DOCUMENT RESUME TM 024 927 ED 393 91, Chongruksa, Jiratha; Thompson, Bruce AUTHOR A Note Regarding the Reliability and Validity of TITLE Scores on Collins and Read's Adult Attachment Scale. Jan 96 PUB DATE I4p.; Paper presented at the Annual Meeting of the NOTE Southwest Educational Research Association (New Orleans, LA, January 1996). Evaluative/Feasibility (142) Reports PUB TYPE Speeches/Conference Papers (150) MF01/PC01 Plus Postage. EDRS PRICE Adults; *Attachment Behavior; *College Students; DESCRIPTORS Foreign Countries; Higher Education; *Likert Scales; Psychometrics; *Scores; Test Construction; Test Reliability; Test Use; *Test Validity *Adult Attachment Scale IDENTIFIERS ABSTRACT Attachment styles have been shown to be enduring and influential, and studies of attachment phenomena require scores that are reliable and valid. The present study investigated the psychometric integrity of scores on the Adult Attachment Scale using data from 283 college students. This instrument is an 18-item Likert scale that is an expansion of the measure developed by C. Hazan and P. R. Shaver (1987, 1990). Results strongly supported the reliability and validity of scores from the measure. These results compared extremely favorably with those from previous studies, and certainly were more ,7_,vorable than those for the Hazan and Shaver measure. (Contains 2 tables and 19 references.) (Author/SLD) Reproductions supplied by EDRS are the best that can be made from the original document. attach.wp1 1/18/96 PERMISSION TO REPRODUCE I HIS U S DEPAIETMENT OF EDUCATION Orace ol Educatronal Research anO Improvement MATERIAL HAS BEEN GRANTED BY EOU ATIONAL RESOURCES INFORMATION CENTER (ERIC) kugE 751DAp4-16evo rep,oducal Tps document has been an reccovect Iron, the person or oconaatron 0.9.w.ga C1 kfinor changes have been made to ,morpre reproaluction ouhly Ppnts 01 view, Or opmons stated en th,s d TO THE EDUCATIONAL RESOURCES ment do not necessarily represent olircrai INFORMATION CENTER (ERIC) OF RI Posdron or poficy A Note Regarding the Reliability and Validity of Scores on Collins and Read's Adult Attachment Scale Bruce Thompson Jiratha Chongruksa Texas A&M University Chaing Mai University (Thailand) and Baylor College of Medicine Adult Attachment RUNNING HEAD: the Southwest Paper presented at the annual meeting of Educational Research Association, New Orleans, January, 1096. MAIL/Ong COPY Adult Attachment -2- Abstract Attachment styles have been shown to be enduring and influential. Studies of attachment phenomena require scores that are reliable psychometric the investigated study present The and valid. integrity of scores on the Adult Attachment Scale. Results compared extremely favorably with those from previous studies, and certainly were more favorable than those for the Hazan and Shaver measure. Adult Attachment -3- Theories about attachment styles (e.g., Hazan & Shaver, 1987) and their development during infancy and early childhood have In theory, more securely attached attracted considerable interest. persons are not distracted by concerns over unmet attachment needs, presumably hecause as secure infants such persons had mothers who were consistently sensitive and responsive Blehar, (Ainsworth, Thus, such people grow up trusting and & Wall, 1978). Waters, displaying optimism towards the world (Erikson, 1950). Sroufe (1983) found that securely attached preschoolers have research Furthermore, problem-solving capabilities. superior suggests that optimists are more likely to engage in active coping and are less likely to disengage from threats and distress (Carver Weintraub, 1989; Scheier, Carver, Scheier, 1990b; 1990a, & & avoidant In adulthood, Carver, 1986). Weintraub, Scheier, & personalities tend to focus on work activity rather than on relationships, as an ego defensive reaction (Hazan & Shaver, 1990). Hazan and Shaver (1987) translated the three infant attachment styles into terms appropriate for adult relationships: secure, The attachment styles have been applied as avoidant and anxious. constructs in several areas of inquiry, but especially in studies interest love--an area scholarly renewed considerable of of (Thompson & Borrello, 1992). attachment styles are fairly According to Bowlby (1980), stable across the life-span, and tend to resist dramatic change. Thus, because attachment styles seem to be both influential and stable, studies of attachment styles and their development and Adult Attachment -4- consequences appear to be warranted. Of course, effective inquiry in this area presumes the availability of instruments that yield circumstances under variety valid scores reliable of and a Too many researchers conduct substantive 1994b). (Thompson, analyzes with unreliable scores, and even with scores that have reliability coefficients that are not only negative,'but that have coefficients less than even -1 (Thompson, 1994a). the investigate conducted study to present was The psychometric integrity of scores on the Adult Attachment Scale (Collins & Read, 1990). This instrument is an 18-item Likert-scale measure, six items of which measure each of three attachment styles typology infant-attachment of the formulated originally in Ainsworth, Blehar, Waters and Wall (1978). is an The Adult Attachment Scale Read, (Collins 1990) & Collins and Read expansion of Hazan and Shaver's (1987) measure. (1990) conducted a factor analysis of their 18-item measure that However, the three factors did not directly yielded three factors. correspond to the three discrete styles of secure, avoidant, and Instead, Collins and Read (1990) labelled their related anxious. dimensions Close, Depend and Anxiety. reflects comfort with closeness and The dimension, Close, Depend reflects trust in others and confidence that intimacy. The dimension, Anxiety, others will be available when needed. Cronbach alpha involves concern with being abandoned or unloved. coefficients for scores on each dimension in their sample were .75 .69 for Close, respectively. and for Anxiety, for Depend, .72 Adult Attachment -5- Test-retest reliability coefficients for Close, Depend, and Anxiety .71, and .52, respectively. were .68, Method Subiects southwestern large a at courses Students enrolled in Data from 283 university were asked to participate in the study. The sample included subjects were used in the present analyses. 197 women and 86 men, ranging in age from 17 to 38,.with a,mean age of 20.56 (SD= 2.88) years. Results Our first research analysis focused on the reliability of the to the data that we attended We collected. the scores for reliability of our own scores, because reliability is a function 'of scores, and not of tests, and is actually affected by the persons Coefficients themselves who take a given test (Thompson, 1994b). alpha for the Adult Attachment Scale dimensions of Depend, Anxiety and Close were .854, .784, and .847 in our sample, respectively. We also conducted item analyses for the measure, as reported correlation item-to-total corrected the All Table in 1. coefficients (Thompson & Levitov, 1985) were sizeable and in the expected direction (i.e., positive), expect the results for item The alpha-if-deleted item statistic for the 11, an Anxiety item. item, prior to our decision to omit this item, indicated that alpha would improve from .713 to .784 if the item was omitted. INSERT TABLE 1 ABOUT HERE. Our second research analysis focused on the validity of scores -6- Adult Attachment role on the scale. Many researchers have acknowledged the prominent that factor analysis can play in efforts to establish construct Joy Guilford's article some 50 years ago illustrates the validity. historical pattern of sometimes even speaking of construct validity as factorial validity: of two kinds... in my opinion The Validity, is by given its test validity is of factorial a loadings in meaningful, common, reference factors. (Guilford, 1946, pp. 428, emphasis added) Similarly, Gorsuch (1983, p. 350) noted that, "A prime use of factor analysis has been in the development of both the operational constructs for an area and the operational representatives for the In short, "factor analysis is intimately theoretical constructs." involved with questions of validity.... Factor anlysis is at the heart of the measurement of psychological constructs" ;Nunnally, 1978, pp. 112-113). Factors here were extracted using the principal components the varimax rotated to factor structure was method and the criterion. Principal components and common factor methods tend to yield equivalent structures when (a) the reliability of scores on the factored entities (here, items) is high or (b) the number of Daniel, is relatively large in (Thompson factored entities & press). Table 2 presents the factor pattern/structure coefficients for The first Adult Attachment Scale the Adult Attachment Scale. factor explained the greatest percentage of the covariance jn Adult Attachment -7- in press). responses (35.2%), prior to rotation (Thompson & Daniel, The second and the third Adult Attachment Scale factors accounted for 13.2%, and 8.1% of the total covariance of the 18 items, again prior to rotation. INSERT TABLE 2 ABOUT HERE. The first factor corresponded to the first factor in the original study, whereas the second and the third factor alternated Differences in their order in comparison with the original study. factor order are generally not considered noteworthy, as long as There were also minor differences in the same constructs emerge. pattern/structure their and items clustered numbers of coefficients. Discussion We directly compared the three varimax-rotated principal components that we extracted with those reported by Collins and Read (1990), as reported in Table 2. The two sets of results were remarkably similar. The alpha coefficients we obtained were larger than those Cronbach alpha coefficients reported by Collins and Read (1990). for scores on each dimension in their sample were .75 for Depend, Coefficients .72 for Anxiety, and .69 for Close, respectively. alpha for the Adult Attachment Scale dimensions of Depend, Anxiety and Close were .854, .784, and .847 in our sample, respectively. the supported strongly results our aggregate, the In reliability and the validity of scores derived from the Adult Scores on this measure are considerably superior Attachment Scale. Adult Attachment -8- to those obtained on the more commonly used measure offered by Hazen and Shaver (1987). Adult Attachment -9- References Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). A psychological study of the strange situation. Hillsdale, NJ: Eribaum. (1980). Attachment and loss: Vol. 3, Loss, sadness and Bowlby, J. depression. New York: Basic Books. (1990a). Origins and functions of Carver, C. S., & Scheier, M. F. process A view. control negative positive affect: and Psychological Review, 97, 19-35. Principles of self- (1990b). Scheier, Carver, S., M. F. C. & (Eds.), Sorrentino Higgins regulation. In M. R. T. E. & Handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 3-52). New York: Guilford. Carver, C. S., Scheier, M. F., & Weintraub, J. K. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56, 267-283. (1990). Adult attachment, working N.L., Read, Collins, S.J. & models, and relationship quality in dating couples. Journal of Personality and Social Psychnlogy, 58, 644-663. (1950). Childhood and society. New York: Norton. Erikson, E. (1983). Factor analysis (2nd ed.). Hillsdale, NJ: Gorsuch, R.L. ErlbauilL evaluation. test New standards for J.P. (1946). Guilford, Educational and Psychological Measurement, 6, 427-439. (1987). Romantic love conceptualized as & Shaver, P. R. Hazan, C. Social Personality and Journal process. of attachment an