one year (M = 12.1, SD = 10.5). Forty-nine percent of partici- invariant across gender; model 3 represented scalar invariance,
pants lived with a partner or had a civil union contract and 51% were which also constrained intercepts to be invariant across gender;
married. Concerning the level of education, 41.3% participants in (Schmitt & Kuljanin, 2008; Vandenberg & Lance, 2000). The last two
our sample had a level two or more years above the baccalaureate, multigroup models were compared to the nested configural model.
19.3% a level below or equal to two years after the baccalaureate, To conclude that there is measurement invariance across gen-
20.4% a bachelor’s degree, 17% a middle-school qualification or less der, the 2 difference between models should be non-significant,
and 2% did not give their level of education. the CFI difference < .01, RMSEA < .015, SRMR < .03 for metric invari-
ance and < .01 for scalar invariance (Chen, 2007). In cases in which
2.1. Procedure the non-invariance hypothesis was rejected, we tested the par-
tial invariance meaning of some invariant parameters (i.e., factor
Participants were contacted by investigators in the university, loading, intercepts), which were constrained to invariance while
on campus, or through personal social networks or Internet. They non-invariant parameters were released to be freely estimated in
were invited to participate in a university investigation on their both groups (Byrne, Shavelson, & Muthén, 1989).
current perception of intimacy in their romantic relationship by
completing a questionnaire. Confidentiality and anonymity were 2.2.4. Step 3: internal consistency reliability
guaranteed. Internal consistency reliability for each of the PAIR subscales
After indicating their age, gender and relationship duration, par- was assessed using Cronbach’s alpha. Alpha values of .70 or higher
ticipants were asked to rate the 36-item PAIR Inventory (Schaefer & indicate acceptable internal consistency (Kline, 2010).
112 E. Constant et al. / Revue européenne de psychologie appliquée 66 (2016) 109–116

3. Results for item 35 on the dimension engagement and communication that


revealing this ambiguity. Factor cross-loading was also identifying
3.1. Factor structure for item 17 (“We like playing and having fun together”) on the com-
munication and engagement dimensions and it was also redundant
The 27-item three-factor structure of the PAIR (model 1) devel- with item 29 (“We seldom find time to do fun things together”).
oped in previous study was tested and revealed a poor fit of the Moreover, when we examined the meaning of this item with regard
model to the data: 2 = 1287.009[321], p = .000, 2 /df ratio = 4.009, to this initial dimension “communication”, we can suggest that
CFI = .808, RMSEA = .073 (90% CI [.069, .077]) and SRMR = .0675. We this item do not correspond to the definition of this dimension. MI
did not confirm the structure proposed by Moore et al. (1998) in revealed also problem of misspecification for the item 3 (“I am satis-
our sample data. fied with the level of affection in our relationship”), particularly the
analysis of its meaning revealed a nonsense and ambiguity in the
communication dimension. Item 10 (“When having a serious dis-
3.2. Development of a revised model of the PAIR
cussion, it seems we have little in common”) revealed high degree
of overlap with three others items.
In front of this poor fit, we chose to conduct post-hoc analyses
With regard to these statistical informations and a seman-
using CFA in a theoretically driven approach in order to obtain a
tic analysis among three researchers, these items seemed to be
stable structure of the scale. To identify the best-fitting structure,
ambiguous in their initial dimension or redundant. We chose to
we examined:
remove them in order to obtain a stable and clear structure of this
scale.
• factor loading of each item to only retain those that strongly
Finally, we respecified the model by adding error covariance
saturate on their dimension; between three pairs of items with the largest MI within a same
• modification indices (MI) option which provide specific informa-
factor after having analyzed semantically each item and hav-
tion about model misspecification and to identify the construct ing determined that they were not ambiguous regarding their
validity of a model that better represents the sample data. dimension or redundant with others items. Respecification of the
hypothesized model of the PAIR structure (Moore et al., 1998) pro-
First, this option is conducted empirically by considering large vided an improvement of the model adjustment and an acceptable
MIs that argue for the presence of factor cross-loadings and error model fit (model 3): 2 = 266.781[129], p = .000, 2 /df ratio = 2.068,
covariance respectively (Byrne, 2009). Second, a reflection on the CFI = .956, RMSEA = .044 (90% CI [.036, .051]) and SRMR = .0395.
content validity was led (i.e., the semantic analysis of the item with Table 1 provides a summary comparison of the fit indices for the
regard to the concept/dimension measured, its inter-relation with three models.
others items and personal considerations between researchers The final version of the scale contains 18 items: 10 in the engage-
about the meaning of the items). ment dimension, five in the communication dimension and three
Based on the first model included 27 items that represented in the shared friends dimension.
the three-factor structure developed by Moore et al. (1998), we
identified five low factor loading (under .40) in the engagement
dimension which were removed (model 2): item 28 “My partner 3.3. Invariance measurement across gender
seldom tries to change my ideas” ( = .033), item 8 “As a couple, we
usually keep to ourselves” ( = .007), item 15 “I am able to tell my To identify the non-invariance for each factor loading, we tested
partner when I want sexual intimacy” ( = .380), item 32 “My part- the 2 difference between the full metric invariance model (all
ner disapproves of some of my friends” ( = .264), item 14 “We have factor loadings were fixed to be invariant, M2) and the full met-
few friends in common” ( = .290). This model did not fit the data ric invariance minus one free factor loading (the same analysis
according to goodness-of-fit indices, but we noted an improvement was conducted for each factor loading). A factor loading was con-
of the adjustment: 2 = 723.338[206], p = .000, 2 /df ratio = 3.511, sidered invariant if the comparison between these models had a
CFI = .884, RMSEA = .067 (90% CI [.062, .072]) and SRMR = .0530. significant 2 difference (we examined also a 2 with a p < .08).
Then, this model was modified examining MI greater than 10 A partial metric invariance model (M2p) was achieved by releas-
indicating a model misspecification in our sample data. Through ing the non-invariant factor loadings. From the partial metric
this procedure, and after verification of theoretical and conceptual invariance model, we proceeded in the same way to identify the
reasons, four items were deleted: one from the engagement dimen- non-invariant intercepts and to construct a partial scalar invariance
sion (item 10) and three from the communication dimension (item model.
35, 17 and 3). The comparison between the configural invariance model
Large MI for error covariances indicated a high degree of overlap and the partial invariance models were based on differences
in item content for item 35 (“We share few of the same interests”) between these models on 2 , CFI, SRMR and RMSEA. The partial
with item 11 (“I share in few of my partner’s interests”) and item 10 measurement invariant model (M2p) was not found to be different
(“When having a serious discussion, it seems we have little in com- from the configural model. Regarding the partial scalar invariant
mon”). Moreover, MI showed the presence of factor cross-loadings model (M3p-M2p ), the 2 difference was near significant with the

Table 1
Goodness-of-fit indices for alternative models of the structure of the PAIR.

Model Overall fit indices


2 df p 2 /df CFI SRMR RMSEA 90% CI

1. Hypothesized model (PAIR 27-item; Moore et al., 1998) 1287.009 321 .000 4.009 .808 .0675 .073 [.069, .077]
2. Model 1 (factor loadings below .40 removed) 723.338 206 .000 3.511 .884 .0530 .067 [.062, .072]
3. Model 2 (correlated uniqueness model) 266.781 129 .000 2.068 .956 .0395 .044 [.036, .051]

PAIR: personal assessment of intimacy in relationships; CFI: comparative fit index; SRMR: standardized root mean square residual; RMSEA: root mean square error of
approximation.
E. Constant et al. / Revue européenne de psychologie appliquée 66 (2016) 109–116 113

Table 2
Goodness-of-fit indices of the models testing measurement invariance and partial invariance.

Overall fit of the models Difference between models and M1

Model 2 df 2 /df CFI SRMR RMSEA RMSEA [90% CI] 2 df p of the 2 CFI SRMR RMSEA

M1 393.289 258 1.524 .957 .041 .031 [.024; .036]


M2 451.081 276 1.634 .945 .053 .034 [.028; .039] 57.792 18 < .001 .012 .012 .003
M3 589.092 294 2.004 .907 .054 .042 [.037; .047] 195.803 36 < .001 .05 .013 .011
M2p 413.037 272 1.519 .955 .049 .030 [.024; .036] 19.748 14 .138 .002 .008 −.001
M3-M2p 552.312 290 1.905 .917 .051 .040 [.035; .045] 159.023 32 < .001 .04 .01 .009
M3p-M2p 427.775 281 1.522 .954 .049 .030 [.025; .036] 34.486 23 .058 .003 .008 −.001

CFI: comparative fit index; SRMR: standardized root mean square residual; RMSEA: root mean square error of approximation; : difference; M1: configural invariance;
M2: metric invariance; M3: scalar invariance; p: partial invariance; M2p: partial metric invariance; M3-M2p : scalar invariance releasing the non-invariant factor loadings;
M3p-M2p : partial scalar invariance releasing the non-invariant factor loading.

configural model (p = .058), but the differences on the other indices friends dimension, all factor loadings and intercepts were invariant
showed no substantial difference between these models. More- and mixed patterns of non-invariant factor loadings and inter-
over, results demonstrated good fit indices for partial invariance cepts were found in engagement and communication dimensions
models (Table 2). (Table 3).
Results from the partial scalar invariance model showed four Because metric and scalar non-invariance may influence latent
non-invariant factor loadings and nine non-invariant intercepts. variable means, they were estimated for the full scalar invariance
Three non-invariant factor loadings in the engagement dimension model (M3), the partial scalar invariance model (M3p-M2p ), and the
were identified – two of which were higher for men than for women free model (M1 or configural model, also called the ‘true’ model).
and one higher for women than for men – and one in the commu- We constructed a model when the means and variance values of
nication dimension – higher for women than for men. Moreover, the latent variables were released and the intercepts were fixed at
six non-invariant intercepts were found in the engagement dimen- 0 and the factor loadings were fixed at 1 for the first invariant item
sion – two of which were higher for men than for women and four for each dimension (items 2, 4 and 11). Results showed (Table 4)
higher for women than for men – and three in the communication that the means estimated in the full and partial invariance models
dimension – two of which were higher for men than for women did not differ much from the means estimated by in the configurale
and one higher for women than for men. To summary, in the shared model.

Table 3
Items assignment, factor loadings and intercepts of the partial invariance model (M3p-M2p ).

Items Factor loading Stand. Intercept

Women Men Women Men

Shared friends
2 We enjoy spending time with other couples (Nous aimons passer du temps avec d’autres couples) .626 .650 3.017 3.017
20 Having time together with friends is an important part of our shared activities (Prendre du temps .853 .868 2.593 2.593
ensemble avec des amis est une part importante de nos activités communes)
26 Many of my partner’s closest friends are also my closest friends (Beaucoup des meilleurs amis de .413 .418 2.284 2.284
mon/ma partenaire sont aussi mes amis les plus proches)
Communication
1 My partner listens to me when I need someone to talk to (Mon/ma partenaire m’écoute quand j’ai .730 .467 3.453 3.564
besoin de parler à quelqu’un)
4 My partner helps me clarify my thoughts and feelings (Mon/ma partenaire m’aide à clarifier mes .702 .732 3.152 3.152
pensées et mes sentiments)
7 I can state my feelings without him/her getting defensive (Je peux exprimer mes sentiments sans .652 .661 2.967 2.967
qu’il/elle soit sur la défensive)
19 My partner can really understand my hurts and joys (Mon/ma partenaire peut vraiment comprendre .621 .716 3.04 3.325
mes blessures et mes bonheurs)
34 We have an endless number of things to talk about (Nous avons un nombre infini de choses à nous .627 .606 2.714 2.531
raconter)
Engagement
9 I feel our level of affection is just routine (Je sens que l’affection que l’on se porte l’un à l’autre est juste .734 .636 3.123 2.931
basée sur la routine)
11 I share in few my of my partner’s interests (Je partage peu les centres d’intérêts de mon/ma partenaire) .412 .397 2.678 2.678
13 I often feel distant from my partner (Je me sens souvent distant(e) de mon/ma partenaire) .801 .700 3.243 3.243
16 I feel “put-down” in a serious conversation with my partner (Je me sens méprisé(e) au cours d’une .546 .522 3.406 3.406
conversation sérieuse avec mon/ma partenaire)
21 Because of my partner’s lack of caring, I “hold back” my sexual interest (En raison du manque d’amour .663 .618 3.266 3.027
de mon/ma partenaire, je freine mon désir sexuel)
22 I feel it is useless to discuss some things with my partner (Je sens que c’est inutile de discuter de .646 .623 3.258 3.258
quelque chose avec mon/ma partenaire)
25 I feel neglected at times by my partner (Je me sens parfois négligé(e) par mon/ma partenaire) .563 .599 2.570 2.816
29 We seldom find time to do fun things together (Nous trouvons rarement le temps de faire des choses .469 .586 2.817 2.636
amusantes ensemble)
31 I sometimes feel lonely when we’re together (Je me sens parfois seul(e) quand nous sommes ensemble) .687 .771 2.879 3.324
33 My partner seems disinterested in sex (Mon/ma partenaire semble désintéressé(e) du sexe) .409 .584 3.152 2.872

Stand.: standardized. The non-invariant values are in bold type.


114 E. Constant et al. / Revue européenne de psychologie appliquée 66 (2016) 109–116

Table 4 revealed a better intimacy concerning listening and understanding


Means and standard errors of the latent variables for women and men for full and
by their partner (item 1 et item 9), whereas women expressed a
partial invariance models and free model as well as estimation bias.
better communication in the variety of subjects raised (item 34).
Full invariant Partial invariant Free These results agreed those of previous studies in that women
Model M3 M3p-M2p M1 are more inclined to emotional closeness, affection and commu-
nication (e.g. emotional self-disclosure), whereas men are more
Mean SE Mean SE Mean SE
inclined to sharing joint activities or leisure time, giving instru-
Women mental help and accord a prominent place to the sexuality in
Engagement 2.666 .055 2.686 .057 2.696 .068
their romantic relationship (Heller & Wood, 1998; Ridley, 1993;
Communication 3.080 .057 3.085 .057 3.045 .059
Shared friends 3.028 .047 3.028 .047 3.016 .053 Tannen, 1990). In addition, women are more listening and under-
Men standing to their partner’s needs (Hook et al., 2003). Our results
Engagement 2.687 .055 2.661 .057 2.645 .075 are consistent with the scientific literature supporting the idea
Communication 3.241 .050 3.218 .054 3.263 .058 that men and women define and experience intimacy in differ-
Shared friends 3.007 .050 3.007 .049 3.020 .059
ent ways, which may be partly explained by gender socialization
SE: standard error. (Gaia, 2002; Heller & Wood, 1998; Hook et al., 2003). Gender
is considered as a social construct in which men and women
3.4. Internal consistency reliability receive gender messages throughout life (Heller & Wood, 1998). It
would contribute to the construction of a gender identity through
Following model respecification and measurement invariance which men report globally higher separateness (i.e., self-other
across gender of the PAIR, Cronbach’s alphas were estimated of the differentiation and independence) whereas women report higher
revised factors as follows: .84 for the revised engagement factor, connectedness with others – and particularly their partner – that
.80 for the revised communication factor and .635 for the shared is essential for themselves (Lang-Takac & Osterweil, 1992). More-
friends factor, which presented a limited internal consistency. over, Gaia (2002) proposed that men’s emotional withdrawal or
independence relative to each other could be explained by their
fear of social rejection because these behaviors would seem to
4. Discussion suggest vulnerably or weakness in accordance to social expecta-
tions and gender norms. Continuing to examine gender differences
This study aimed to confirm the previous three-factor structure in intimacy appears to be important in order to determine how
based on the original scale in general population (Moore et al., 1998) women and men conceive intimacy and its impact on the evolution
and to assess its measurement equivalence to test whether the scale of the relationship (e.g., pattern of communication, problem solving
has the same function across gender. This previous three-factorial techniques).
structure that has been developed was not confirmed because it In view of this lack of metric and scalar invariance, assess-
presented a poor model fit. Therefore, the PAIR revised structure ment of its impact on the estimation of latent variables means
based on this previous study was created using post-hoc analysis in was important in order to compare the means of intimacy for men
order to detect misfitting parameters and to obtain a structure with and women. A lack of invariance may impair the interpretation
a good adjustment. Finally, nine items in the engagement dimen- of gender differences especially when the pattern of invariance is
sion and one item in the communication dimension were removed. uniform and the proportion of non-invariant items is high (Chen,
The obtained solution of the PAIR contained 18 items distributed 2008). In this present study, the non-invariant factor loadings were
over three revised factors: minors, mixed and the non-invariant intercepts were not uniforms.
The estimation of means did not show high differences of means
• engagement (10 items); between the full, the partial and the configurale models. There-
• communication (5 items); fore, we recommend the use of this scale to assess intimacy but we
• shared friends (3 items), and presented a good fit according to encourage further studies to examine the validity of the compari-
the goodness-of-fit statistics. son of means between men and women if one does not take into
account of the lack of invariance.
The present study provided complementary results concerning Finally, internal consistency was acceptable for the engagement
gender differences in the factorial structure of the revised version of and communication subscale. The shared friends subscale also pre-
the PAIR. Findings related to measurement invariance concerning sented a limited internal consistency. This may be explained by the
factor loadings indicated that the scale presents a lack of invari- presence of few items in these subscales (Kline, 2010).
ance for three-factor loadings in the engagement dimension with To our knowledge, this is the first study in which measure-
a mixed pattern of invariance: mutual activities with the partner ment invariance across men and women on the PAIR is examined.
(item 29) and sexuality (item 33) are more explained by engage- Strength of the present study provides valuable information
ment factor for men than women, and this was the case for women regarding the construct of intimacy across gender as assessed by the
concerning the emotional proximity with the partner (item 13). PAIR. Although findings revealed a lack of invariance across women
Moreover, one factor loading in the communication dimension con- and men in the construction of intimacy, these differences are very
cerning the listening of partner (item 1) is more explained by the interesting from theoretical and clinical perspectives. It would be
communication factor for women than for men. particularly interesting to take into consideration these gender
Findings revealed also nine non-invariant intercepts in a mixed differences in future research that investigates the link between
pattern of invariance, but more non-invariant intercepts are higher intimacy and other couple variables (e.g. marital satisfaction and
for men than women. In the engagement dimension, four intercepts marital adjustment) or individual variables of emotional function-
were higher for men that mean a lower intimacy than women con- ing (e.g. empathic abilities, emotional regulation), notably because
cerning the sexuality area (item 21 and item 33) or the sharing of of the importance of emotional transactions between partners in a
common and various activities (item 9 and item 29). By contrast, romantic relationship.
women expressed more negligence by their partner (item 25) or Further studies are required to specify gender differences
loneliness when they are together (item 31). Finally, in the com- regarding the contribution of each element to intimacy and their
munication dimension, two intercepts were higher for men that respective levels (e.g. sexual, feeling neglected).
E. Constant et al. / Revue européenne de psychologie appliquée 66 (2016) 109–116 115

The version of the PAIR proposed in this study and the identifi- Gaia, A. C. (2002). Understanding emotional intimacy: A review of conceptualiza-
cation of gender differences in the theoretical meaning and level tion, assessment and the role of gender. International Social Science Review, 77,
151–170.
of intimacy across men and women confirm the importance to Greeff, A. P., & Malherbe, H. L. (2001). Intimacy and marital satisfaction in spouses.
consider gender differences in future studies trying to understand Journal of Sex & Marital Therapy, 27(3), 247–257.
the dynamic of couple relationships. The multidimensional opera- Gregorich, S. E. (2006). Do self-report instruments allow meaningful comparisons
across diverse population groups? Testing measurement invariance using the
tionalization of intimacy is important in order to investigate the confirmatory factor analysis framework. Medical Care, 44(11, Suppl. 3), 78–94.
contribution of each component of intimacy and to analyze gender Harper, J. M., & Elliott, M. L. (1988). Can there be too much of a good thing? The rela-
patterns of intimacy and their mutual influence on marital adjust- tionship between desired level of intimacy and marital adjustment. American
Journal of Family Therapy, 16, 351–360.
ment. Using inter-influence analyses between partners, it would be
Hetherington, S. H., & Soeken, K. L. (1990). Measuring changes in intimacy and sex-
important to explore these gender differences regarding the per- uality: A self-administered scale. Journal of Sex Education and Therapy, 16(3),
ception and definition of intimacy in relation to partners’ emotional 155–163.
Heller, P. E., & Wood, B. (1998). The process of intimacy: Similarity, understanding
functioning in order to deepen understanding of romantic rela-
and gender. Journal of Marital and Family Therapy, 24(3), 273–288.
tionships. Another development will concern the characterization Hook, M. K., Gerstein, L. H., Detterich, L., & Gridley, B. (2003). How close are we? Mea-
of the patterns of interaction, in intimate situations, using behav- suring intimacy and examining gender differences. Journal of Counseling & Devel-
ior analyses and somatic/physiological markers. Nevertheless, it opment, 81(4), 462–472. http://dx.doi.org/10.1002/j.1556-6678.2003.tb00273.x
Hopkins, K. D., & Weeks, D. L. (1990). Tests for normality and measures of skewness
seems important to consider gender differences on the means of and kurtosis: Their place in research reporting. Educational and Psychological
engagement and communication subscales with caution. Measurement, 50, 717–729.
Our results may also have clinical implications in couple Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance struc-
ture analysis: Conventional criteria versus new alternatives. Structural Equation
counseling to help couples increase their level of intimacy and Modeling: A Multidisciplinary Journal, 6(1), 1–55. http://dx.doi.org/10.1080/
understand gender differences that may impact the quality of their 10705519909540118
romantic relationship. Marital therapists may target their interven- Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.).
New York: The Guilford Press.
tion on the understanding of one partner’s intimacy representation Kline, R. B. (2010). Principles and practice of structural equation modeling (3rd ed.).
and needs – particularly on the engagement and communication New York: The Guilford Press.
dimensions – and thus facilitate the expression and feelings of inti- Kouneski, E. F., & Olson, D. H. (2004). A practical look at intimacy: ENRICH couple
typology. In D. J. Mashek, & A. Aron (Eds.), Handbook of closeness and intimacy
macy in romantic relationships (Heller & Wood, 1998; Schnarch,
(pp. 117–133). Mahwah, NJ: Psychology Press.
1991). Therapists could also discuss with couples about the rela- Lang-Takac, E., & Osterweil, Z. (1992). Separateness and connectedness: Differ-
tionship dynamic and how gender socialization may partially ences between the genders. Sex Roles, 27(5–6), 277–289. http://dx.doi.org/
10.1007/BF00289929
explain their difficulties in the expression of intimacy.
Laurenceau, J.-P., Barrett, L. F., & Rovine, M. J. (2005). The interpersonal process
This short instrument may facilitate its use in applied research. model of intimacy in marriage: A daily-diary and multilevel modeling approach.
It also could be helpful to assess couple intimacy routinely and offer Journal of Family Psychology, 19(2), 314–323. http://dx.doi.org/10.1037/
a complementary useful guidance for therapies. 0893-3200.19.2.314
Laurenceau, J.-P., Rivera, L. M., Schaffer, A. R., & Pietromonaco, P. R. (2004). Intimacy
as an interpersonal process: Current status and future directions. In D. J. Mashek,
Disclosure of interest & A. Aron (Eds.), Handbook of closeness and intimacy (pp. 61–78). Mahwah, NJ:
Psychology Press.
Maslow, A. H. (1954). Motivation and personality. New York: Harper.
The authors declare that they have no competing interest. McAdams, D. P., & Vaillant, G. E. (1982). Intimacy motivation and psychosocial
adjustment: A longitudinal study. Journal of Personality Assessment, 46, 486–493.
Miller, R. S., & Lefcourt, H. M. (1982). The assessment of social intimacy.
Acknowledgements Journal of Personality Assessment, 46(5), 514–518. http://dx.doi.org/10.1207/
s15327752jpa4605 12
The authors wish to extend their grateful thanks to all the par- Moore, K. A., McCabe, M. P., & Stockdale, J. E. (1998). Factor analysis of the Personal
Assessment of Intimacy in Relationships Scale (PAIR): Engagement, commu-
ticipants of the study. The authors also thank The Conseil Régional
nication and shared friendships. Sexual and Marital Therapy, 13(4), 361–368.
du Nord-Pas-de-Calais and the SIRIC ONCOLille, (grant INCa-DGOS- http://dx.doi.org/10.1080/02674659808404254
Inserm 6041) for their help. Moss, B. F., & Schwebel, A. I. (1993). Defining intimacy in romantic relationships.
Family Relations, 42(1), 31. http://dx.doi.org/10.2307/584918
Nezhad, M. Z., & Goodarzi, A. M. (2011). Sexuality, intimacy, and marital satisfac-
References tion in Iranian first-time parents. Journal of Sex & Marital Therapy, 37(2), 77–88.
http://dx.doi.org/10.1080/0092623X.2011.547336
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal Patrick, S., & Beckenbach, J. (2009). Male perceptions of intimacy: A qualitative study.
attachments as a fundamental human motivation. Psychological Bulletin, 117(3), The Journal of Men’s Studies, 17(1), 47–56. http://dx.doi.org/10.3149/jms.1701.47
497–529. http://dx.doi.org/10.1037/0033-2909.117.3.497 Prager, K. J. (1995). The psychology of intimacy. New York: Guilford Press.
Borsboom, D. (2006). The attack of the psychometricians. Psychometrika, 71(3), Reis, H. T., Collins, W. A., & Berscheid, E. (2000). The relationship context of
425–440. http://dx.doi.org/10.1007/s11336-006-1447-6 human behavior and development. Psychological Bulletin, 126(6), 844–872.
Van den Broucke, S., Vertommen, H., & Vandereycken, W. (1995). Construction and http://dx.doi.org/10.1037//0033-2909.126.6.844
validation of a marital intimacy questionnaire. Family Relations, 44(3), 285–290. Reis, H. T., & Shaver, P. (1988). Intimacy as an interpersonal process. In S. W. Duck
http://dx.doi.org/10.2307/585527 (Ed.), Handbook of personal relationships (pp. 367–389). Chichester, England:
Byrne, B. M., Shavelson, R. J., & Muthén, B. (1989). Testing for the equivalence Wiley.
of factor covariance and mean structures: The issue of partial measure- Reiter, M. D., Richmond, K., Stirlen, A., & Kompel, N. (2009). Exploration of intimacy in
ment invariance. Psychological Bulletin, 105(3), 456–466. http://dx.doi.org/ intercultural and intracultural romantic relationships in college students. College
10.1037/0033-2909.105.3.456 Student Journal, 43(4), 1080–1083.
Byrne, B. M. (2009). Structural equation modeling with AMOS: Basic concepts, applica- Ridley, J. (1993). Gender and couples: Do men and women seek different kinds
tions, and programming (2nd ed.). New York: Routledge. of intimacy? Sexual and Marital Therapy, 8(3), 243–253. http://dx.doi.org/
Chen, F. F. (2007). Sensitivity of goodness-of-fit indexes to lack of measure- 10.1080/02674659308404971
ment invariance. Structural Equation Modeling: A Multidisciplinary Journal, 14(3), Schaefer, M. T., & Olson, D. H. (1981). Assessing intimacy: The pair inventory*.
464–504. Journal of Marital and Family Therapy, 7(1), 47–60. http://dx.doi.org/10.1111/
Chen, F. F. (2008). What happens if we compare chopsticks with forks? The impact j.1752-0606.1981.tb01351.x
of making inappropriate comparisons in cross-cultural research. Journal of Schmitt, N., & Kuljanin, G. (2008). Measurement invariance: Review of practice and
Personality and Social Psychology, 95(5), 1005–1018. http://dx.doi.org/10.1037/ implications. Human Resource Management Review, 18(4), 210–222.
a0013193 Schnarch, D. M. (1991). Constructing the sexual crucible: An integration of sexual and
Descutner, C. J., & Thelen, M. H. (1991). Development and validation of a Fear of marital therapy (1st ed.). New York: W.W. Norton & Company.
Intimacy scale. Psychological Assessment, 3(2), 218–225. Sternberg, R. J. (1986). A triangular theory of love. Psychological Review, 93(2),
Elliott, M. L., Bingham, R. D., Nielsen, S. C., & Warner, P. D. (1986). Marital intimacy 119–135. http://dx.doi.org/10.1037/0033-295X.93.2.119
and satisfaction as a support system for coping with police officer stress. Journal Sternberg, R. J., & Grajek, S. (1984). The nature of love. Journal of Personality and Social
of Police Science and Administration, 14, 40–44. Psychology, 47(2), 312–329. http://dx.doi.org/10.1037/0022-3514.47.2.312
Erikson, E. H. (1950). Childhood and society. New York: Norton. Sullivan, H. S. (1953). The interpersonal theory of psychiatry. New York: Norton.
116 E. Constant et al. / Revue européenne de psychologie appliquée 66 (2016) 109–116

Tannen, D. (1990). Gender differences in topical coherence: Creating involve- relationships questionnaire. Psycho-oncology, 23(3), 346–349. http://dx.
ment in best friends’ talk. Discourse Processes, 13(1), 73–90. http://dx.doi.org/ doi.org/10.1002/pon.3416
10.1080/01638539009544747 Waring, E. M., & Reddon, J. R. (1983). The measurement of intimacy in mar-
Vandenberg, R. J., & Lance, C. E. (2000). A review and synthesis of the measure- riage: The Waring Intimacy Questionnaire. Journal of Clinical Psychology, 39(1),
ment invariance literature: Suggestions, practices, and recommendations for 53–57.
organizational research. Organizational Research Methods, 3(1), 4–69. Worthington, E., Buston, B., & Hammonds, T. (1989). A component analysis of mar-
Walker, L. M., Hampton, A., & Robinson, J. W. (2014). Assessment of rela- riage enrichment: Information and treatment modality. Journal of Counseling
tional intimacy: Factor analysis of the personal assessment of intimacy in and Development, 67(10), 555–560.

You might also like