Professional Documents
Culture Documents
Variable
Collins and Read (1990, p. 644) “developed a multi-item scale to measure dimensions
underlying adult attachment styles to replace Hazan and Shaver’s discrete, categorical measure.
Based on multi-sentence type descriptions, Collins and Read (1990) had argued that there were
limitations to a categorical measure.” Since each type description contained statements about
more than one aspect of relationships (e.g., being comfortable with closeness and being able to
depend on others), respondents therefore had to accept an entire description that may not have
reflected their position on all facets, and likewise, researchers were unable to assess the degree to
which a particular style characterized a particular person. Moreover, the categorical measure
assumed that there were three mutually exclusive attachment styles, making it difficult to
examine possible relations among styles or to evaluate whether there were actually just three
adult attachment patterns. For these reasons, Collins and Read (1990) sought to develop a more
sensitive measure of adult romantic attachment styles, which was subsequently revised by
Collins (1996).
Description
The 18-item AAS was originally constructed by Collins and Read (1990), based on
Hazan and Shaver’s (1987) Following factor analysis with oblique rotation (N = 406), three
subscales were elucidated (each with 6 items): (a) the Close subscale concerns comfort with
closeness and intimacy; (b) the Depend subscale assesses comfort with depending on others and
2
the belief that partners can be relied upon when needed; and (c) the Anxiety subscale measures
the extent to which a person is worried about being rejected and abandoned (Collins & Read,
Table 2; cf. Collins, 1996, pp. 814-815). Respondents rate each item on a 5-point Likert-type
scale. Collins (1996) revised the scale by replacing a few items to improve reliability and
validity. The revised 18-item version correlated .98 with the original version (N = 295).
Sample
Collins and Read (1990) collected data from 406 undergraduates (184 males; 206
females) ranging in age from 17 to 37 years (M = 18.8 years). In a second study, Collins and
Read utilized a sample of 118 undergraduates ranging in age from 17 to 24 years (M = 18.6
years). In a third study, Collins and Read employed a sample of 71 dating couples ranging in age
from 18-44 years (M = 22 years). Subsequently, Collins (1996) utilized samples of 135
undergraduates (53 male; 82 female) ranging in age from 18 to 25 years (M = 18.7 years), of
whom, 55% were in a romantic relationship. In a replication study, Collins (p. 821) utilized a
further sample of 129 undergraduates (56 male; 73 female) ranging in age from 17 to 30 years
(M = 19.2 years).
Reliability
Internal Consistency
Collins and Read (1990, p. 646) reported Cronbach alpha coefficients for the three
subscales (Close, Depend, and Anxiety) of .69, .75, and .72, respectively (N = 406).
Subsequently, Collins (1996, p. 814) reported alpha coefficients of .77, .78, and .85, respectively.
3
In the Collins (1996) replication study (N = 129), alpha coefficients for the three subscales were
Test-Retest
A subset (N = 101) of Collins and Read’s (1990) sample completed the AAS again two
months later. Stability coefficients for Close, Depend, and Anxiety subscales were found to
Validity
Convergent/Concurrent
Collins and Read (1990, Table 8, p. 652). Subsequently, in a sample of 135 undergraduates, the
Close and Depend subscales of the AAS correlated .53 with each other (Collins, 1996, p. 814). In
the replication sample of 129 undergraduates, the Close and Depend subscales correlated .67
(Collins, p. 821). Also, positive associations between the Close and Depend subscales and other
measures of adult attachment have been reported. Thus, the Close subscale correlated positively
(.25 with love/security, .29 with responsive/dependable .16 with self-worth/reliance, .15 with
trust, .30 with partner warmth/closeness, and .26 with minimizing negative impact). Likewise,
the Depend subscale correlated positively (.27 with love/security, .16 with
responsive/dependable, .14 with self-worth/reliance, .17 with trust, .38 with partner
warmth/closeness, .28 with minimizing negative impact) (Collins, 1996, p. 816). Additional
Divergent/Discriminant
had been reported by Collins and Read (1990, Table 8, p. 652). Subsequently, Collins (1996, p.
814) reported that the Anxiety subscale correlated -.34 with the Close subscale, and -.46 with the
Depend subscale. In the replication sample, the Anxiety subscale correlated -.28 with the Close
subscale, and -.46 with the Depend subscale (Collins, p. 821). As well, Collins (p. 816) reported
(N = 135) that the Anxiety subscale correlated -.25 with love/security, -.21 with
responsive/dependable, -.20 with self-worth/reliance, -.18 with trust, -.34 with partner
warmth/closeness, and -.27 with minimizing negative impact. Additional negligible or negative
correlations with various other measures indicative of divergent/discriminant validity also have
Construct/Factor Analytic
As indicated above, in constructing the original AAS, Collins and Read (1990) conducted
an exploratory factor analysis with oblique rotation (N = 406) based on the 21 x 21 item
intercorrelation matrix and extracted three factors that clearly defined the AAS structure (see
Collins & Read, Table 2, p. 647, for the factor loadings on each of the original 198 items).
“Factor 1 (Depend) and Factor 3 (Close) were moderately correlated (.41)” suggesting some
measurement overlap between these two AAS subscales. Subsequently, Collins (1996, pp. 814-
815, N = 295) confirmed the tripartite structure of the AAS (based on a factor analysis of the
revised items).
5
Criterion/Predictive
Evidence of predictive validity was reported by Collins (1996, Table 5, p. 823). For
example, as predictors of scores on the Anxiety subscale, the standardized beta coefficients were
(-.23), partner warmth/closeness (-.29), and minimizing negative impact (-.21). In contrast, there
were no significant predictors for scores on the Close/Depend subscales. Further evidence of
Location
Collins, N. L. (1996). Working models of attachment: Implications for explanation, emotion, and
The three AAS subscales can be used as continuous measures of the dimensions
underlying differences in adult attachment patterns (comfort with closeness, comfort with
depending on others, and fear of rejection). Second, scores on the AAS can be used to place
the Close and Depend subscales correlate closely with each other, in some studies (e.g., Collins
et al., 2000) they have been collapsed into a single dimension, making the resultant two
Notes.
Items are rated on a 5-point Likert-type scale ranging from: 1 = Not at all characteristic to 5 =
Very characteristic.
7
Items 1, 5, 6, 12, and 14 are reverse-keyed prior to computing the subscale scores.
The Close score is computed by averaging items 1, 6, 8, 12, 13, and 17. Higher scores indicate
greater discomfort with closeness and intimacy.
The Depend score is computed by averaging items 2, 5, 7, 14, 16, and 18. Higher scores reflect
greater discomfort depending on others.
The Anxiety score is computed by averaging items 3, 4, 9, 10, 11, and 15. Higher scores reflect
greater fear of being rejected or unloved.